Sector playbooks

Six playbooks (construction first), then a value model per sector in the customer's money, then the trigger feed that tells you who to ring this week. The same product sells differently in a kitchen, a depot, a cleaning contract, a factory and a care home: different buyer, different law, different calendar. One playbook per sector, with the regulation quoted from the primary source.

Each one answers the same seven questions in that sector's own words: who buys and who actually signs, which regulation turns proof of receipt from a nice-to-have into a compliance purchase, what event makes them buy this month rather than next year, what they are using instead, how to price it, what they will push back with, and what the first thirty days look like.

Companion documents: the 200 targets (ranks 1–100 construction-led, 101–200 the other sectors; CSV downloads on that page). Product, pricing and sales-kit detail live in Chapter 1, Chapter 3 and Chapter 8; the price points quoted below are from the pricing chapter.

How to read the regulatory tables. Every quotation is from primary legislation or a government publication, linked. Quote the words, not a paraphrase, and let their compliance lead map it to their own obligations. Never give legal advice. Where a figure is an estimate or an inference it is marked est. or flagged in the sentence.

Construction

The sixth playbook, and the one the target list is built around. the sector playbooks (this chapter) covers the five non-construction sectors; this one covers UK construction and answers the same nine questions in the same order, so the two documents read as one set.

Construction is different from the other five in one structural way, and everything below follows from it: the people you need to reach mostly do not work for the people who will pay you. A housebuilding site with 120 people on it on a Tuesday might have eight direct employees. The duty to inform them is not eight people wide - it is 120 people wide, and CDM 2015 says so twice. That gap between the payroll and the site is the product.

Companion documents: the sector playbooks (this chapter) · the 200 targets (the construction-led target list, ranks 1-100) · the value models (below) (what the downside costs in pounds) · the trigger feed (below) (how to find the trigger events below without asking) · research-commercial.html §3.3, §3.6, §3.7 · research-competitors.html §2.14, §2.16, §3.1.

How to read the regulatory table. Every quotation is from primary legislation or a government publication, linked. Quote the words, not a paraphrase, and let their compliance lead map it to their own obligations. Never give legal advice. Estimates are marked est. or flagged in the sentence.

1. Who buys, who signs

There are three different companies in UK construction and they buy in three different ways. Work out which one you are in before the first call, because the signature sits in a different place in each.

(a) The housebuilder division

The regional or divisional business of a volume housebuilder: 9-14 active sites, a divisional MD with a real P&L, a regional H&S team, and a group function above them. (9-14 sites verified across five UK housebuilders - research-commercial.html §3.7.)

Who really signs: the Divisional MD, on the Head of H&S's recommendation. Group is the second sale, not the first. Do not start at group - a plc group function will put you into a procurement process before you have a reference.

(b) The main contractor

Tier 1 and upper tier 2: a group SHEQ function, a project-based operating model, a formal supply chain, and procurement that is genuinely load-bearing above a threshold.

Who really signs: for a first deal, the Project Director on one job, out of preliminaries. That is the single most under-used route into a tier 1 in this document. A project's prelims already carry welfare, signage, inductions and site security; a comms channel that produces induction and briefing records is the same budget line, and it does not touch group procurement.

(c) The subcontractor

Groundworks, RC frame, M&E, drylining, scaffolding, cladding, fit-out. Typically 60-600 operatives, owner-managed or PE-backed, and the population most likely to be multilingual.

Who really signs: the owner-MD, in one meeting, usually within three weeks. Subcontractors are the fastest deal shape in construction and the most under-prospected. They are also the population your housebuilder and main-contractor customers will ask you about in month two.

The multi-thread rule for this sector. Never run a construction deal on one thread. The minimum viable set is the H&S lead (champion) + the person with the P&L (signer) + one site manager (adoption). If you cannot name all three by the second call, you do not have a deal, you have a conversation.

2. The regulatory hook

Construction has the strongest regulatory hook of any sector on the target list, because CDM 2015 imposes the information duty twice - once on the principal contractor for the site, and once on every contractor for its own workers - and the second one is the answer to "the subbies aren't our employees" before the buyer has finished saying it.

Say section 40 accurately

Section 40 is the most powerful sentence you have and the easiest one to overstate. An H&S lawyer in the room will correct you, and you will lose the meeting.

What it does: where a duty is expressed as "so far as is practicable", "so far as is reasonably practicable" or "best practicable means" - which covers most of HSWA and most of CDM - and once the prosecution has established the duty and the risk, the burden shifts to the defendant to prove that it was not reasonably practicable to do more than was in fact done. The reverse burden bites only on that limb, to the civil standard (balance of probabilities).

What it does not do: it does not mean an employer is guilty until proven innocent. The prosecution still has to prove the duty applied and that a risk to health or safety arose. It does not apply to absolute duties. It is not a general presumption of guilt.

The sentence to actually say:

That is accurate, and it is still the strongest thing anyone will say to a construction H&S director all week.

The evidence base: how multilingual UK construction actually is

Lead with these two numbers. They are current, primary, and construction-specific.

Why these and not the old ones. The 2018 ONS construction migration article rests partly on the 2011 Census, and a construction H&S lead in 2026 will ask what the number looks like post-Brexit. The APS figure is the same statistical family, UK-wide, includes the self-employed - which matters enormously in construction - and carries a trend line: 14.0% (2018) → 15.4% (2023) → 15.7% (Oct 2023-Sep 2024). The migrant share of UK construction has gone up, not down. Say that; most buyers assume the opposite.

Three caveats to hold in your head, because a good compliance lead will find them:

• ONS's own note on the APS series: "The population totals used for APS estimates do not take into account any changes in migration, birth rates, death rates, and so on, since June 2021, so levels estimates may be under- or over-estimating the true values and should be used with caution."
• The London figure is a Census 2021 custom table, not a published ONS table - there is no published ONS table crossing industry with country of birth. ONS also warns: "As Census 2021 was during a unique period of rapid change, take care when using Labour Market data for planning purposes." The same Census table for England and Wales gives 16.7% against the APS's 15.7%, so the two sources corroborate each other.
• Both measure country of birth, not nationality, and not language. A worker born abroad may read English perfectly. The number establishes that the question is live; it does not answer it for their sites.

A more recent figure exists, and you should know why not to lead with it. HMRC PAYE Real Time Information for December 2025 (published 22 May 2026) shows 12.2% of UK payrolled construction employments and 45.7% of London payrolled construction employments held by non-UK nationals (gov.uk). It is newer, but it counts payrolled employments only: its UK construction base is 1.33m against the APS's 2.14m, so roughly 800,000 construction workers - the self-employed - are missing, and that group is disproportionately migrant. The Migration Observatory makes the same point in its briefing updated 15 May 2026: "because these figures only include employees, they do not fully capture working patterns in sectors with higher levels of self-employment or cash-in-hand work, such as construction" (migrationobservatory.ox.ac.uk). Use it as a London corroboration if challenged; do not use it as the headline.

But the point is not the percentage. It is the shape of the question: whatever the number is on their sites, they do not know it, and their toolbox talk is in one language. Ask the site manager, never HR - the site manager's answer is always longer.

3. Trigger events

4. The incumbent

What is actually there today, in order of prevalence: a WhatsApp group per site run on the site manager's personal phone; a printed toolbox-talk sheet with signatures on a clipboard in the site office; a noticeboard in the canteen; a paper or tablet induction at the gate; and, for staff only, Microsoft Teams and an intranet that nobody in a hi-vis has ever opened.

Above that, one or more of:

Read the gap out loud. Nobody in the construction stack translates the message, and nobody captions video. Everybody has a site layer TeamTalk does not have. That is the trade, and it argues for co-existence and integration rather than replacement - which is also, conveniently, what removes the rip-and-replace objection.

The positioning line: "Procore runs the project, HammerTech runs the site, Biosite knows who's through the turnstile. None of them can tell you that the Romanian steel fixer read the lift plan - in Romanian - and confirmed it. That's the only thing we do."

5. Three opening lines

1. "CDM regulation 15(8) says every contractor must give each worker under their control appropriate instructions and information. On your last site, how many of the people on it were your employees - and how did you evidence the other ninety per cent?"
2. "HSE publishes its own safety guidance in nineteen languages. What language was your last toolbox talk in, and could you print who read it?"
3. (subcontractor) "Which principal contractor is currently giving you the hardest time about evidencing briefings - and what are you sending them?"

A fourth, for a housebuilder that publishes a training KPI: "You publish a 92% mandatory-training completion figure. Can you evidence it per person, per briefing, in the language they read it in?" Where a prospect has a published number, set the pilot target against their commitment, not one you invented. (Barratt publishes a 92% mandatory-training-completion KPI - research-commercial.html §3.7.)

6. The pricing unit

The site is the unit for a housebuilder and a main contractor. The company is the unit for a subcontractor. Do the arithmetic in front of them; construction buyers assume software vendors are trying it on, and showing the cheaper of two options you could have quoted buys more credibility than any discount.

The three rules that hold in construction:

1. Never price per seat. The workforce on a site turns over faster than a billing cycle and most of it is not on the payroll. A per-seat price makes the customer's finance team ration exactly the people the regulation is about. Say this explicitly - it is the difference between you and Mitti at USD 24 a seat.
2. Re-band only at renewal. A division that wins two sites in March must not receive a bigger invoice in April.
3. Enrol subcontractors at no cost, and say so before they ask. It is the whole product thesis and it is also HammerTech's model, so it will not surprise them - it will reassure them.

On the price ceiling. Microsoft 365 F1 lists at GBP 2.30 per user per month and the Viva Employee Communications and Communities add-on at GBP 1.54, both paid yearly on Microsoft's UK pages as at 28 August 2026 (F1, Viva) - GBP 3.84 combined. Every quote above lands under it. Treat GBP 3.84 as where price objections start, not as a hard ceiling: Microsoft's frontline pricing moved on 1 July 2026 and will move again, and an F1 licence still gives a site operative nothing they can use without a work account. Re-check both pages before putting the number in writing.

7. Objections specific to this sector

8. The calendar

Two things to ask on every construction discovery call:

1. "When does your financial year start?" - housebuilder year-ends vary (June, September, October and December all appear, and several changed after recent mergers). Diary the budget conversation for six weeks before it. Do not work from a table; verify per account.
2. "Has your business designated a close-down, and when?" - because the industry answer no longer exists.

9. First 30 days of a pilot

Scope it at three sites, not a division. A division of 9-14 sites is the right commercial target and the right thing to price - but it is a rollout, not a pilot, and one founder cannot support it while also selling. Three sites gives you a spread and a defensible number; one gives you an anecdote. The rest is the expansion you have already priced into the pilot agreement.

