One register for evidence notes, instruments, field briefings, corridor playbooks and decision conversations. Every item shows its shelf and its status, and the publication classes below set out the evidence a piece must carry before it may claim anything.
18 items are open right now: 3 evidence notes, 8 instruments you can print and run against a live decision, 4 methods of record and 2 platforms. A further 3 field briefings are not open downloads: we send them on request, within one working day.
Version 1.1.0 · Last reviewed · .
The evidence note read most often, and the instrument that goes with it. Read the argument, then print the waterfall and run it against one customer, one product, one channel.
Pairs as
Notes, instruments, briefings, corridor playbooks and the method of record, in a single list with its class, its status and what you can do with it. Filter by what you need rather than by where it happens to live.
A shelf is where an item sits in the register: what kind of thing it is. A publication class is the evidence standard a piece is published to, and the classes are set out in full below. Every item shows its shelf; class assignment is recorded editorially and is not published row by row.
33 of 33 publications shown
The Discipline · No. 01
The one rule that reorders every other rule in the firm.
AI, honestly · No. 01
The frame for the whole framework: no vendor pitches, no unattributed claims.
MarginLeak · No. 01
How true profitability by customer, product, channel and payment behaviour hides in the data you already own.
MLX-01 · A4, printable
Follow one euro from list price to cash-adjusted contribution, and mark every silent leak.
DRT-01 · A4, printable
Five versions of the target, and the gap between the one sold and the one you inherit.
DRT-02 · A4, printable
The buyer-side model on one page, for a board that has ten minutes.
TS-01 · A4, printable
Seven rungs from anecdote to independently verified value. Locate where a claim actually sits.
EDC-01 · A4, printable
Six gates from a claim to committed action. The method on a single sheet.
TCT-01 · A4, printable
Growth, margin and cash on the same page. Approve only if all three hold.
BB-01 · A4, printable
Seven handovers between a business case and the P&L, and where the benefits leak.
BAC-01 · A4, printable
One page a board needs before it commits, with the authorise gate made explicit.
English 21 pages, French 26 pages · CFO, Group Investment Committee
The one-page defence a board should ask for before approving a capex line.
English 15 pages, French 16 pages · Commercial Director, CFO
How to test a price move against the four axes that decide whether it survives a quarter.
English 26 pages, French 29 pages · Group CTO, CFO
The evidence trail that turns an AI pilot into a defensible budget line.
Board chair, company secretary
One page per decision, six decisions a year. For boards that want evidence, not comfort.
How we know what we claim
The evidence ladder, the verdict standard and the closed-loop review, written into every engagement letter.
Redacted specimens
Redacted specimens of the artefacts we produce: verdicts, exhibits and engagement briefs.
What we added and retired
What we added, retired and refined. Instruments are dated, and retirement is recorded.
Nine gates, three domains, one memo
Nine gates across value leakage, AI readiness and decision quality. Each domain read on its own, ending in a printable memo. About two minutes.
WAEMU to euro · Abidjan, Dakar, Paris · English 15 pages, French 17 pages
Engine · Profit and pricing
Francophone West Africa · 13-Week Cash Decision Twin, then Benefits Bridge
Engine · Cash and capital
Abidjan, Lomé, Douala · Hidden Logistics Cost Observatory, then Margin Leakage Waterfall
Engine · Operations and supply
Corridor wide · Capex Evidence Gate, then Evidence-to-Decision Cycle
Engine · Leadership, governance and risk
WAEMU and CEMAC · Growth Constraint Decision Twin, then Three-Corner Test
Engine · Growth and commercial
Paris, Geneva to Abidjan, Lomé · Board Authorisation Canvas, then Truth Stack
Engine · Leadership, governance and risk
Abidjan, Dakar, Casablanca · AI Workflow Experiment Studio, then Benefits Bridge
Engine · AI and productivity
Corridor wide, francophone Africa · Growth Constraint Decision Twin, then Benefits Bridge
Engine · Leadership, governance and risk
Abidjan, Douala, Casablanca · Hidden Logistics Cost Observatory, then Capex Evidence Gate
Engine · Operations and supply
Platform · live
The index every AI programme is scored against before it earns a budget line.
Platform · standing reference
The instrument set we run inside Evidence Express and the Decision Sprint, listed with the engagement that uses it.
Instrument · in preparation
The gate a capital request must clear before a board authorises it. Named in three corridor playbooks; the specimen is in preparation.
Instrument · in preparation
Where a plan loses time and margin between decision and delivery. Named in the corridor playbooks; the specimen is in preparation.
Instrument · in preparation
The structural view behind the Margin Leakage Waterfall: where leakage is designed into the operating model. The specimen is in preparation.
Thirteen further evidence notes are commissioned and listed with their publication month on the Insights ledger. Fifteen decision conversations are indexed below.
