Every promise and decision with its trail: who proposed it, who agreed, what happened after. A suggestion nobody replied to stays a suggestion here — that distinction is the difference between an answer and a fight.
At the 90-day review (22 May 2025) the reported headline results were: forecast accuracy up 6.2 points versus baseline, ambient inventory down 9%, order automation at 91% — all ambient, pilot-cohort numbers. Robert Kahn’s response was to ask where the fresh waste reduction was: “that is the case I approved”.
Sofia Almeida is the only person at Acme who understands the fresh categories — she said so herself, unprompted, at the March 2024 demo; the point sat in three risk registers “in slightly different words”, which Charlotte Meyer’s verdict on was: then it is in none.
Whether the warehouse goods-receipt data can ever be obtained is unresolved. The archive records who owns the system (Nils Ackermann, logistics IT), that Priya asked twice (2024 and 2025), that Kahn escalated in June 2026 (“Send me when the…”), and that no answer exists in the corpus.
About 60% of bakery is baked in store — a production decision made twice a day — and 40% is delivered, which behaves like the rest of fresh. The split varies heavily by format: big stores bake most of it, convenience almost none.
In June 2026 Nadia Haddad compared the extract totals with the till numbers for January–March at Sofia’s April request: they do not reconcile, the extract is lower, and it is “not a rounding difference”. She had held the result for two days, did not yet trust it, and declined to give a figure in the meeting. No cause was identified in the archive.
On the signature-clock reading the twelve-month expectation was not met: signature was July 2024, fresh went live in December 2025, and in June 2026 Marco Rossi answered Robert Kahn’s question with “On that reading, no.” An alternative reading (clock from fresh go-live, ending December 2026) exists in the conversation, but Marco declined to argue for it.
In September 2024 the pilot was reduced to 25 stores: two of the three regions could not support 40 during their peak, and Lena Fischer agreed the reduction with them. Ana Duarte reissued the plan on that basis.
At the ordering-logic design workshop on 12 November 2024, service levels of 98 and 95 were agreed for the pilot cohort — proposed by Tomas Lindholm as a reduction from 99, accepted by Lena Fischer (“Fine. Agreed.”), with the expected effect that inventory falls.
A joint engineering session (held 4 June 2025) produced a permanent-fix plan with a date: 15 September 2025, gated on Acme’s quarterly change window because the fix requires a middleware upgrade on Acme’s side.
On 4 September 2025 Nadia Haddad (RELEX) proposed an approximation for remaining shelf life: total shelf life from MHDHB minus average days-in-warehouse per article group, derived from stock turn.
The fresh cutover runbook carried on-call mobile numbers with distribution restricted to a named group; Lena Fischer forwarded the operational sections without the contact page, but Nadia Haddad later forwarded the file she had — with the contact page — to Meridian, unaware two versions existed. Priya Nair: “My mobile is now with a consultancy I have no contract with. Please ask them to delete it and tell me when they have.”
Meridian received extracts rather than read access to the RELEX test environment: there is no category-level restriction in the current configuration (“read access is read access”), a separate environment would take roughly two weeks, and Priya Nair ruled it out — “I cannot explain to my data protection officer why a third party had read access to everything.”
The DC-2 cover gap: Kwame Boateng was the only person watching the feed and was on holiday 10–24 March 2025; the monitoring fired at him in Portugal. Ana Duarte acknowledged she raised the cover question in January and did not close it.
In November 2025 Sofia Almeida asked in writing for the removal of the note recorded against her name in handover document Acme-fresh-handover-v3 (page 4, key stakeholders): “out October to November, surgery”. She had said it in the March 2024 demo because it was relevant to the timeline, not so it could follow her for two years; she asked to keep the fact that she is the only person at Acme who knows the fresh categories, “because that is a real project risk and it is the reason I said anything at all”.
The Acme DPO (Katarina Voss) recorded four actions after the April 2026 review: written confirmation of the hosting region and its naming in the agreement; details of the fifth sub-processor; confirmation that the November erasure request is fully closed with a list of every location; and a stated retention period for meeting recordings and transcripts. All four were wanted within four weeks.
Fresh was committed as Phase 2 in writing at the March 2024 demo: Marco Rossi committed to put it in the statement of work with a target window rather than leaving it open, after Lena Fischer insisted it not become “a someday item”.
Robert Kahn set the investment frame in March 2024: return on investment inside twelve months of signature, with the case built on fresh waste because that is what the board asks him about following a newspaper story.
DC-2 went into the technical assessment as a named risk with monitoring to be designed around it from day one — Tomas Lindholm’s stated approach: build assuming it fails rather than discover it.
Case-pack remediation finished on 19 December 2024 — all 1,317 values. It took Priya’s team three weeks of overtime; for two unresponsive suppliers the case pack was inferred from six months of actual delivered quantities, which she recorded as “not ideal but it is right for 190 of the 200”.
The operator identifier was first raised in the September 2024 master-data workshop: Kwame Boateng reported an unexpected operator column on write-off movement records and said he did not need to receive it; Priya Nair agreed to take it out (“a five minute change”), and Ana Duarte noted it as an action.
A file-size check was added the week after the 14 March failure: any file under 1KB now fails the job loudly. Since then the archive records no silent failures — four loud ones by December 2025.
First full fresh quarter (January–March 2026, produce and dairy, 25 pilot stores): fresh waste 4.1% → 3.4% of fresh net sales, ~118 tonnes avoided, ~296 tonnes CO2e on Acme’s own factors, availability +1.9 points, order automation 88% in scope. Chilled ready meals is not in these numbers (went live too late in the quarter).
Fresh Phase 2 slipped six weeks for two reasons: UAT resourcing on Acme’s side, and a calculation defect in the fresh allocation module that produced negative quantities in test. The defect was understood, fixed in days, and verified by November.
At the 17 November 2025 go/no-go, Lena Fischer decided: go live on produce, dairy and chilled ready meals; bakery comes later as its own workstream. Fresh go-live was set for 15 December 2025.
For bakery, Meridian consumes a RELEX demand forecast at store, article and day and owns the production plan; RELEX owns the demand signal. Nadia Haddad was to define the interface points by mid-February 2026.
Tomas Lindholm withdrew his own cutover recommendation in March 2025: he had said a file-size check would catch this class of failure, and it would not have — both files were the size he would have expected. “A file can be the right size and still be the wrong file.” The withdrawn recommendation still sat in the cutover pack, runbook and hypercare plan, because it was written once and copied; nobody went back to take it out of three documents.
Marco Rossi proposed on 11 June 2025 delivering fresh in three waves (produce and dairy, then chilled, then bakery). Acme never replied; the plan was nonetheless built on the three-wave basis, justified internally as “discussed verbally in the June steering and nobody pushed back”.