1. Days 1-3 - Three sites, two named sponsors, criteria in writing. Choose the three sites with the Head of H&S: one high-performing, one average, one difficult. Include the divisional office so managers see the traffic. Get two names on the pilot document - the H&S lead (owner) and the person with the P&L (sponsor). Agree in writing: 70% activated within 14 days · 85% confirmed receipt on mandatory posts within 72 hours · at least 2 posts per site per week, committed by the customer · and one compliance export the H&S lead accepts as an evidential briefing record. Where they publish a training-completion KPI, set the confirmed-receipt target against their own number. Put the post-pilot divisional price in the document on day one.
2. Days 4-7 - Get the subcontractor list. This is the critical path. The site manager has a complete list of directly employed operatives and an incomplete one of everyone else, and this is where the pilot numbers go wrong. Ask for the site access records - Biosite, Insite, the turnstile, the signing-in book - not the payroll. Build groups by trade and by gang, not by employer: groundworks, RC frame, M&E, drylining, scaffold, plant, site management, subcontract supervision. Ask the site managers which languages are actually on site; the HR answer is always shorter than the truth. Brand the tenant.
3. Days 8-10 - Onboard at the gate, at the induction, and at the canteen. Put the join card in the induction pack so every new starter enrols on day one without you needing their number in advance. Stand at the turnstile for two consecutive 6am starts and one afternoon shift change. QR posters in the canteen, the welfare unit and the drying room. Enrol every subcontract operative on the same card - it costs nothing and it is the entire point. Target the whole site inside 48 hours.
4. Days 11-21 - Post the work that already exists, in the right order. Take last month's toolbox talks and safety alerts and publish them as must-reads: (i) a genuine site-rules or emergency-procedure item, translated, as the first mandatory post; (ii) a real risk briefing tied to an actual activity - a lift plan change, a temporary works change, an excavation or a scaffold handover - with a captioned 60-second video shot on a phone; (iii) one hazard report demonstrated by an operative on their own phone, in front of the site manager. Run the chase list at 48 hours and have the site manager speak to the last few names at the morning brief. A named senior person posts message #1 - if the Divisional MD posts it, activation runs materially higher.
5. Days 22-30 - Produce the artefact, then book the decision. Print the per-person confirmation record for the mandatory post - named individual, delivery time, confirmation time, and the language they read it in - and put it in front of the Head of H&S with one question: "if HSE, your insurer or your principal contractor asked how you evidence that every worker on this site received this, would this answer it?" Their answer is the pilot outcome; nothing else is. Then book the day-60 decision meeting before you leave site, with the person who owns the P&L in the room, and ask for the site count for the division in the same conversation.

Two failure modes to design against. First, the pilot dies from the manager end, not the worker end - if nobody posts, workers stop opening. Chase the posters weekly, by phone, and treat a silent site manager as a red flag in week two, not week six. Second, subcontractor enrolment is where the numbers go wrong - treat the subcontractor list as the critical path from day four, and re-count it at day 14 against the access records.

The five sharpest construction hooks, ranked

1. CDM 2015, reg 15(8) - "A contractor must provide each worker under their control with appropriate supervision, instructions and information..." Every contractor, every tier, every worker under their control. This is the answer to "the subbies aren't ours" and it is also why the subcontractor is your next sale. (source)
2. HSE L153 ¶70 - "Any information or instruction provided should be in simple, clear English (and/or other languages where appropriate)... The use of photographs or diagrams in written communication can be very helpful." The regulator's own CDM guidance describing translated, illustrated briefings. (PDF)
3. HSWA 1974, s.40 - once the duty and the risk are established, "it shall be for the accused to prove... that it was not... reasonably practicable to do more than was in fact done." Say it with the qualifier; it is still devastating. (source)
4. CDM 2015, reg 13(4)(a) - the principal contractor must ensure "a suitable site induction is provided", and L153 ¶133(f) says that induction should cover "arrangements for briefing workers on an ongoing basis, eg toolbox talks." An absolute duty covering every site worker, plus the regulator's expectation that you explain the ongoing mechanism. (reg 13, L153)
5. MHSWR 1999, reg 10 with reg 12 - "comprehensible and relevant information", owed also to "any person working in his undertaking who is not his employee." The two clauses on which the whole translation feature rests. (reg 10, reg 12)

Runner-up worth carrying into any 18m+ residential conversation: the golden-thread information principles in SI 2023/909 reg 31 - accurate, up to date, intelligible to the intended reader, access-controlled and change-logged with who and when - because that is a specification for a records system and it reads like a product requirements document. (source)

Sourcing: the multilingual-workforce evidence, ranked

Not used, and why. CITB's Migration and Construction (6th edition, June 2023) is the obvious industry source, but its PDF now returns 404 on citb.co.uk and the landing page redirects, so the figures circulating from it (a migrant share of 9.8% in 2021) could not be checked against the primary document and are on an undocumented, much narrower definition than the ONS measure. Do not quote CITB migration figures until the document itself can be read. CITB's newest research (26 August 2026) is about domestic workforce mobility, not migration.

What this playbook does not claim

• Nothing here is legal advice. Quote the words, link the source, let the customer's compliance lead map it to their obligations.
• Section 40 does not reverse the burden of proof generally. It shifts it onto the defendant only on the "practicable / reasonably practicable / best practicable means" limb, once the prosecution has established the duty and the risk. Stated any more strongly, it is wrong, and an H&S lawyer will say so.
• TeamTalk is not the golden thread, and the golden-thread duties in SI 2023/909 attach to building information, not to workforce communications. The parallel between the golden-thread information principles and a defensible briefing record is a rhetorical one. Position it as "your golden thread covers the building; this covers the people" and stop there.
• BSR expectations are drawn from BSR's published strategic plan and the gov.uk announcement of its move to a standalone body, not from any BSR statement about workforce communications. There is no BSR requirement for a comms record. Do not imply one.
• HSE does not "endorse" or "recommend" translation software. The phrase "use professional translation software or free online tools" appears as part of one item in a list of options on HSE's help-with-language-issues page, introduced by "Some options to ensure effective communication when people in a workplace are not all fluent in the same language include:". HSE's own HTML for that page merges two separate options into a single malformed list item, so the phrase renders mid-bullet. Quote the fragment; do not build a sentence around HSE recommending anything. The duty sentence - "Employers have a duty to provide comprehensible information to workers" - is the strong claim, and it needs no help.
• The "19 languages" figure is the count of languages that have a translated-information page on hse.gov.uk as at 24 February 2025. The volume of material behind each language varies and has not been audited. Say "translated safety information in 19 languages", not "a full library in 19 languages".
• Competitor claims (HammerTech, Mitti/SafetyCulture, Procore, Fieldwire, HandsHQ, Biosite) are from those vendors' own public pages on the dates recorded in research-competitors.html §2.14, and reflect what is published, not what is possible on a bespoke enterprise contract. Ask the prospect what their contract actually includes rather than asserting what it does not.
• The GBP 3.84 comparison is Microsoft's UK list price for F1 plus the Viva Employee Communications and Communities add-on as at 28 August 2026. Microsoft changed frontline pricing on 1 July 2026 and will change it again. Re-check both pages before quoting it, and treat it as where objections start rather than a ceiling.
• Trigger-event frequencies, buyer-signature patterns and calendar windows other than the CIJC dates are practitioner estimates drawn from sector contract cycles, not measured data. Treat them as prospecting heuristics. The CIJC winter break dates and the discontinuation of the summer shutdown are verified from the CIJC/Build UK document itself.
• No incident-reduction claim appears anywhere in this playbook, because none is supported. See the value models (below) for what the evidence base actually does and does not contain.

The five non-construction playbooks

The argument that does not change between sectors

Three things are true everywhere on this list, and they are the spine of every pitch below.

1. The duty is to communicate, and once a duty and a risk are shown, the burden of proving you did what was reasonably practicable is on you. Section 40 of the Health and Safety at Work etc. Act 1974: "it shall be for the accused to prove... that it was not practicable or not reasonably practicable to do more than was in fact done to satisfy the duty or requirement" (legislation.gov.uk). Food law has the identical shape - section 21 of the Food Safety Act 1990 makes it "a defence for the person charged to prove that he took all reasonable precautions and exercised all due diligence" (legislation.gov.uk). In both regimes the defendant supplies the evidence. A confirmed-receipt export is that evidence.

2. English-only is not automatically "comprehensible". Regulation 10 of the Management of Health and Safety at Work Regulations 1999 requires "comprehensible and relevant information" (legislation.gov.uk). Regulation 8(3) of PUWER 1998 requires equipment information to be "readily comprehensible to those concerned" (legislation.gov.uk). HSE itself publishes safety material in nineteen languages, including Polish, Romanian, Lithuanian, Latvian and Portuguese (hse.gov.uk) - the regulator's own practice concedes the point.

3. The duty follows the person, not the payroll. MHSWR regulation 12 extends the comprehensible-information duty to "any person working in his undertaking who is not his employee" (legislation.gov.uk); COSHH regulation 12(4) does the same for "any person (whether or not his employee) who carries out work in connection with the employer's duties" (legislation.gov.uk). This is the answer to "the agency staff aren't ours" in every sector below, and the reason TeamTalk's pricing counts agency and subcontract workers without charging per seat for them.

1. Contract catering and hospitality

Kitchens, front of house, housekeeping floors and event teams are the purest version of the problem: a workforce that is young, high-churn, heavily non-UK-born, split across sites the head office never visits, on shift patterns that guarantee half of them are asleep when the notice goes out - and working under a food-safety regime where the evidence of briefing is the defence.

Who buys, who signs

Who really signs: in contract catering the Operations Director or divisional MD, because the GBP 2,500 pilot is sized for a divisional budget (confirm the delegated-authority limit on the first call; the supplier pack goes to procurement in parallel). In hotel and pub groups the signature is usually higher (Ops Director or COO) and the Area Manager is the gatekeeper you must win first.

The regulatory hook

Trigger events

The incumbent

WhatsApp groups run by unit managers on personal phones, plus a paper allergen folder and a laminated notice board in the back-of-house corridor. Above that sits either nothing or a rota tool with comms bolted on: Fourth (dominant in UK hospitality; workforce management, HR/payroll and inventory, with "Employee Engagement" listed but no published acknowledgement feature - uk.fourth.com), Deputy (Lite $5, Core $6.50, Pro $9 per user per month, Messaging & News Feed on all tiers - deputy.com/pricing), Planday (from GBP 2.99 per user per month, "Team communication", Messenger/SMS and Newsfeed - planday.com/uk/pricing) and Rotageek (multi-site, 400+ employees, AI scheduling, no published pricing - rotageek.com).