A publication class is not a content format. It is a promise about what the piece is allowed to claim, and the evidence it must carry before it may claim it. A conversation may report what an executive believed. A benchmark may not, unless a governed dataset stands behind it.
Rows are ordered by the minimum method grade a class demands, from no minimum up to F5. The floor is an entry price, not a ranking of authority: a Final Evidence Standard at F4 governs the practice, while a Benchmark or Atlas at F5 needs a governed comparable dataset before it may exist at all.
| Class | Method | Finding | Mandatory condition |
|---|---|---|---|
| CEO Decision Conversation | No minimum | SEG-0 to SEG-1 | Separate testimony from verified fact |
| Executive Evidence Brief | F2 | SEG-1 to SEG-3 | Evidence ledger and limitations |
| Method Note | F2 | SEG-1 to SEG-2 | Disclose whether logic is tested |
| Decision Dossier | F2 | SEG-1 to SEG-4 | Triangulation and claim-permission table |
| Field Experiment Note | F3 | SEG-3 to SEG-4 | Predefined hypothesis and outcome measures |
| Reference guide | F3 | SEG-2 to SEG-5 | Grade material findings individually |
| Final Evidence Standard | F4 | SEG-4 to SEG-6 | F3 versions must be labelled draft standards |
| Benchmark or Atlas | F5 | SEG-5 | Governed comparable dataset |
Method grades are FEM (Framework Evidence Maturity). Finding grades are SEG (Specific Evidence Grade). The two are never merged, for the reason set out below.
These are not open downloads, and that is not a capture tactic: they are the evidence checklists and decision templates we use on engagements. A briefing applied to the wrong situation does damage, so a partner checks the fit first.
Why on request
A partner reads your context before anything is sent.
You receive the briefing within one working day, or the one that fits better.
No sales sequence is triggered behind the request.
Extent — 12 to 30 editorial pages per edition. Page counts are measured from the rendered edition, never declared in advance. English and French editions differ; both are stated.
The one-page defence a board should ask for before approving a capex line — with the gates we run against every project brief.
Practitioner-grade briefings. We send by email within one working day.
How to test a price move against the axes that decide whether it survives a full quarter — before it reaches the sales floor.
Practitioner-grade briefings. We send by email within one working day.
The evidence trail that turns an AI pilot into a defensible budget line, and the handovers where value most often leaks.
Practitioner-grade briefings. We send by email within one working day.
One page per decision. The template we ship to boards that want evidence, not comfort.
Treated under client confidentiality: never reused, benchmarked or aggregated.
A named framework is not validated because it has been published. Internal use is not field validation. Three client examples are not a benchmark. So we grade each material finding on its own, on one published scale, and a document's grade is the strongest finding it rests on — shown with the range beneath it.
How strong is this particular finding, on a scale from SEG-0 to SEG-6, applied to a single conclusion rather than to a brand.
A well-made method does not strengthen a weak finding. A strong finding does not validate the method used to interpret it. The grade belongs to a version, a decision type and a scope, never permanently to a name.
A grade without its scope and its limitations is decoration, so we never show one alone. No asset may be self-certified. A grade is published with its version, validated scope, evidence period and limitations, and a document-level grade is published with the range of grades beneath it, or it is not published at all.
Where a grade has been earned, it appears on the asset's own page as the panel opposite, complete with the version and scope it applies to, the evidence behind it, and what has not been established.
Every graded asset displays this.
Illustrative field values, not a live grade.
A commissioned set of CEO decision conversations from francophone Africa and its corridors. Each one opens on the trade-off rather than the outcome, because the trade-off is the part that transfers. The index is published so you can tell us which one you would read first.
Distribution
Cash and currency
Capital
Infrastructure
Regional scale
Working capital
Localisation
Informality
AI and data
Governance
Talent
Strategy
M&A
Board decisions
Legitimacy
Titles are final. Publication dates are not, and we will not invent them. Nominate a decision if yours belongs in the set.
A playbook sequences existing instruments against one recurring decision in one named corridor, and states the local conditions that change the answer.
A playbook publishes when the sequence has been run often enough to prescribe, with its grade and its limitations on the page.
Request a playbook →Every piece opens with a claim it could be wrong about. If the claim is safe, the piece does not ship. That one test removes most of what a firm is tempted to publish.
Sources, samples and workings live in the piece, not in an appendix. If it cannot be re-run, it has not been written.
Every piece closes on a rule you can apply to your own decision on Monday morning, not an invitation to think differently.
We document consequential decisions from the two arenas we work in. Nomination is not promotion: a decision enters the register only if it can be evidenced, and the executive who owned it is willing to be held to the record.
What qualifies
A decision already taken, with a consequence that has landed.
A named owner prepared to separate what they knew from what they assumed.
Documents, not recollection, for anything presented as fact.
Willingness to publish the decision that did not work as well as the one that did.
Anonymised on request. Nothing is published without written approval of the final text.
Subscribe to receive our evidence notes and decision briefings.
