One consequential decision, examined end to end.
What you leave with: a board-ready decision report, the decision frozen into one sealed record, and a review booked for the date reality answers — then a learning report on why it went the way it did. Three fixed packages, priced by what changes the work, delivered by people with the LEXUN engine underneath. Choose the package by the decision, not by the deliverables list.
Three packages
Pick by the decision in front of you. Every package produces the same kind of frozen record; they differ in scope, scenarios, stakeholders and how long we stay with the outcome.
For one business unit, up to four stakeholders.
Typical decision a hire, an equipment purchase, a lease.
For up to eight stakeholders, with a half-day workshop.
Typical decision a second location, a debt facility, a restructuring.
For the board, with a full-day workshop and a board presentation.
Typical decision acquiring a competitor, a major contract, a market exit.
Who does what: the engine and the people
The engine computes
- every probability band, on a released model, at a stated seed
- the scenarios, stress tests and sensitivity, one input at a time
- breakpoints: the value at which the answer flips
- the receipts, so your own team can reproduce every number
A language model never produces one of these numbers.
People do
- the framing workshop and the success test
- the evidence inventory, and classifying each input as fact or assumption with you
- the reference-class review and the independent challenge
- the words of the report, the freeze, and the outcome review on the date
Judgement is labelled as judgement wherever it appears in the report.
What every package delivers: the fourteen deliverables
- Decision framing workshop — one question, a checkable outcome, the alternatives named
- Evidence inventory — every document, its source, owner, date, applicable period and expiry
- Facts-versus-assumptions classification — each input labelled and priced by the engine
- Reference-class review — what happened to comparable decisions, with provenance and its limits
- Scenario modelling — do nothing, expected, the proposer’s case, downside, severe downside, upside, delay
- Stress testing — the shifts that take the decision to its floor
- Independent challenge — the strongest case against, written by someone who did not propose it
- Sensitivity analysis — the input that moves the answer most, in points
- Decision thresholds — breakpoints, margins and the value at which the answer flips
- Risk and dependency register — early warnings, stop conditions, owners
- Frozen decision record — sealed under one hash, with its reproducibility receipt
- Executive decision report — the paper the board or the client files
- Scheduled outcome review — a date, an evidence owner and a resolution source
- Post-outcome learning report — why reality differed, which assumption mattered, what the organisation now does differently
Exactly what changes the price
| Variable | Moves the price when |
|---|---|
| Number of decisions | Linked decisions each need their own frozen record |
| Number of scenarios | Beyond the package allowance; each custom scenario is modelled and documented |
| Evidence volume | More than roughly 40 documents to inventory, or evidence that needs reconciling |
| Stakeholder count | Interviews and workshop seats beyond the package |
| Data preparation | Management accounts or exports that need cleaning before facts can be verified |
| Workshop requirements | On site rather than remote; more than one session |
| Custom modelling | A question the released models cannot answer honestly is declined or scoped as bespoke analysis, stated as such |
| Reporting requirements | Branded, multiple audiences, presentation to the board |
| Follow-up period | Checkpoints beyond the package; a resolution date beyond 12 months |
| Governance requirements | Non-disclosure terms, data-handling constraints, formal sign-off stages |
Boundaries, stated explicitly
- Not regulated advice. The Decision Audit does not provide financial, legal, tax or investment advice and is not a substitute for it. Where a decision needs a regulated adviser, the report says so and stops there.
- Not a guarantee. A probability band is a statement about a model on stated inputs; the report records the band, its basis and its known weaknesses, and never a promised outcome.
- Not a black box. Every number in the report carries its seed, model version and input hash, so your own team can reproduce it.
- Only released models. If the decision falls outside the three released models, the engine is not stretched to fit; the audit is scoped as structured analysis and labelled as such.
- Your data stays yours. Evidence is handled under the data-processing agreement, which sets out where it is held and for how long, and none of it is used to train anything. The organisation workspace itself is planned, not live: during an audit the work is delivered by people, as the organisations page sets out.
How it runs
| Stage | Focused | Strategic | Board |
|---|---|---|---|
| Framing workshop | 90 minutes, remote | Half day | Full day |
| Evidence and classification | 1 week | 2 weeks | 3 weeks |
| Modelling, stress, challenge, sensitivity | 1 week | 2 weeks | 3 weeks |
| Report and freeze | 3 working days | 1 week | 1 week, plus board presentation |
| Outcome review and learning report | At 3 months | At 6 months | Checkpoints, then at 12 months |
Prices are quoted exclusive of VAT; LEXUN is not currently VAT-registered. Payment by invoice: 50% on booking, 50% on delivery of the frozen record; the learning report is included, not billed again.
Request a Decision Audit
Describe the decision in a sentence and tell us which package looks right. A person replies within one working day with a scope, a fixed price within the package range, and dates.