How to position against them, honestly. These are scheduling products. Their feed exists to tell people about shifts. Ask two questions: does it translate the post into the reader's language, and does it produce a per-person printable record an EHO or a client auditor will accept? If the answer to either is no, the rota tool is not the compliance channel - and you are not asking them to replace it. TeamTalk sits alongside the rota. Say that out loud; it removes the rip-and-replace objection entirely.

Three opening lines

1. "When a supplier recall or an allergen change goes out to your kitchens, how do you know every kitchen has read it - and could you print that on the day an EHO asks?"
2. "Section 21 of the Food Safety Act puts the due-diligence burden on you, not the council. On your last recall, what evidence did you keep that the message landed?"
3. "You mobilise a new contract and inherit sixty TUPE'd staff with no work email and four first languages. How does head office reach them in week one?"

The pricing unit

The kitchen or the unit is the site. That is how a caterer already budgets and how a hotel already thinks.

• Single-site or small multi-site operator (an independent caterer, a hotel, a stadium): quote the Site plan at GBP 180 per site per month, up to 100 enrolled workers, overage GBP 1.60 per worker per month. A 70-person hotel reads as GBP 2.57 per worker.
• Contract caterer with an estate: quote the Company band. Three or four kitchens and 150-250 staff is the GBP 550 band; eight kitchens and ~300 staff is the 251-500 band at GBP 950 a month, GBP 11,400 a year, every kitchen, every language.
• Pub and restaurant groups: use bands, not sites. A 25-person pub on the Site plan is GBP 7.20 per worker per month, which is where price objections start (the internal sanity check in Chapter 3 is Microsoft F1 + Viva at GBP 3.84 a user a month, annual, ex VAT, before Microsoft's July 2026 rise; never quote it to a customer). A 150-250 person group is the GBP 550 band; a 400-unit group with 8,000 staff is Enterprise, quoted per estate.
• Hotel groups: hotel = site works up to roughly 12 hotels; above that the band is cleaner and re-bands only at renewal, which matters because housekeeping headcount swings seasonally.

Rule of thumb: if the average unit has fewer than 50 people, price by band; if it has more than 50, price by site.

Objections specific to this sector

The calendar

Schools and education caterers run on the academic year, so their buying year is roughly six months ahead of everyone else's. B&I caterers run on the client's financial year. Ask which they are in the first five minutes.

First 30 days of a pilot

1. Days 1-3 - Pick the unit and write the criteria down. One region, up to three kitchens or hotels, up to 250 workers. Agree in writing: 70% logged in within 14 days, 85% confirmed receipt on mandatory posts within 72 hours, two posts per site per week, and one compliance export the Food Safety lead accepts as evidence. Name the internal owner - usually the Compliance Manager, not Ops.
2. Days 4-7 - Build the tenant and the language list. Brand colours and logo. Ask the unit managers, not HR, which languages are actually spoken on shift; the HR answer is always shorter. Create user groups for kitchen, front of house, housekeeping and night shift.
3. Days 8-10 - Onboard at the shift handover. Print the join card, put it on the pass and by the time clock, and stand there for two consecutive shift changes. Target the whole unit in one break. Do not email an invitation to anyone.
4. Days 11-21 - Post real work. A genuine allergen or recipe change as the first must-read, a hygiene reminder with a video and captions, one hazard report demonstrated by a chef. Use the chase list on day two of each must-read and have the unit manager phone the last three names.
5. Days 22-30 - Produce the artefact and book the review. Print the per-person confirmation record for one must-read and put it in front of the Food Safety lead with one question: "would this satisfy an EHO or a client auditor?" Their answer is the pilot outcome. Book the day-60 decision meeting before you leave the building, with the Operations Director in the room.

2. Logistics and warehousing

Drivers, warehouse operatives, yard staff, cold-store pickers and depot night shifts. Two distinct populations with one problem: the warehouse population is dense, multilingual and shift-based; the driver population is dispersed, never at head office, and legally the subject of a documented briefing regime. Neither has a work email.

Who buys, who signs

Who really signs: the Operations Director. But the Transport Manager is the most motivated human in the building, because O-licence undertakings and DVSA scrutiny attach to them by name. Multi-thread both.

The regulatory hook

Trigger events

The incumbent

WhatsApp groups per depot and per shift, a printed noticeboard in the transport office, a toolbox-talk sign-in sheet on a clipboard, and increasingly a driver app supplied by the telematics or TMS vendor that reaches drivers but not the warehouse. Larger 3PLs have Microsoft Teams for salaried staff and nothing for the floor. Where a workforce management system exists it is typically Deputy, Quinyx or a bespoke rota, none of which translates a post or produces a per-person compliance record.

The positioning line: "Your telematics app reaches the cab and Teams reaches the office. Nothing reaches the pick face."

Three opening lines

1. "Your O-licence undertakings are in the Transport Manager's name. When a defect-reporting instruction goes out, what proof exists that each driver received it?"
2. "How many first languages are on the night shift at your biggest DC - and what language was the last forklift safety alert written in?"
3. "Peak starts in eight weeks and you'll take on several hundred agency staff. How do they get a safety notice on day one without a work email?"

The pricing unit

The depot or DC is the site - but only for the fixed population. Drivers, multi-drop rounds and field engineers have no site, and roughly a fifth of this sector is like that.

• Warehousing / DC operator: Site plan at GBP 180 per site per month up to 100 enrolled workers. A large DC with 600 people is GBP 180 + 500 x GBP 1.60 = GBP 980 a month, GBP 1.63 per worker - comfortably below where per-head objections start (Chapter 3's internal GBP 3.84 sanity check). Check this against the band before quoting; for a single big DC the Site plan with overage is usually the honest, cheaper number.
• Multi-depot 3PL: Company band (GBP 550 up to 250 enrolled workers, then GBP 950, 1,600, 2,600), unlimited sites, banded by enrolled workers, re-banded only at renewal. This is the right answer when headcount swings 40% at peak - a busy October must not produce a surprise invoice.
• Driver-only or round-based operators (parcels, multi-drop, pallet trunking): Company band, organised by depot, hub or round rather than by building.
• Pallet networks (Palletways, Pallet-Track, Palletline, TPN, Fortec, UPN): the member depots are separate legal entities. Do not try to sell one contract across the network. Sell the network central operation a pilot, then ask for a slot at the members' conference. Treat the network as a channel, not an account.
• Ports and bus operators: the terminal or the garage is the site, and both are usually 200-800 people, so quote the band.

Objections specific to this sector

The calendar

Food wholesale runs a second peak into the pre-Christmas delivery window and a hard stop over the shutdown. Bus and coach operators buy against the timetable change dates rather than the calendar year - ask when their next major service change is.

First 30 days of a pilot

1. Days 1-3 - One depot, two sponsors, written criteria. Pick a single DC or depot of 100-250 people. Get the SHEQ Director and the Transport Manager both named on the pilot document - the warehouse and the driver population are different problems and each needs an owner. Criteria in writing: 70% logged in within 14 days, 85% confirmed on mandatory posts within 72 hours, two posts per week, and one export the SHEQ lead accepts as a briefing record.
2. Days 4-7 - Build groups that match the shift pattern, not the org chart. Days, lates, nights, drivers, yard, agency. Ask the shift managers which languages are actually on nights; the HR answer will be shorter than the truth. Brand the tenant.
3. Days 8-10 - Onboard at the gate and at the transport office. Put the join card at the turnstile, in the canteen and on the transport-office counter. Catch three consecutive shift changes in 24 hours, including the night handover. Enrol agency workers on the same card - it costs nothing and it is the whole point.
4. Days 11-21 - Post the work that already exists. Take the last month's toolbox talks and safety alerts and publish them as must-reads: an MHE or pedestrian-segregation alert, a defect-reporting reminder for drivers, and one genuine operational change. Run the chase list at 48 hours and have the shift manager speak to the last few names at handover.
5. Days 22-30 - Put the export in front of the Transport Manager. Print the per-person record for the driver briefing and ask: "if DVSA or the Traffic Commissioner asked how you evidence that every driver received this, would this do it?" That answer is the pilot result. Book the day-60 decision meeting with the Operations Director and ask for the depot count in the same conversation.

3. Facilities, cleaning and security

Structurally the best product fit on the entire target list. A commercial cleaning contractor employs people who work alone or in small teams, at night, inside somebody else's building, with no company premises, no email, and - in London especially - a first language that is very often not English. The employer's only channel to them today is the supervisor's personal phone.

Who buys, who signs

Who really signs: the Operations Director, with HSEQ as champion. In security, the ACS compliance lead carries unusual weight because approval is commercially load-bearing - public-sector and large corporate buyers require it.

The regulatory hook

Trigger events

The incumbent

Almost entirely WhatsApp, plus printed assignment instructions in a folder in the cleaners' cupboard or the security gatehouse, plus a supervisor doing a phone round. Some larger contractors have an intranet nobody on shift can reach, and a few have bought a frontline app - if they have, it is most likely Blink. Rota tools are less embedded here than in hospitality; where they exist, they are usually a time-and-attendance system (biometric clock or a call-in line) rather than a comms channel.

The positioning line: "Your operatives work at 5am in a building you don't own. The only company channel that reaches them is a supervisor's personal WhatsApp - and when the supervisor leaves, so does it."

Three opening lines

1. "You mobilise a new contract and inherit 200 TUPE'd cleaners on day one. How do you get an assignment instruction to all of them in week one, in the languages they actually read?"
2. "COSHH regulation 12 requires information about the substances people are exposed to. When you changed chemical supplier, how did you evidence that every operative got the new safety data sheet - and read it?"
3. "When your ACS assessing body samples your operative management, what do you hand them for briefings? A folder, or a per-person record?"

The pricing unit

Not the building. The contract - and usually not even that. This is where the site model breaks and you must say so before the buyer does. A cleaning contractor with 900 small contracts cannot buy 900 sites.

• Default: Company band, by enrolled workers. A 700-operative regional cleaner is the 501-1,000 band at GBP 1,600 a month, GBP 2.29 per worker. Unlimited "sites" lets them organise by contract, building, round or supervisor - which is exactly how they already think.
• Exception - one big building: an airport terminal, a stadium, a hospital or a single corporate HQ contract with 100+ operatives can be sold as a Site at GBP 180 per month with overage. This is a genuinely useful land-and-expand: sell one flagship contract, then take the reference to the Operations Director for the estate.
• Security: band by licensed operatives (a 150-250 officer contractor is the GBP 550 band; 251-500 is GBP 950). Guarding contracts are people, not places.
• Waste and grounds: the depot is the site and the round is the group. A depot of 120 with eight rounds is one Site plus a bit of overage; a national operator is a band.
• Laundries: the plant is the site, and plants run 200-500 people, so quote the band or the Site-plus-overage arithmetic, whichever is lower and honest.

The line that closes it: "You don't pay per cleaner, and you don't pay per building. You pay one number for the whole workforce, and it only changes at renewal - so winning a contract in March doesn't cost you anything until we next sit down."

Objections specific to this sector

The calendar

Waste and recycling operators buy against local-authority contract cycles, which are typically seven or eight years with break points - ask which contracts are in their last two years, because that is where the evidence pressure is highest.

First 30 days of a pilot

1. Days 1-3 - Pick a mobilisation, not a contract. The best pilot in this sector is a contract that is about to start, because the population arrives on a known date with no incumbent channel. If nothing is mobilising, pick one flagship building or one region of 200-500 operatives. Criteria in writing, with the HSEQ lead as owner: 70% logged in within 14 days, 85% confirmed within 72 hours, two posts a week, one export their assessing body or client would accept.
2. Days 4-7 - Build groups by contract and by shift, and load the languages. Cleaning: early, day, evening, night, periodic. Security: static, mobile, control room. Waste: by round. Ask the supervisors which languages, not HR.
3. Days 8-10 - Onboard where they actually are. For cleaning, that is the cupboard, the signing-in book and the 5am start; for security, the gatehouse handover; for waste, the depot yard before the rounds go out. Print the join card into the mobilisation pack so TUPE'd staff enrol themselves on day one without you needing their phone numbers in advance.
4. Days 11-21 - Lead with COSHH. Make the first must-read a real chemical or safety data sheet change, translated, with a captioned 60-second video of the correct dilution. Follow with an assignment-instruction update and one lone-working reminder. Demonstrate a hazard report from a cleaner's own phone.
5. Days 22-30 - Turn the export into a bid asset. Print the per-person record and take it to the Operations Director with one question: "could you put this in your next tender as evidence that every operative was briefed, in their own language?" If the answer is yes, you are no longer selling software, you are selling win rate. Book day 60 with the Operations Director and the bid lead in the room.

4. Food manufacturing

A factory floor of 400-900 people, running two or three shifts, with a documented nationality mix that is frequently published in the company's own modern slavery statement, audited annually against a standard that specifically examines training and competence records. Of all five sectors this is the one where the customer already agrees with you about the problem - the question is only whether they will spend money on it this year.

Who buys, who signs

Who really signs: the Operations Director or Site Director, but the Technical Director opens the door, because they are the one who has just been asked by an auditor to produce evidence that everyone was briefed on the allergen changeover.

The regulatory hook

Trigger events

The incumbent

Line noticeboards, laminated SOPs at the workstation, a pre-shift huddle with a sign-in sheet, and a WhatsApp group per line or per shift run by the team leader. Some sites have digital signage in the canteen. Larger groups have SAP or a Workday HR system that reaches nobody on the floor, and a growing number have bought a factory-floor tablet system (SafetyCulture/Mitti and similar) whose comms module is an afterthought. Where a frontline app has been bought it is usually Blink.

The positioning line: "The huddle sheet proves attendance at a meeting. It does not prove that the Romanian packer understood the allergen changeover. Your auditor knows the difference."

Three opening lines

1. "Your modern slavery statement says you employ people of [N] nationalities. What language was your last allergen changeover instruction written in?"
2. "When your BRCGS auditor asks for evidence that every operative was briefed on the new procedure, what do you hand them - a signing sheet, or a per-person record?"
3. "Talk me through your last recall. How long did it take to reach every shift on every line, and how do you know it did?"

The pricing unit

The factory is the site - but a factory is bigger than the Site plan's 100-worker allowance, so do the arithmetic in front of them.

• Single-site manufacturer, 400-900 people: Site plan GBP 180 + overage GBP 1.60 per worker above 100. At 600 workers that is GBP 980 a month (GBP 1.63 per worker), against a 501-1,000 Company band of GBP 1,600 (GBP 2.67 per worker). Quote the lower number and say why - it buys enormous credibility in a sector that assumes software vendors are trying it on. Both are well below where per-head objections start (Chapter 3's internal GBP 3.84 sanity check). A 150-250 person plant or group is the GBP 550 band.
• Multi-site group, 2,000+ workers: Enterprise from GBP 3,600 a month, quoted per estate. Groups this size will want SSO and a security review, which is what Enterprise is for.
• Seasonal packhouses and produce growers: Company band, re-banded at renewal only. This matters more here than anywhere else - a soft-fruit business can go from 300 to 2,500 people between April and August. Say explicitly: "your August headcount does not change your August invoice."
• Group with several factories: band by total enrolled workers, then let them create a "site" per factory, per line and per shift at no extra cost.

Objections specific to this sector

The calendar

First 30 days of a pilot

1. Days 1-3 - One site, the Technical Director as champion, criteria written down. Pick one factory. Get the Technical Director to define what "acceptable audit evidence" means to them, in their words, and put that sentence in the pilot document alongside the numbers: 70% logged in within 14 days, 85% confirmed within 72 hours, two posts a week.
2. Days 4-7 - Groups by line and by shift, languages from the shift leaders. Days, nights, weekend crew, hygiene team, engineering, agency. This is the sector most likely to have a published nationality count - find it in their modern slavery statement before the meeting and use it to build the language list.
3. Days 8-10 - Onboard in the changing room and the canteen, never on the line. Phones do not belong on the production floor and you should say so first. Put the join card at the clock, in the changing rooms and in the canteen, and catch every shift within one 24-hour cycle including the weekend crew.
4. Days 11-21 - Run a real allergen changeover and a recall drill. Publish an actual changeover instruction as a must-read with a captioned video. Then, with the Technical Director's agreement, run a mock recall: publish, watch the reach page, and time how long it takes to reach 100% across three shifts. That number is the deliverable.
5. Days 22-30 - Hand the audit pack over. Print the per-person confirmation record for both posts and give it to the Technical Director to show their certification body or their retailer's technical team. Ask directly whether it closes the training-evidence gap. Book day 60 with the Operations Director, and ask which of their other factories goes next.

5. Care and housing repairs

Two workforces that look different and buy for the same reason. Care staff are shift-based, heavily internationally recruited, working in a home rather than an office, under a regulator that inspects records. Housing repairs operatives are mobile trades in vans, working inside residents' homes under a fast-tightening safety regime. Neither has a work email; both are subject to a duty to prove that instruction reached them.

Who buys, who signs

Who really signs: in care, the Director of Operations, with Quality & Compliance as champion and the Registered Manager network as the real adoption risk. In housing, the Director of Property Services, and repairs budgets are usually larger and less contested than comms budgets.

The regulatory hook

Trigger events

The incumbent

In care: a policy management system (Quality Compliance Systems, Radar Healthcare and similar) that emails policies to managers, a handover book on the unit, a noticeboard in the staff room, and a WhatsApp group per home. The gap is always the same - the policy tool reaches the Registered Manager and stops. In housing DLOs: a job-management or dynamic-scheduling system (a repairs mobile app on the operative's device) that carries jobs but not notices, Microsoft Teams for office staff, a toolbox-talk clipboard, and a supervisor's WhatsApp group per trade.

The positioning line, care: "Your policy system proves the policy exists and that the manager received it. Regulation 17 asks what records you keep about the people employed in the activity. Who read it?"

The positioning line, housing: "Your scheduling app sends jobs to the van. It doesn't send the asbestos alert, and it doesn't tell you who read it."

Three opening lines

1. "Regulation 17 asks you to keep records relating to the persons employed in the regulated activity. When you change a safeguarding procedure, what record shows every care assistant on every shift actually received it?"
2. "You've recruited internationally this year. What language did your last medication-safety alert go out in, and how do you know the night staff read it?"
3. (housing) "Awaab's Law has been running since October 2025 and the timescales start when the resident reports. When a case lands, how fast does the operative on the estate know - and can you show that they did?"

The pricing unit

The care home is the site, and it is the cleanest fit anywhere in this document. A typical home has 60-120 staff, which sits squarely inside the Site plan's 100-worker allowance.

• Care homes: GBP 180 per home per month. At 80 staff that reads as GBP 2.25 per worker. A three-home group with 150-250 staff is the GBP 550 band; a 25-home group is the 1,001-2,000 band at GBP 2,600 a month if headcount is around 2,000, or Site-by-site if they want to start with one home. Start with one home. It is the easiest pilot to run in the entire target list because the Registered Manager is a single, identifiable owner.
• Domiciliary and supported living: there is no site. Company band, organised by branch, round or supervisor. A 3,000-carer domiciliary provider is Enterprise territory; a 400-carer regional provider is the 251-500 band at GBP 950.
• Housing association DLOs: the DLO is the unit, not the housing association. A repairs workforce of 300-800 operatives is a Company band. Do not try to price the whole association - the tenant-facing side is not your buyer and the frontline you want is the trades.
• Healthcare facilities providers: the hospital contract is the site; a large acute contract runs 300-800 porters, cleaners and caterers, so use Site-plus-overage or the band, whichever is lower.

Objections specific to this sector

The calendar

First 30 days of a pilot (care)

1. Days 1-3 - One home, one owner, written criteria. Pick a single home with a confident Registered Manager and 60-120 staff. Agree in writing: 70% logged in within 14 days, 85% confirmed on mandatory posts within 72 hours, two posts a week, and one export the Head of Governance accepts as a regulation 17 record.
2. Days 4-7 - Tenant, brand, languages, groups. Ask the Registered Manager which languages are on the night rota. Build groups for care staff, nurses, night shift, housekeeping and kitchen, and maintenance.
3. Days 8-10 - Onboard at handover. Join card on the staff-room noticeboard and at the handover desk. Catch both day and night handovers on the same 24 hours. Enrol agency staff on the same card - it costs nothing and it is the point.
4. Days 11-21 - Post real governance. One genuine policy or procedure update as a must-read; one safeguarding refresher with a captioned video; one infection-control notice. Use the chase list at 48 hours and have the manager speak to the remaining names at handover.
5. Days 22-30 - Hand the export to Governance. Print the per-person confirmation record and ask the Head of Quality & Compliance one question: "if an inspector asked how you evidence that a procedure change reached every member of staff, would this answer it?" Book the day-60 decision meeting with the Director of Operations, and ask for the second and third homes in the same conversation.

