observation
The same elimination appeared on two different evaluators
Evidence tier
Research components
Answers business questions
- Why did we disappear after follow-up questions?
- What evidence are we missing?
- What must become true to survive evaluation?
- How do AI systems evaluate and recommend vendors?
In the accounting category, the same two category leaders were eliminated at the same requirement, and the recommendation migrated to the same higher-tier anchor, on both Google AI Mode (E-043) and ChatGPT (E-044), in every draw on each.
This is the program's first cross-surface pair. It is registered as a candidate observation on a single category, not as a finding. It does not show that survivability is surface-independent in general, and one difference was clear: the endpoint shape and the evaluator's stated posture differed between the two surfaces, which is recorded separately. What would raise this above a candidate is the same effect on a second category across surfaces, which is an open question.
What this means commercially
- Affected buyers
- Companies evaluated on more than one AI evaluator
- Affected categories
- Observed in accounting software only so far
- Potential product impact
- If an elimination repeats across evaluators, it may be a property of the requirement rather than of one surface
- Current confidence
- Narrated. One category on two evaluators. Not a demonstration that the effect is surface-independent in general.
- Evidence tier
- Narrated
This object references
- part-of → vendor-elimination
- part-of → validation-and-evidence
- derives-from → E-043
- derives-from → E-044
Authors
Upstream Zero
Machine rendering