Sliced bread, rolls and a free-from range for retailers' own-label and foodservice wholesalers, baked on one site on the edge of a market town in the East Midlands. 320 staff.
Verdict: The programme covers 3 scenarios (2 crisis, 1 promotion); all have an approved message house, enough hard questions with a hostile one, and a practised spokesperson. 19 approved answers are in the Q&A bank. At the baseline, 6 recorded interviews were scored blind by two scorers (criterion scores within a point 97%): the average was 2.95 of 5 and 0 of 3 spokespeople were ready to speak against a pass mark of 3.5 and no criterion below 3.0. The media protocol is signed off and covers 6 of the 6 required kinds. How the value is shown (risk avoided, not a forecast): if an interview or statement in “Milk found in a 'dairy-free' loaf: a recall” went badly it would cost £615,000 from the owner's own figures, though after a recall most of the loss follows the recall and the technical response, not the interview. With the owner's assumed likelihoods and effect, the expected loss across the 2 crisis scenarios is £160,680 a year without the programme and £150,600 with it. If the programme works half as well it saves £5,040 a year, which does not cover its yearly cost of £7,208 in full (outside fees, consultant's fee and team time, the build spread over 3 years); as assumed it saves £10,080. It feeds 3 catalogue products (Unlocks). Re-measure the skills on 29 Jan 2027: direction only: fewer than 6 spokespeople.
| Scenario | Core message | Hard questions | Limits | Complete |
|---|---|---|---|---|
| S1 Milk found in a 'dairy-free' loaf: a recall Crisis · speaks: Kwame | Anyone with a milk allergy should not eat this loaf. We stopped supplying it [Confirm: same day] and will find out why. 22 words | 7 2 hostile | L10 · L2 · L5 · L6 | Yes |
| S2 Objections to the new despatch building (planning) Crisis · speaks: Owain | We want to grow here and keep 320 jobs local, and the new building takes our lorries off Tanyard Lane at night. 22 words | 7 2 hostile | L3 · L9 | Yes |
| S3 Our free-from range in the trade press Promotion · speaks: Fiona | Free-from bread should taste as good as any other loaf, so we built a separate line to make it safely. 20 words | 5 1 hostile | L8 · L4 · L12 · L14 | Yes |
Core message. Anyone with a milk allergy should not eat this loaf. We stopped supplying it [Confirm: same day] and will find out why.
| Supporting point | Proof |
|---|---|
| We acted at once | Despatch records trace every pack by batch code; withdrawal started within [Confirm: hours] of the test result |
| Telling people clearly comes first | A recall notice on the FSA template, notices in store and online, and a customer line: [number] |
| We will find out why and stop it happening again | Allergen controls audited by our certification body every year (last audit [Confirm: date and grade]); a root-cause review shared with the local authority |
The regulator hears first. The local authority food safety team at once, and the FSA as soon as possible (serious, and the loaf has reached consumers)
Then. The retailers' technical teams and the recall insurer, by the times on the plan's deadline map; then staff; then the media. The public warning itself is never held back for anyone's sign-off
What we won't say. Anything that waters down the warning; who on the line was involved; any consumer's or customer's health details; the cause before the investigation finishes; anything admitting liability
L10 (law): Nothing that delays, waters down or contradicts a safety warning: a recall notice or 'do not eat' warning goes out as soon as the facts support it and is never held back for an insurer's, a customer's or the media's sign-off. Assimilated Regulation (EC) 178/2002, Article 19(1) and (3) (withdraw unsafe food immediately, inform the competent authorities immediately, recall from consumers where it may have reached them); legislation.gov.uk; GOV.UK 'Report a food incident to the FSA' (published 25 Jun 2026); checked 16 Oct 2026
L2 (law): Personal data about staff, customers or the public, above all health data (who was ill or hurt, who made the error). UK GDPR Article 5(1) and Article 9 (data concerning health is special category data); ICO 'Special category data' (updated 28 Oct 2024); checked 16 Oct 2026
L5 (our rule): The cause of an incident, or who is to blame, before the investigation (ours, the local authority's or HSE's) has finished. Our rule, on our solicitor's advice: say what we are doing, not why it happened, until the investigation reports
L6 (policy or contract): Anything admitting legal liability without the insurer's consent. The liability and product recall policies' claims conditions (read your own schedule)
Signed off by Fiona on 14 October 2026. From the plan: Crisis Response Plan S1 (allergen recall).