(For housing repairs, substitute: pick one DLO trade group; make the first must-read a genuine asbestos-register or gas-appliance notice; and hand the export to the Head of Health & Safety rather than Governance.)

The five sharpest hooks, ranked

If you only remember five sentences from this document, make them these.

1. Food Safety Act 1990, s.21 - "a defence for the person charged to prove that he took all reasonable precautions and exercised all due diligence". The burden of proof is on the caterer or manufacturer, in food law exactly as in health and safety law. (source)
2. HSWA 1974, s.40 - "it shall be for the accused to prove... that it was not practicable or not reasonably practicable to do more than was in fact done". Say it in every discovery call, in every sector. (source)
3. COSHH 2002, reg 12 - information, instruction and training on hazardous substances, "adapted to take account of significant changes", owed to "any person (whether or not his employee)". Every chemical change in every kitchen, warehouse, cleaning contract and factory is a dated, evidenceable event. (source)
4. Health and Social Care Act 2008 (Regulated Activities) Regulations 2014, reg 17 - providers must "maintain securely such other records as are necessary... in relation to persons employed in the carrying on of the regulated activity". Inspected, unannounced, in every care home in England. (source)
5. Hazards in Social Housing (Prescribed Requirements) (England) Regulations 2025, in force 27 October 2025 - statutory investigation and written-summary timescales for social landlords, with a defence only where "all reasonable endeavours" were used. The newest deadline on the list and the one nobody has finished implementing. (source)

Runners-up worth carrying: MHSWR 1999 reg 10 "comprehensible and relevant information" and PUWER 1998 reg 8 "readily comprehensible to those concerned" - the two phrases the translation feature hangs on; Regulation (EC) 852/2004 Annex II Ch. XII, the continuous food-handler instruction duty; and BRCGS Issue 9 clause 7.1, which is commercial rather than legal but decides whether a retailer will buy from the site at all.

What this document does not claim

• BRCGS clause wording is summarised from BRCGS's own published guideline material and secondary interpretation, not the paywalled Standard. Confirm with the customer's technical team before putting it in writing.
• SIA ACS briefing-record requirements are an inference. The public gov.uk ACS page sets out licensing, fitness and director requirements but does not specify briefing evidence; the detail sits in the ACS self-assessment workbook. Ask the compliance lead what their assessing body actually samples.
• Competitor feature claims (Fourth, Deputy, Planday, Rotageek) are taken from those vendors' own public pages on the dates fetched and reflect what is published, not what is possible on a bespoke enterprise contract. Ask the prospect what their contract actually includes rather than asserting what it does not.
• Trigger-event frequencies and calendar windows are practitioner estimates drawn from sector contract cycles, not measured data. Treat them as prospecting heuristics.
• Nothing here is legal advice. Quote the words, link the source, and let the customer's own compliance lead map it to their obligations.

Value models: the customer's downside in pounds

What the downside actually costs, in pounds, from public data. One page per sector.

The pilot criteria in the sector playbooks - 70% activated, 85% confirmed, two posts a week - are good pilot criteria and rare in a first-time founder's kit. But they are adoption metrics, and none of them is denominated in the customer's money. Without a number in their currency, £950 a month is a cost line, and the first finance push-back at renewal kills the account.

This document supplies the other half. For each of the six sectors it puts a pound figure on the downside from published sources, shows the arithmetic, labels every estimate as an estimate, and ends with the same sentence: "so £950 a month against …".

Companion documents: the sector playbooks (this chapter) · the construction playbook (this chapter) · the trigger feed (below) · the market sizing (Chapter 2).

How to use this, and the three rules that keep it honest

Rule 1 - never claim a fine you would have avoided. You cannot say "TeamTalk would have prevented this £2m fine". You do not know that, and any competent H&S director will say so. What you can say is what the sentencing guideline says about culpability and mitigation, and let them do the arithmetic on their own turnover.

Rule 2 - never claim an incident reduction. There is no credible published study supporting a quantified "briefings reduce incidents by X%" claim. See §8. If you make one, you will be asked for the source, and there isn't one.

Rule 3 - put the customer's own numbers in. Every model below has stated assumptions. In a real meeting, replace them with their turnover, their site count, their manager cost, live, on a napkin. The model is a structure to fill in, not a slide to present.

The two published tables that do most of the work

(a) HSE enforcement, FY 2024/25 - from the HSE Annual Report and Accounts 2024/25 (HC 1407), published 20 November 2025 (gov.uk):

Average fine per prosecution: ≈ £134,000. est. This is our arithmetic (£33m ÷ 246), not an HSE statistic - HSE does not publish an average. It rests on a rounded input ("over £33 million"), so treat it as a floor. Per conviction (96% of 246 ≈ 236) it is ≈ £140,000 est.

⚠️ Note for anyone updating this document: HSE stopped publishing enforcement statistics in its annual statistical release - hse.gov.uk/statistics/enforcement.htm says so directly. The only source is now the Annual Report and Accounts, which appears each November. 2025/26 figures are not yet published.

(b) The Sentencing Council definitive guideline, in force since 1 February 2016, applies to all organisations sentenced on or after that date. Turnover bands are identical for health and safety and for food safety and hygiene offences:

Health and safety offence range: £50 to £10 million. Food safety and hygiene offence range: £100 to £3 million. Both unlimited fine, triable either way. (H&S guideline · food guideline)

The two sentences in the guideline that this product speaks to directly. Among the High culpability factors:

And among mitigating factors: "Effective health and safety procedures in place", "Good health and safety record", "Evidence of steps taken voluntarily to remedy problem". The food guideline's High culpability wording is "ignoring concerns raised by regulators, employees or others".

Why that matters commercially, and it is the single best argument in this document. Culpability is one of the two axes that set the starting point. Look at what one step does on the Large-organisation H&S table at harm category 2: High culpability starts at £1,100,000; Medium starts at £600,000. One culpability step is worth half a million pounds at the starting point - and £1.4m at the top of the range (£2,900,000 against £1,500,000).

A hazard-reporting channel with a timestamped record is direct documentary evidence that concerns raised by workers were not ignored. A confirmed-receipt record after an incident is direct documentary evidence that appropriate changes were made and communicated. Neither prevents a prosecution. Both are evidence on the axis that moves the number most.

The one other rate you need

Fee for Intervention: £188 an hour, the rate from the 2026/27 fee year (hse.gov.uk/enforce/charging/rates.htm, updated 1 April 2026). HSE's own words: "£188 an hour for the time it takes HSE to identify what is wrong and to help you put things right", triggered by a material breach - "something which an inspector considers serious enough that they need to formally write to the business requiring action". Chargeable time includes the visit, writing the reports, obtaining specialist advice, post-visit correspondence and consulting workers. A single notice with the investigation around it runs to a four-figure invoice before anyone has been to court.

And the reputational clock nobody mentions

• Enforcement notices are published on HSE's public register 5 weeks after being served - before any appeal is concluded - and stay visible for 5 years (resources.hse.gov.uk/notices).
• Convictions are published 9 weeks after conviction, on the main register for 1 year, then transferred to the conviction history register for a further 9 years - ten years of public visibility (resources.hse.gov.uk/convictions).

Ten years is longer than most of the contracts these companies are bidding for. In every sector below, the client's PQQ asks about enforcement history.

The admin-time model, stated once and reused

Every sector page below uses the same arithmetic with a different site count. The assumptions, all in one place, all labelled:

⚠️ This model deliberately excludes the compliance manager's time assembling evidence for an audit, an inspection or a claim, which every customer says is the worst part and none of them measures. Ask for it; do not estimate it for them.

And say the honest thing about it. The admin saving is real but it is not why they should buy. It is the number that makes the price look reasonable to a finance director. The reason to buy is the record. Lead with the record, close with the arithmetic.

1. Construction

The shape of the downside: an HSE prosecution against a turnover band that is usually Large or Medium, on a culpability axis where "ignoring concerns raised by employees" is a High-culpability factor; plus prohibition notices that stop work on a programme with liquidated damages attached; plus a ten-year public conviction record read by every PQQ in the supply chain.

Enforcement

Take a Large contractor (turnover £50m+) and a harm category 2 offence - a serious injury risked, medium likelihood. On the published table:

The gap between High and Medium culpability at the starting point is £500,000. For a Medium turnover contractor (£10m-£50m) at the same harm category, it is £450,000 versus £240,000 - £210,000. For a Small contractor (£2m-£10m), £100,000 versus £54,000 - £46,000.

Say it as a question, not a claim: "Your barrister will be arguing culpability. One of the High-culpability factors is 'ignoring concerns raised by employees or others'. What will they hand the court to show you didn't?"

Notice costs

Approximately 1,200 notices prohibiting work activity were issued across all industries in 2024/25, and construction is consistently among the most heavily enforced sectors. A prohibition notice does the damage before any fine:

One half-day stoppage on one site costs more than five months of the £950 band. That is the sentence for the Operations Director, and unlike the fine numbers it needs no legal framing at all.

Admin time

A regional division of 12 sites (9-14 verified across five UK housebuilders): 12 × £1,702 = £20,424 a year est. in site-manager time on paper briefings, against a £2,600 Company band for 1,001-2,000 enrolled workers = £31,200 a year, or £1,600 for 501-1,000 = £19,200 a year.

Note honestly what this does and does not show: at the £1,600 band the admin saving alone roughly covers the licence; at £2,600 it does not. Do not oversell it. Say: "the admin time roughly pays for it at your headcount; everything else - the record, the languages, the subcontractors - is what you're actually buying." That is a much stronger position than a spreadsheet that suspiciously nets to positive.

So

2. Catering and hospitality

The shape of the downside: the FHRS rating, which is public, displayed, and - uniquely in this document - directly visible to the customer's customer. Plus a food-safety prosecution regime with its own sentencing table, and a due-diligence defence under s.21 Food Safety Act 1990 that the operator has to evidence.

What an FHRS drop actually costs

The rating is built from three components, each scored separately, lower being better:

CIM carries the largest single possible score of the three - 30 against 25 - and it is the most subjective. It is explicitly a judgement on "the likelihood of satisfactory compliance being maintained in the future", driven by track record, willingness to act on previous advice, complaint history, management attitude, and documented procedures plus evidence of their implementation. (FSA, Food Hygiene Rating Scheme Brand Standard, 104pp, created 6 March 2026 - PDF)

And the mechanic that makes it expensive. The rating is not a simple total. From Table 3 of the Brand Standard:

A business scoring 0, 5, 10 has a total of just 15 - which would be a 5 - but because one individual score is 10, it is capped at 4. A business scoring 5, 5, 20 (total 30) is capped at 1. A weak CIM score alone can cost between one and four rating points regardless of how clean the kitchen and the building are. That is the commercial argument in a single mechanic, and it is published.

