Fictional advisory example · prepared draft
See the client output and the work behind it
You asked for an example of how interview evidence becomes a client recommendation. Below is a completed fictional case: a decision question, three findings with their sources, a conditional recommendation and the decisions a reviewer keeps. Read it as work to inspect, not advice to adopt.
What you are looking at
Larch, the participants and the eight interview notes are invented. The original draft comes from a dated AI drafting run; the reviewer return shown later was scripted for teaching. Nothing was installed in a real firm or professionally approved.
- Start with the recommendation, then trace each finding to its source IDs
- Notice where a single account is prevented from becoming a general claim
- Watch for the decisions that stay with a person at every step
How to use it
Compare it with one service you already sell. The optional reading sheet gives five questions you can answer privately; no call or files are needed.
See the client output and the work behind it
Prepared teaching draft · CV07 excerpt, 5 October 2026. Larch, the participants and the case are fictional. This extract illustrates an advisory evidence-to-decision method. The original report derives from a dated model drafting run; the reviewer return and revision are separately scripted teaching material. It is not an installed system, professional approval or a client outcome.
The decision the client pays to make
Should a components business test a local replacement-part offer for non-critical processing components, and what buyer conditions should a pilot require?
The teaching case uses eight invented interview notes. It compares no launch, a broad speed-led launch and a bounded compatibility-led pilot. It does not establish market size or population-level demand.
The completed report excerpt
Finding 1 — Trial depends on approval and the current alternative. A maintenance manager would consider a trial only with engineering compatibility sign-off. A procurement lead reports that the incumbent delivered nine of ten orders within the agreed window. These accounts support examining approval conditions and incumbent performance; they do not establish a general unmet need. Sources: N01 and N02.
Finding 2 — Willingness differs by use case. A workshop lead would trial a documented substitute for low-risk fittings while retaining approved suppliers for safety-critical parts. Another operations manager reports no recent stockout. Trial interest is conditional, and urgency cannot be assumed across the market. Sources: N03 and N07.
Finding 3 — One premium purchase does not establish general willingness to pay. One repair buyer reports a single urgent purchase at a higher price. A distributor has no record linking faster delivery to switching, and another buyer has committed neither an order nor a premium. Sources: N06, N04 and N08.
Conditional recommendation: Consider a pilot of one non-critical component with willing sites, engineering compatibility review and a separate supplier-capability check. Do not justify it through a general switching or price-premium claim. No launch remains reasonable if qualification effort or capability makes a trial unattractive.
The notes are not transcripts or verified purchase records. Supplier capability, costs and a representative market sample remain absent. The recommendation is a teaching draft for review, not professionally accepted advice.
How the work reaches a reviewer
| Step | Prepared work | Decision a person retains |
|---|---|---|
| Define the job | One decision question, method, included evidence and exclusions | Approve scope and the method |
| Organise evidence | Preserve each note, source ID and use condition | Decide whether sources are permissible and adequate |
| Prepare analysis | Compare approval, urgency, alternatives and pricing evidence | Challenge interpretation and missing evidence |
| Prepare the output | Findings, options, conditional recommendation and limits | Review the exact version before professional issue |
| Handle a return | Keep the reason and revise the affected claims | Decide whether the revision addresses the return |
In the separately scripted review example, a reviewer rejects an introduced general-switching claim and narrows the price interpretation. That return was authored for teaching; it is not an actual reviewer accepting this output, nor evidence that the original model draft made that error.
What this helps you examine
If your firm repeatedly turns qualitative evidence into client decisions, this is a specimen of the work an AI implementation could help prepare and check. It lets you inspect the output, trace its basis and consider the judgement you would retain before deciding whether to investigate your own service.
The method structure and checking questions are possible reusable components. Each new case still needs its own sources, use rights, interpretation and review. This example does not include a completed second engagement or measure reduced effort.
The accompanying reply supplies the single buyer question.
Source edition: Original CV07 teaching packet, Parts B and C; source IDs identify the fictional records. The separate interactive browser demonstration uses authored deterministic rules and current-session states. It does not run live AI or demonstrate client integration. The source edition’s scenario fees and finance records are omitted; they are not Build n Bloom pricing.
This is a generic excerpt. No version of your firm’s actual method, operating environment or client engagement has been designed here.
The next useful step
Use the example privately, or tell us which part relates to a service you already sell. A call is optional.