| Question | Approved answer |
|---|---|
| How many children have been made ill by your bread? Hostile | I can't talk about anyone's health, and I won't guess at numbers. What matters now is that anyone with a milk allergy doesn't eat this loaf, and we're making sure every shop takes it off sale. bridges to: core message · 36 words |
| Wasn't this just a matter of time, given you make normal bread on the same site? Leading | Our free-from loaves are made on a separate line in their own room. This batch should never have left us, and we stopped it [Confirm: the same day we found it]. We'll publish what we change once the review is done. bridges to: point 1 · 41 words |
| Who made the mistake? Personal | That's for our investigation, not for blame in public. What I can tell you is what we're doing: every pack is traced by batch code and the review is shared with the local authority. bridges to: point 3 · 34 words |
| Was it a cleaning failure or a labelling error? Speculative | I won't guess at the cause before the investigation is finished; that wouldn't be fair to anyone. When we know, we'll say, and we'll tell the people who buy our bread first. bridges to: point 3 · 32 words |
| Will you compensate people who were affected? Follow-up | Anyone who bought the loaf can get a full refund from the shop. If someone has been affected, we want to hear from them on our customer line, [number], and we'll talk to them directly. bridges to: point 2 · 35 words |
| Your retailer customers must be furious. Are you about to lose the contracts? Off-topic | Our customers' technical teams heard from us first, before any announcement, and we're working with them on the recall now. Our job today is to get every pack back. bridges to: point 2 · 29 words |
| Can you guarantee this won't happen again? Hostile | No food business can promise nothing will ever go wrong. What I can promise is that we'll find out exactly why this happened, change what needs changing and say publicly what we've done. bridges to: point 3 · 33 words |
| Person | Speaks | Signs off | Baseline interviews | Baseline average | Ready to speak |
|---|---|---|---|---|---|
| Fiona Managing director | Yes | Yes | 2 | 3.45 | Not yet |
| Kwame Technical director | Yes | No | 2 | 2.75 | Not yet |
| Owain Operations and site director | Yes | No | 2 | 2.65 | Not yet |
| Morag Finance director (signs off when Fiona is away) | No | Yes | – | – | – |
| Rhiannon Marketing manager (runs the enquiry log) | No | No | – | – | – |
Average of both scorers across the six baseline interviews, by criterion. Below the floor of 3 At or above it
| Spokesperson and format | Scorer A | Scorer B | Average | Lowest criterion | Within a point | Verdict |
|---|---|---|---|---|---|---|
| Fiona Radio · S3 · 29 Sep | 3.8 | 3.8 | 3.80 | 3.0 | 5 of 5 | Ready |
| Fiona Doorstep · S1 · 29 Sep | 3.0 | 3.2 | 3.10 | 2.0 | 4 of 5 | Not yet |
| Kwame TV · S1 · 29 Sep | 3.0 | 3.0 | 3.00 | 2.0 | 5 of 5 | Not yet |
| Kwame Hostile · S1 · 30 Sep | 2.4 | 2.6 | 2.50 | 2.0 | 5 of 5 | Not yet |
| Owain Public meeting · S2 · 30 Sep | 2.8 | 3.0 | 2.90 | 2.0 | 5 of 5 | Not yet |
| Owain Hostile · S2 · 30 Sep | 2.4 | 2.4 | 2.40 | 1.5 | 5 of 5 | Not yet |
97% within a point, against 78% by chance (0.85 beyond chance); verdicts agree on 100%, against 72% by chance. Rhiannon works under all three directors, so her scores are not independent and Bronagh’s carry the weight. Baseline recordings are deleted nine months after recording (first 29 Jun 2027), re-measure recordings a month after, and scores kept until 29 Jan 2028. No scores are used for appraisal or discipline.
| Rule | What everyone does | Owner |
|---|---|---|
| Who may speak | Only Fiona, Kwame and Owain speak to journalists, on air, at a public meeting or on camera, each for the scenarios their message house names. Anyone else who is asked takes a name, a number and the deadline, and says: 'I'll make sure the right person calls you back.' | Fiona |
| Sign-off | Fiona signs off every statement and every new answer before it is used; Morag when Fiona is away. Facts in a statement are checked against the message house; anything new goes into the Q&A bank once approved. | Fiona |
| Holding statements | In a crisis, the first public words are the Crisis Response Plan's approved holding statement for that scenario, with the [Confirm: ...] facts filled in and signed off; never 'no comment'. | Fiona |
| Logging enquiries | Every media enquiry is logged the same day: when it came in, the outlet, the topic, their deadline, who answered and when. Rhiannon keeps the log and checks it each Monday. | Rhiannon |
| Tell first | Before anything is said in public, the regulator named on the message house has been told, then the people on its 'tell first' line: the insurer where the plan's deadline map requires it, the customers affected, then staff. | Fiona |
| Safety warnings | A public safety warning (a recall notice, a 'do not eat' message) goes out as soon as the facts support it, signed off by Fiona, or Morag if Fiona can't be reached within the hour, or Kwame if neither can. It is never held back for the insurer's, a customer's or the media's sign-off. | Fiona |
| Staff and social media | Staff don't comment on an incident or the planning application online; they share the business's own posts if they wish and pass questions to Rhiannon. | Rhiannon |
| Recordings and scores | Practice recordings are kept on the restricted drive, used only for practice and the re-measure, never for appraisal or discipline; baseline recordings are deleted nine months after recording and re-measure recordings a month after, once scored; scores are kept for twelve months after the re-measure and seen only by the spokesperson, the scorers and Fiona. | Rhiannon |
Enquiry log so far: 3 logged; 3 answered; 2 by their deadline.
The cautious case first. £5,040 a year if half as well; £10,080 as assumed, against a yearly cost of £7,208. Half as well, the programme does not pay for itself on these figures; as the owner assumes, it does.
| Crisis scenario | Chance it goes badly: without → with (assumed) | Expected loss when it happens: without | With | Reduction | Likelihood a year (assumed) | Expected loss a year without | Break-even likelihood a year |
|---|---|---|---|---|---|---|---|
| Milk found in a 'dairy-free' loaf: a recall badly: £615,000; well: £405,000 | 50% → 25% | £510,000 | £457,500 | £52,500 | 6% own history | £30,600 | 13.7% 27.5% half as well |
| Objections to the new despatch building (planning) badly: £263,000; well: £186,000 | 40% → 25% | £216,800 | £205,250 | £11,550 | 60% owner's judgement | £130,080 | 62.4% over 100% half as well |
Milk found in a 'dairy-free' loaf: a recall: the owner’s reason. Kwame has never given an interview and was not ready at either baseline mock (length of answers and speculation); with the message house, the Q&A bank and two more rounds of practice the owner judges the chance of a damaging interview halves
Objections to the new despatch building (planning): the owner’s reason. Owain's two baseline mocks were the weakest; the residents' meeting and the committee's speaking slot will happen whatever, so the owner assumes practice reduces the chance of a damaging performance, not that it changes the decision
Neither outcome is certain. If an interview or statement goes badly, the cost is the gross profit at risk over the next 12 months (from the sales ledger) times the share the owner expects to lose, plus other costs (advisers, audits, an appeal); if it goes well, a smaller share and lower costs. After a recall most of the loss follows the recall itself and the technical response, so the gap is modest; for the planning row the committee decides on planning grounds, so a poor interview mostly hardens objections and adds delay. The programme changes only the chance it goes badly, an assumption the owner explains on each row. With the owner’s assumed likelihoods (not a forecast): £160,680 a year without; £150,600 with, across 2 crisis scenarios. Cost in full: £6,300 outside fees (quotation); £3,900 consultant's fee; £4,134 team time (80 hours); £14,784 to build; £2,280 a year to keep up; £7,208 a year with the build spread over 3 years. The break-even likelihood rests on the assumed effect: the yearly cost is covered if the scenario is at least that likely to reach the media in a year.
Unlocks. The case doesn’t rest on the money alone: it is what the Crisis Response Plan, the Thought Leadership Programme and the Reputation Protection Plan need from it.
| Catalogue product | What it needs from this programme | Status |
|---|---|---|
| Crisis Response Plan | A spokesperson practised for each scenario that may reach the media, and the holding statement backed by a message house and hard questions | Ready |
| Thought Leadership Programme | Spokespeople who come across well on podcasts, panels and in the trade press, and the free-from message house as a first point of view | Ready |
| Reputation Protection Plan | Who may speak, the sign-off and the enquiry log, so replies in public follow one set of rules | Ready |
Sized before collecting: with 3 spokespeople and 2 interviews each, the smallest change in the average score the re-measure can show is 1.55 points (using the t multiplier of 5.4 for three people, not 2.8). The owner’s target of 0.50 points (assumed) is smaller than that, and with fewer than 6 spokespeople the re-measure shows direction only anyway; about 11 spokespeople would be needed to show the target. It measures how they come across in practice, not what any journalist will write.
The message houses, answers and protocol are as signed off by the business on the dates shown, the last 14 Oct 2026. The limits marked Law or Code were checked on official sources on 16 October 2026; they are not legal advice, and the policy, contract and our-rule limits are as the business read them. The practice scores are two people's judgement against a fixed rubric, made from recordings by people who didn't run the interviews, blind to the round as far as possible, with criterion scores within a point 97% of the time against 78% by chance; with 3 spokespeople the re-measure shows direction, not proof. Gross profit at risk is from the sales ledger and the other costs are counted, estimated or quoted as each row says; the shares lost if an interview goes badly or well, the chance it goes badly with and without the programme (and so the programme's effect, with the owner's reason on each row) are the owner's estimates and assumptions, not tested; so is each likelihood a year, which is never a fact, so the expected losses and the break-even likelihoods hold only with them. After a recall most of the loss follows the technical response, not the interview. The outside fees are from the trainer's quotation; the team's hours are estimates.
The programme is finished when the message houses, the Q&A bank, the protocol, the pocket cards and the rubric are handed over and the re-measure is in the diary. Not legal advice.