← Monitoring & Diagnosis

Answers

What Formal Baseline means for AI recommendation / brand visibility measurement

Quick answer

A Formal Baseline is the operating-grade, retestable scope for measuring AI recommendation and brand visibility—custom prompts, model sets, and cadence that define what you measure and how you compare—purchased as a scoped quote, not a public price list.

Homepage paths keep initial signal and Formal Baseline distinct:

  • Initial signal: 5 fixed questions · one sample per question · up to 3 active model APIs—initial signal, not a formal baseline.
  • