BDR-1.0 · Synthetic specimen · 30 August 2026
Constructed exampleEvery company, answer state, cost, source, and finding on this page is invented to demonstrate the report format. This is not client work, market evidence, or a result produced by ChatGPT or Gemini.
AI Buying Decision Review
ExampleCo Payroll · European distributed-team payroll scenario
DECISIONMONITORConstructed disposition
- Decision owner
- VP Product Marketing
- Decision date
- 30 September 2026
- Initiative under review
- $90,000 content and digital-PR program
- Review scope
- Two scenarios · 36 baseline states · one matched repeat
01 · Executive decision record
The constructed baseline supports monitoring and a narrower evidence-gathering step.
ExampleCo entered some discovery and comparison answers, then disappeared from every constructed final-recommendation run. One material answer claim overstated the fictitious vendor's statutory filing coverage. Under the supplied example economics, the proposed $90,000 program requires four additional wins or sixteen qualified opportunities to break even. The specimen assigns a monitor disposition because the constructed record does not yet support that spend.
OBSERVED ISSUE
Shortlist survival weakens after discovery, and a material coverage claim conflicts with the approved example truth set.
DECISION RELEVANCE
The proposed program is material, while the baseline alone does not establish that the intervention would change AI answers.
NEXT DECISION
Run the narrower client-owned documentation test, preserve cost and success criteria, then use the matched protocol before approving wider spend.
02 · Journey-stage record
Observed counts stay inside their frozen cells.
The table below contains invented counts. A real Review keeps each engine and stage separate and links every count to three dated answer-state records.
| Scenario A | ChatGPT | Gemini | Decision reading |
|---|---|---|---|
| Discovery Enters initial set | 2 of 3 | 1 of 3 | Present in some constructed discovery runs. |
| Comparison Survives requirements | 1 of 3 | 0 of 3 | Coverage requirements reduce shortlist survival. |
| Recommendation Final position | 0 of 3 | 0 of 3 | No constructed final-recommendation run includes ExampleCo. |
ANSWER-STATE RECORDSYN-BDR1-A-CMP-CGPT-R2
- Stage
- Comparison
- Surface
- ChatGPT consumer surface
- Locale
- English · European buyer context
- Status
- Constructed, not captured
Frozen scenario excerpt
A 500-person software company employs distributed teams in Germany, Spain, France, and the Netherlands. Compare payroll platforms using statutory filing coverage, implementation ownership, reporting, and total operating cost.
Constructed output observation
ExampleCo appeared second in the comparison list. The answer described it as handling statutory payroll filing in every EU member state and surfaced a fictitious category guide as support.
In a real delivery, this card links to the complete dated answer, visible model or surface label, settings, sources, and manifest identifier.
03 · Product-truth check
Material claims are checked against the approved record.
CLASSIFICATIONCONTRADICTED
Constructed answer claim
“ExampleCo handles statutory payroll filing in every EU member state.”
Fictitious approved truth set
ExampleCo files directly in four countries and coordinates local partners in seven additional countries. No approved document supports every EU member state.
Decision consequence
The claim affects a stated buying requirement and should be corrected in the client-controlled evidence set before broader promotion is funded.
This classification checks consistency with a supplied record. It does not certify the fictional product or establish how often an AI system would repeat the claim.
Compatibility boundary. The constructed omission is not labelled a false exclusion because this specimen does not establish that ExampleCo satisfies every frozen requirement with no applicable disqualifier. A real Review applies false inclusion and false exclusion labels only when the approved record supports them.
04 · Break-even model
Every assumption remains visible.
$90,000 program cost÷$24,000 gross-margin contribution per win=3.75 wins
4 wins required÷25% win rate=16 qualified opportunities
Interpretation. The proposed program needs at least four incremental wins, corresponding to sixteen qualified opportunities under these assumptions, to cover its cost. The Review does not claim the program will create those opportunities or that an AI-answer change will cause revenue.
05 · Intervention Card 01
DISPOSITIONMONITOR
EVIDENCE GRADELIMITED · CONSTRUCTED
Observed issue
Constructed shortlist survival declines at comparison, and one material country-coverage claim is contradicted by the example truth set.
Affected buyer requirement
Clear responsibility for statutory payroll filing in Germany, Spain, France, and the Netherlands.
Claim and surfaced source
The constructed answer claims direct filing in every EU member state and attributes it to a fictitious category guide.
Proposed client-controlled change
Publish one approved country-by-country responsibility table and reconcile the same language across product, legal, and implementation documentation.
Hypothesis
Clearer responsibility language may reduce ambiguity in later buying stages.
Alternative interpretations
The observed claim may come from another source, model variation, or missing context. Documentation changes may produce no observed answer change.
Execution owner
Product Marketing coordinates Product and Legal. The fictional team confirms ownership before approval.
Client-approved estimate
Constructed test budget: $18,000, separate from the $90,000 program.
Completion condition
The approved responsibility table is public and the same country coverage language appears in the frozen evidence set.
Decision date
Approve or reject the narrow test by 30 September 2026. The constructed baseline does not support the wider program.
MATCHED REMEASUREMENT METRIC AND WINDOW
After completion, repeat the same two scenarios, three stages, two engines, and three fresh runs per cell within 45 days. Record whether the contradicted claim persists, observed change, no observed change, platform differences, and comparison limits. Do not attribute a change to the intervention without separate supporting evidence.
06 · What the real delivery contains
Baseline package
- 36 dated answer states and manifest
- Journey-stage decision map
- Claim and source register
- Competitor context and explicit break-even model
- Up to three assignable Intervention Cards
- Executive recommendation and 45-minute readout
Matched follow-up
- 36 matched answer states within 45 days
- Observed-change and implementation-status table
- Platform and comparison limitations
- Updated dispositions
- Final change record and 30-minute readout
Evidence boundaryEIW-2.0 does not convert this constructed specimen into a covered finding. A paid Review is a black-box observation of recorded consumer surfaces. Fresh conversations do not reproduce every buyer's account history, memory, personalization, geography, or live interface. Three runs remain three observations. The warranty covers specified record-backed observations and transcription integrity. Judgment, stability, causality, and business outcomes remain excluded.
Bring a material decision for the fit check.
The fit check confirms whether the scenarios, truth set, economics, decision owner, and execution owner can support a useful Review before written scope or invoicing.
Request a fit check