What the drop is worth

The FSA does publish consumer data. From Food and You 2: Wave 11, FSA Official Statistics, published 31 March 2026 (fieldwork May-Aug 2025, n=5,898 adults) (gov.uk):

A majority of consumers who know the scheme will not accept below a 4. That is an official statistic, it is four months old, and it is the strongest FHRS line available.

⚠️ What we will not claim. This is stated preference in a survey, not observed trading behaviour. It is not a revenue-impact measurement, and there is no published £ or % figure for what a drop costs an operator. We looked for the Welsh and Northern Irish mandatory-display evaluations, which would be the obvious source, and could not locate a live published evaluation quantifying commercial effect. Do not put a revenue number on an FHRS drop. Put the 52% on the table and let them put the revenue number on it - they know their covers.

The cost of getting the rating back

• A re-rating visit can be charged for. FSA, verbatim: "Yes… powers available to local authorities in England under the Localism Act 2011 allows for the recovery of costs of re-inspections/re-visits made at the request of the FBO." Each authority sets its own charge. ⚠️ There is no national fee figure and we could not verify one - it must come from a named authority's schedule.
• The free route is slow. Where no fee is charged, a three-month standstill applies and the re-visit must happen within three months of it ending: "six months is the maximum amount of time a business should have to wait", and one requested re-visit per planned intervention. Where a fee is paid, there is no standstill, the visit happens within three months, and there is no limit on requests.
• It can go down. FSA: "the food hygiene rating could go up, down or remain the same."
• The appeal window is "in writing within 21 days of being notified about your food hygiene rating. This period includes weekends and public holidays."

Six months of a displayed 2 across a unit, on a scheme 92% of the public recognise, in exchange for not being able to evidence management control. That is the sentence.

Prosecution

Food safety and hygiene, Large organisation (£50m+), harm category 2: High culpability £230,000 (range £90,000-£600,000) against Medium £90,000 (range £35,000-£220,000). One culpability step, £140,000. For a Medium turnover caterer at the same category: £90,000 against £35,000 - £55,000. The food guideline's High-culpability wording is "ignoring concerns raised by regulators, employees or others".

Admin time

A contract caterer with 40 kitchens: 40 × £1,702 = £68,080 a year est. of unit-manager time on paper briefings, against the £950 band (251-500 workers) at £11,400 a year.

So

3. Logistics and warehousing

The shape of the downside: HSE enforcement on a workforce with one of the highest agency shares in the economy, plus the O-licence, which is the only regulatory instrument in this document that attaches personally to a named individual and can remove the operator's ability to trade at all.

Enforcement

Warehousing incidents - forklift and pedestrian interactions, falls from vehicles, load shift, racking collapse - land squarely in harm categories 1 and 2. For a Large 3PL (turnover £50m+), harm category 1:

£1.1 million between High and Medium at the starting point. For a Medium turnover operator: £950,000 against £540,000 - £410,000.

The agency-labour point is the one to press. Where a substantial share of a shift is agency, and the only channel to them is the shift manager's phone, the question "how do you evidence that the agency picker received the racking alert?" has no good answer today - and MHSWR reg 12 and COSHH reg 12(4) both say the duty reaches them anyway.

Notice costs

That last row is the one for the Commercial Director, not the SHEQ lead. A notice appears publicly before the appeal is heard, and stays five years - across the life of most 3PL contracts.

Admin time

A 25-depot operator: 25 × £1,702 = £42,550 a year est. against the £1,600 band (501-1,000 workers) at £19,200 a year or the £2,600 band at £31,200.

So

4. Facilities, cleaning and security

The shape of the downside: this sector's money is not primarily in fines. It is in contract retention, because the whole business is contracts, margins are thin, and the client's own compliance team audits the evidence. The regulatory exposure is real but secondary.

Retender and mobilisation loss - estimate, clearly labelled

⚠️ Everything in this subsection is an estimate. There is no published dataset of UK contract-cleaning retender win rates or the cost of losing one. What follows is a structure to fill in with the customer's own numbers in the meeting - and doing it that way is far more persuasive than presenting our arithmetic anyway.

Losing one mid-sized contract at retender is worth roughly £54,000 of gross profit est. - about five years of the £950 band. And a single scoreable line in a bid - "every operative confirmed receipt of the method statement and the safety data sheet, in their own language, and we can produce the per-person record" - is a real differentiator in a market where every bidder claims the same things and none can evidence them.

This is why the pitch in this sector is a win-rate pitch, not a cost pitch. Take the export to the Operations Director and the bid lead together and ask whether it could go in the next tender. If the answer is yes, you are not selling software.

Enforcement, secondarily

Cleaning and security contractors are frequently Medium or Small turnover bands. At harm category 3, Medium turnover: High culpability £210,000 against Medium £100,000 - £110,000 on one step. COSHH exposure - chemical changes, dilutions, new safety data sheets, delivered to a night-shift workforce that frequently does not read English - is the specific risk, and it is exactly what a translated, confirmed post is for. Add £2,256 est. of Fee for Intervention per material breach at £188/hour.

Admin time

Cleaning contractors have supervisors rather than site managers, and the ratio is different. A 700-operative contractor with 35 supervisors each running a briefing round: 35 × £1,702 = £59,570 a year est., against the £1,600 band (501-1,000) at £19,200 a year - £2.29 per operative per month.

Here the admin arithmetic genuinely does stand on its own, which is unusual in this document. Say so, and say it plainly, because in a 4%-margin business it is the number the Finance Director cares about.

So

5. Food manufacturing

The shape of the downside: two regimes stacked. The statutory one (food safety plus machinery safety, both with published sentencing tables) and the commercial one (the retailer's technical standard), which is faster, harsher and entirely private.

Statutory

Food manufacturers are usually Large or Medium turnover. Food safety, Large, harm category 1: High culpability £500,000 (range £200,000-£1,400,000) against Medium £200,000 (range £80,000-£500,000) - £300,000 on one culpability step, with an offence range reaching £3 million.

And separately, health and safety: food manufacturing has one of the worst machinery-injury records in UK industry, and guard removal, lock-out/tag-out and clean-down instructions are exactly the kind of instruction that must reach a mixed-language, multi-shift workforce. Large turnover, harm category 1: High £2,400,000 against Medium £1,300,000 - £1.1 million on one step.

Plus £2,256 est. of Fee for Intervention per material breach.

Retailer delisting - estimate, clearly labelled

⚠️ This is the biggest number in this document and the least verifiable. The BRCGS Global Standard is paywalled, retailer technical standards are confidential commercial documents, and no supermarket publishes delisting criteria or the commercial consequences. Nothing in this subsection is a published figure. It is a structure to complete with the customer, and doing it in front of them is the whole point - because they know the number and it frightens them.

And the mechanism that connects it to this product. BRCGS Issue 9 made pre-operational training mandatory for all staff and requires a documented record of training needs for every employee, with competence tested for control measures including allergen controls and CCPs; clause 7.1 is a critical clause, so a non-conformity has certification consequences. ⚠️ This summary is drawn from BRCGS's own published guideline material and secondary interpretation, not from the paywalled Standard itself. Confirm the clause wording with the customer's technical team before putting it in writing. (BRCGS Issue 9 interpretation guideline, sample PDF)

The honest framing: an audit non-conformance on training records does not cost £8m. It costs a corrective action with a deadline, reported to the retailer. But it is the first step on a path whose end the Technical Director can see, and it is the reason the Technical Director will champion this and the Operations Director will sign it.

Admin time

A three-site manufacturer, but with shift leaders rather than a single manager - assume 9 briefing owners (three shifts × three sites): 9 × £1,702 = £15,318 a year est. Against a £1,600 band at £19,200. In this sector the admin arithmetic does not carry the purchase, and you should say so before they work it out: "the time saved doesn't pay for this. The audit evidence does."

So

6. Care and housing

The shape of the downside: in care, a regulator that inspects records and publishes ratings, with published fixed-penalty amounts and unlimited fines on prosecution. In housing, the newest statutory clock in this document.

CQC - published penalty amounts

Unusually, CQC publishes exact fixed penalty notice amounts (cqc.org.uk, penalties, fines and offences):

On prosecution, fines are unlimited for failure to comply with quality-and-safety regulations, carrying on unregistered, breach of registration conditions, and suspension or cancellation offences. LASPO 2012 removed the previous £5,000 summary cap. There is a defence of all reasonable steps and due diligence under reg. 22(4) - which is, once again, a defence the provider has to evidence.

Note the second column. The registered manager is personally liable, at £2,000 for a quality-and-safety breach and £625 for a notification failure. That is not an abstraction to a Registered Manager, and it is why the Registered Manager - who is the adoption gatekeeper in every care pilot - has a personal reason to want this to work.

The rating

⚠️ We could not verify, and will not claim, any figure for the effect of a "Requires Improvement" or "Inadequate" rating on local-authority placements or embargoes. CQC publishes no such dataset, and no CQC, DHSC or ADASS publication quantifying it could be located. Placement suspensions are decided and recorded by individual local authorities and are not centrally collected. Likewise, CQC does not publish aggregate counts or values of fixed penalty notices issued or prosecutions brought - those appear only in narrative form in State of Care (latest edition published 24 October 2025).

So do not put a number on it. Ask the question instead: "What happened to admissions at the last home in your group that went to Requires Improvement, and how long did it take to come back?" They will tell you, it will be specific, and it will be their number rather than ours - which is better.

Housing repairs

The Hazards in Social Housing (Prescribed Requirements) (England) Regulations 2025 - "Awaab's Law" - have been in force since 27 October 2025, with prescribed statutory timescales for investigation, a written summary to the resident, emergency action and making safe (legislation.gov.uk). The enabling power is s.42 Social Housing (Regulation) Act 2023, enforceable "through actions for breach of the covenant" (legislation.gov.uk), with a defence available only where the landlord "used all reasonable endeavours".

"All reasonable endeavours" is an evidential standard, and the clock starts when the resident reports. Every hour a repairs operative does not know about a case is an hour off a statutory clock, and the defence requires the landlord to show what it did. We are not aware of published data on the cost of a breach, and do not estimate one.

Admin time

A 20-home care group: 20 × £1,702 = £34,040 a year est. of registered-manager time, against the £2,600 band (1,001-2,000 workers) at £31,200 a year - or £180 per home per month on the Site plan, which at 20 homes is £3,600 a month, £43,200 a year. Quote the band.

The specific saving here is sharper than elsewhere, because it removes work from the single person whose cooperation the whole deployment depends on: "today your registered manager chases fourteen people by phone. Tomorrow they open a list of the four who haven't confirmed."

So

7. The one-page version

Put this on a single slide and let the buyer pick their own row.

8. What the evidence does not contain

There is no credible published study linking workforce safety communication, briefings or toolbox talks to a quantified reduction in incident rates. This was searched for properly and the negative result is reported as a finding, not hidden.

What was checked:

• HSE's Research Report (RR) series, all twelve range pages, RR1-RR1154 (hse.gov.uk/research/rrhtm). Relevant titles exist - RR306 Investigating practices in communication, RR516 An investigation of approaches to worker engagement, RR660 Behaviour change and worker engagement practices within the construction sector, RR785 Improving health and safety: an analysis of HSE's risk communication in the 21st century, RR880/RR1067 Leadership and Worker Engagement Toolkit, RR964, RR1031, RR1089 and others. None is titled or framed as a measurement of incident-rate change attributable to communication or briefings; they are process evaluations, qualitative studies, toolkit development and case studies. ⚠️ Individual abstracts could not be retrieved (reports older than six years are on The National Archives mirror, which blocks automated retrieval), so this is title-level evidence only and no claim is made about what any individual report concluded.
• IOSH's research library (iosh.com/research) - the library is JavaScript-driven with client-side filtering, so only the default listing was visible. Nothing linking communication to incident rates surfaced. Not exhaustively checked.
• Peer-reviewed literature. A PubMed search for the exact phrase "toolbox talks" returns six indexed papers in total: Kearney et al. (2025) Front Public Health 13:1690149; Kara & Sönmez (2025) J Nurs Scholarsh 57(3):439-451; Somerset, Evans & Blake (2021) IJERPH 18(8):4184; Eggerth, Keller & Cunningham (2018) Am J Ind Med 61(12):997-1004; Kaskutas, Jaegers & Dale (2016) Professional Safety Jan:33-37; Kaskutas et al. (2013) J Safety Res 44:111-118. None of the six measures incident or injury rates as an outcome. They measure knowledge, recall, engagement, compliance behaviour, or - in the 2025 nursing paper - quality indicators in a non-safety domain. The 2013 paper is explicitly a pilot.
• The one number that circulates. A 2022 paper in Occupational Health Science (doi:10.1007/s41542-022-00118-8) found that texting toolbox talks to residential construction supervisors improved supervisors' safety-meeting compliance by 19.39%. ⚠️ That is an adherence measure, not an incident-rate measure. If it appears in a sales document as an injury-reduction figure it is a misrepresentation. Do not use it.
• A 2025 US scoping review reaching the same conclusion - positive findings on worker knowledge, effectiveness never systematically evaluated against injury outcomes - was surfaced in search but is not indexed in PubMed and its citation could not be verified. Do not cite it.

The defensible adjacent claims, if pressed, are that safety climate correlates with injury outcomes in meta-analytic work, and that supervisor-to-worker communication quality is associated with safety climate. Both are correlational, not causal, and neither is about toolbox talks specifically. ⚠️ The standard citations (Christian et al. 2009; Clarke 2006) were not verified in this pass - verify before printing either.

So what do you say when a buyer asks whether this reduces accidents? Say this:

That answer will win you more credibility than any statistic, and it is the only one that is true.

Sources

Enforcement and sentencing: HSE Annual Report and Accounts 2024/25 (HC 1407), published 20 Nov 2025 · Sentencing Council, health and safety offences - organisations · Sentencing Council, food safety and hygiene offences - organisations (both in force 1 Feb 2016) · HSE Fee for Intervention rates, updated 1 Apr 2026 · HSE convictions register · HSE notices register

Food: FSA, Food Hygiene Rating Scheme Brand Standard, created 6 Mar 2026 (PDF) · FHRS guidance for businesses, updated 30 Jun 2026 · Food and You 2: Wave 11, published 31 Mar 2026 · BRCGS Issue 9 interpretation guideline (sample)

Care and housing: CQC penalties, fines and offences · CQC enforcement policy · CQC, using CQC data, updated 5 Aug 2026 · Hazards in Social Housing (Prescribed Requirements) (England) Regulations 2025 · Social Housing (Regulation) Act 2023 s.42

Wages: ONS, Annual Survey of Hours and Earnings, April 2025, published 23 Oct 2025

All URLs verified live on 28 August 2026. Several sources in earlier drafts had moved: HSE's enforcement-statistics page no longer carries statistics, the Sentencing Council restructured its guideline URLs, food.gov.uk migrated its guidance to GOV.UK, and CQC moved its enforcement-policy paths. Re-check before quoting in writing.

The trigger feed: a weekly ranked call list from free public data

A weekly ranked call list, built from free public feeds.

Every trigger event in the sector playbooks is currently discovered by asking - which means you must already be in the conversation to find out that the conversation is worth having. This document inverts that. It turns the 200 named accounts on the target lists into a ranked weekly call list of maybe fifteen to thirty companies where something has just happened, so that Monday's outreach references a dated, verifiable, public event rather than a hopeful hypothesis.

Everything below was tested on 28 August 2026. Endpoints were called, responses inspected, and the things that turned out not to exist or not to be free are called out as plainly as the things that work. Two of the seven are worse than they look on paper; one does not usefully exist at all. Read §8 before building anything.

Companion documents: the sector playbooks (this chapter) · the construction playbook (this chapter) · the 200 targets · the 200 targets · verification-log.md · the value models (below).

1. The seven feeds at a glance

If you only build two things, build Find a Tender and Contracts Finder. They give you dated, valued, company-number-identified events, with no scraping, no key and no per-request cost.

2. Feed by feed

2.1 Find a Tender - the strongest single feed

Why it is the best feed here: it carries the supplier's Companies House number, natively. A real record from the test pull:

``json {"name":"NATURAL WORLD PRODUCTS LIMITED","identifier":{"scheme":"GB-COH","id":"NI023268"}} ``

That means an award notice can be joined to the target list exactly, with no fuzzy matching. Buyers carry the post-Procurement-Act GB-PPON identifiers. Also available per release: awards[].value (amount, amountGross, currency), address.postalCode, tenderPeriod / awardPeriod / awards[].date, CPV codes via tender.items[].classification, tender.aboveThreshold, mainProcurementCategory, procurementMethod, lots.

⚠️ Use the API, not the website. find-tender.service.gov.uk/Notice/Search returns 403 to a script - the HTML UI has bot protection. The API does not.

2.2 Contracts Finder - the easiest to ingest

How it differs from FTS after the Procurement Act 2023 (in force 24 February 2025): FTS is the above-threshold central platform carrying the new notice types and GB-PPON identifiers; Contracts Finder is the legacy below-threshold and general UK service. They overlap. Build on FTS as primary for the identifiers; keep Contracts Finder for lower-value and legacy coverage, and because a daily CSV is a fifteen-minute integration. Its own API help page flags "Contracts Finder API is updating. Version 2..." - expect churn.

2.3 Companies House - the join key for everything else

What to use it for: (a) resolve every target to a company number once, permanently; (b) watch the streams for new charges (new lending, often new capex), insolvency (disqualify), officer changes (a new HSEQ or Operations Director is the most buyable human in any of these sectors - see the construction playbook's trigger table); (c) enrich anything the other feeds hand you as a bare name.

2.4 FSA Food Hygiene Rating Scheme - free, keyless, and the easiest win in catering

⚠️ Two gotchas. First, an unfiltered query returns HTTP 403 with {"Message":"This is a CPU intensive query: please use one of the documented filters..."} - that is a throttle, not an auth wall. Always pass at least one filter. Second, there is no company number and no SIC code, and BusinessName is frequently a trading name, not the registered entity - so joining to the target list is name-plus-postcode fuzzy matching. Budget for that.

Why the sub-scores matter more than the headline. ConfidenceInManagement is the most subjective and most management-facing of the three components, and it is exactly what a per-person briefing record speaks to. A site whose overall rating held but whose ConfidenceInManagement score worsened is a better call than one that simply dropped to a 3, and nobody else is watching that field.

2.5 CQC - free bulk files, API now needs a key

The API changed and the old advice is wrong. api.cqc.org.uk is retired (404). The current host is https://api.service.cqc.org.uk/public/v1/ and it is behind Azure API Management: an unauthenticated call returns a misleading HTTP 502, and with any key header the true error surfaces as 401 - "Access denied due to invalid subscription key", with www-authenticate: AzureApiManagementKey ... name="Ocp-Apim-Subscription-Key", type="header".

• Get a key at api-portal.service.cqc.org.uk (sign up, then Products / APIs). ⚠️ Flagged as unverified: the products page sits behind sign-in, so free-ness could not be confirmed without creating an account. CQC has historically published this under the Open Government Licence and states that the reason for authentication is usage analytics and change notification rather than charging - so it is very likely free, but check before relying on it.
• The API's useful primitive for a trigger feed is /changes/provider and /changes/location - timestamp-windowed deltas. It also carries linked organisations including the previous provider, which makes ownership changes detectable.

The bulk spreadsheets are free, keyless, and were verified downloading:

Directory columns: Name, Also known as, Address, Postcode, Phone number, Website, Service types, Date of latest check, Specialisms/services, Provider name, Local authority, Region, Location URL, CQC Location ID, CQC Provider ID. No company number, but Provider name is usually the registered company name ("Midshires Care Limited"), so Companies House matching is far more tractable here than with FSA.

⚠️ CQC's own live data-quality warning, and it bears directly on using ratings as a trigger: CQC is migrating systems, which has "caused some delays generating the files" and "some updates are currently taking longer than usual to be reflected in the data... For example, the cancellation of provider registrations." Multi-service locations now show as "Not Rated" in the filters file under the new assessment approach; their ratings sit only in the ratings file. Do not treat a "Not Rated" as a signal.

2.6 HSE convictions and notices - the highest-intent trigger, the worst plumbing

Both registers are live, free and public. Neither has an API, an export button, or an open-data release. Be honest with yourself about the cost.

Verdict: usable, and worth it - an improvement or prohibition notice is the single highest-intent trigger in the construction and logistics playbooks - but it is an HTML scraper you build and maintain, paginated ten results at a time. Budget it as an engineering project, not a data source. The pragmatic v1 is a weekly name-search against your 200 targets (200 name queries a week is trivial) rather than a full crawl of 30,000 records.

2.7 Street Manager - free, but push-only

⚠️ The architecture is the catch, and it is a big one. This is AWS SNS publish/subscribe - push, not pull. There is no REST endpoint you can GET. You must operate a public HTTPS POST endpoint able to receive requests from AWS's IP range, and confirm the subscription by calling the SubscribeURL in the SubscriptionConfirmation message within three days or re-register. Filtering happens on your side after receipt. That is materially more engineering than every other feed here.

Topics (prod, eu-west-2): prod-permit-topic, prod-activity-topic, prod-section-58-topic. Eighteen event types, including work start/stop, permit submitted / granted / refused / revoked / cancelled, permit alteration granted, activity created/cancelled/updated, and the Section 58 lifecycle.

Fields: promoter_organisation, promoter_swa_code, highway_authority, work_reference_number, permit_reference_number, works_location_coordinates, street_name, area_name, usrn, work_category, traffic_management_type, proposed and actual start/end dates, work_status, activity_type, event_time. So: contractor names yes, geography yes, dates yes. No company number, no postcode, no SIC, no value.

❌ There is no open-data S3 bucket of daily or monthly exports. The "data export" feature inside Street Manager is a reporting tool for registered highway authorities and works promoters - parties to the works - not an open-data release. Do not plan around a downloadable archive.

⚠️ The docs warn that the message specification "is subject to extension/modification with no prior notice" and recommend extracting only the properties you need rather than deserialising the whole message.

3. Filtering to the 200 targets

The whole feed is worthless as an undifferentiated firehose. Everything below assumes one preparatory step, done once:

Step 0 - resolve every target to a Companies House number, once. Take the 200 rows in targets-100.csv and targets-nonconstruction-100.csv, and for each one record: the group company number, the numbers of the trading subsidiaries you would actually sell to, the registered postcode, and the trading names the operating sites use. This is a day's work and it is the foundation of everything else. verification-log.md already did the equivalent exercise against two registers and is the model to follow; note that its own attempt at Companies House found that recent accounts are filed as image-only PDFs with no text layer, so read the API's structured company profile, not the filings.

Then, per feed:

A rule worth writing down. Four of the seven feeds carry no company number, so entity resolution is the real build, not the API calls. Keep a single aliases table mapping every observed name string to a company number, and grow it by hand every week when you review the list. After two months it stops being work.

4. Event → playbook trigger mapping

This is the table that turns a data row into a sentence you say on the phone. The playbook column tells you which document to open before dialling.

Ranking rule. A = something bad or big has just happened to them, with a date; call within five working days. B = something has happened that creates a window; call within two weeks. C = context that improves a call you were making anyway; never a reason to call on its own.

The tie-breaker, in order: (1) an A-rank event beats everything; (2) among A-rank, prefer the account where you can name the champion; (3) then the account whose band is £950 or above; (4) then the account with a warm path in the target list; (5) then recency.

5. The weekly routine

Ninety minutes, Friday afternoon, so that Monday starts with a list rather than a decision. Do it in this order.

1. Pull (20 min, mostly automated). Yesterday's Contracts Finder daily CSV; FTS ocdsReleasePackages?updatedFrom= the last run's timestamp; the FSA bulk CSV or the per-LA XMLs for the authorities your targets sit in; the Companies House stream buffer for the week; the CQC ratings file if it has changed; and a name-search pass over the HSE notices register for your 200 names.
2. Match (15 min). Join on company number where you have one; run the fuzzy matches and dump anything ambiguous into a review queue rather than the call list. Add new aliases to the alias table as you go - this is the fifteen minutes that makes next week faster.
3. Rank (10 min). Apply the A/B/C rule and the tie-breakers. Cap the list at 25. A list of sixty is a list of zero.
4. Enrich (20 min). For each of the top 25: open the relevant playbook section, write one sentence naming the event and the date, and confirm the named human you are calling. If you cannot name the human, the account drops to next week.
5. Load (10 min). Create or update the deal record, set the task for Monday, paste the sentence into the notes.
6. Review last week (15 min). Which triggers actually produced a conversation? Kill any trigger type that has produced nothing in six weeks. Especially do this for the expensive feeds - if the HSE scraper has not produced a meeting in two months, it is not earning its maintenance.

Two discipline rules. First, an event without a named human is not a lead - it is a research task. Second, never reference a bad event in the first line of the first message. "I saw you got a prohibition notice" reads as ambulance-chasing and it will end the relationship before it starts. The event tells you when to call and what they are thinking about; it is not the opening line.

6. Tooling: start with a spreadsheet

You are one person. Do not build a data platform.

Stage 1 - a spreadsheet and six scheduled scripts (week one, and it will last months)

• One Google Sheet, four tabs: targets (the 200, with company numbers, aliases, postcodes, band, champion name, warm path), events (append-only: date, feed, company number, event type, one-line description, source URL), aliases (name string → company number), calllist (this week's ranked 25, generated by formula from events).
• Six small scripts on a cron or a scheduled GitHub Action, writing rows into events via the Sheets API: Contracts Finder daily CSV; FTS OCDS; FSA bulk CSV diff; CQC ratings file diff; Companies House stream consumer; HSE notices name-search. Write the first three; add the rest only when the first three have produced a meeting.
• The whole of stage 1 fits in about 300 lines of Python and costs nothing to run.

Stage 2 - HubSpot, once you have deals to lose track of (the free tier is enough)

• Create a custom object or, more simply, custom properties on Company: last_trigger_type, last_trigger_date, last_trigger_url, trigger_rank.
• Push events from the same scripts via the HubSpot API. Set a workflow that creates a task when trigger_rank = A.
• Use a saved view sorted by last_trigger_date as the call list, and retire the calllist tab.
• Do not attempt bidirectional sync, deduplication rules or a data warehouse. The value is entirely in the ranked list; everything else is procrastination with a plausible cover story.

What to log for every call

Company number, trigger type, trigger date, whether you reached the named human, and whether the trigger was mentioned by them unprompted. That last field is the only real measure of whether a feed is working, and it costs nothing to record.

7. Planning approvals: the honest answer

There is no free national feed of individual planning decisions in England. This was tested rather than assumed, and the negative result matters more than the positives above, because "planning approvals for large sites" is the intuitive trigger for construction and it is the one that does not exist.

• planning.data.gov.uk is free, keyless and genuinely good - 25,353,102 entities across 339 datasets under OGL v3, with a clean API (?dataset=, ?limit=, ?offset=, JSON/GeoJSON/CSV). But almost every one of those 339 datasets is a planning constraint or designation: conservation areas, article 4 directions, flood risk, green belt, listed buildings, brownfield land, tree preservation orders, local plans. Excellent for site-suitability work. It is not a decisions feed.
• The planning-application dataset exists and is not usable. The platform's own dataset page says: "The planning application dataset is incomplete and is not yet ready for use." Phase: alpha/beta. Provider note: "Data created by MHCLG. We will replace this with data from authoritative sources when it is available." It holds 100,627 entities against roughly 400,000+ planning applications per year in England, and a sample across twelve offsets spanning the whole dataset found only four local planning authorities represented, with Camden accounting for the overwhelming majority. The schema is right; the coverage is a four-council pilot.
• The planning-decision dataset is empty - entity.json?dataset=planning-decision returns {"entities":[],"count":0}. It is a schema placeholder.
• Planning application statistics on gov.uk are aggregates only - MHCLG live tables and the PS1/PS2 district returns, quarterly counts by authority. Useful for market sizing; zero individual applications, no addresses, no applicant names.
• infrastructure-project-decision does exist and is national and free, but it covers only NSIP / DCO projects - a few dozen live nationally significant infrastructure schemes, not ordinary large sites.

So the realistic options are: (a) scrape ~330 local planning authority portals individually - Idox, Northgate, Arcus, no common API, many with bot protection - which is high cost and permanently breaking; (b) pay a commercial aggregator (Barbour ABI, Glenigan, LandTech, Landmark), which is a real budget line and should be treated as one, not smuggled in as "a free feed"; or (c) use planning.data.gov.uk for constraints and NSIP decisions only, and get your construction site-start signal from Street Manager permits and FTS award notices instead.

MHCLG is actively building towards a national dataset, so this may change. Re-check before the next planning cycle. As of 28 August 2026 it is a four-council alpha and you should not build on it.

8. What is not as good as it looks

Six things to say out loud before anyone commits engineering time.

1. No free national planning-decisions feed for England. See §7. The intuitive construction trigger is the one that does not exist.
2. HSE has no API, no open data and no export. The registers are free and public but HTML-only and paginated ten rows at a time. Any monitoring is a scraper you own forever. Start with a weekly 200-name search, not a crawl.
3. CQC's API now requires a subscription key, the old api.cqc.org.uk host is retired, and an unauthenticated call misleadingly returns 502 rather than 401. ⚠️ Whether the key is free could not be confirmed without creating an account - very likely yes, historically OGL, but verify before relying on it. The bulk spreadsheets are unambiguously free and cover most of the need.
4. Street Manager is push-only with no export bucket. You must register and operate a public HTTPS receiver for AWS SNS, and confirm the subscription within three days. Its message spec can change without notice.
5. Four of the seven feeds carry no company number (FSA, CQC, HSE, Street Manager). Entity resolution against Companies House is the real build. Plan for a hand-maintained alias table and accept that it will never be perfect.
6. "Daily" is not daily everywhere. The FSA bulk job runs daily; individual local authorities publish on their own schedule and some were five weeks stale on the test date. CQC has an open data-quality warning about migration delays, including to registration cancellations, and multi-service locations now show as "Not Rated" in the filters file for reasons that have nothing to do with their actual rating. Never treat an absence as a signal.

And one thing that is better than it looks: Find a Tender carries the supplier's Companies House number natively, which makes public-sector contract awards the only trigger in this document that needs no fuzzy matching at all. That is where to start.

9. Verification note

Every endpoint, file and claim in this document was tested on 28 August 2026: HTTP status codes checked, responses inspected, record counts and file sizes read off real downloads, and negative results (planning decisions, HSE export, Street Manager bucket, CQC free-ness) confirmed by looking rather than assuming. Two items remain unverified and are flagged in place: whether the CQC API subscription key is free, and HSE's actual publication lag to the notices and convictions registers, which is nowhere stated. Feeds change. Re-test the whole set before any quarter in which you plan to spend money on it.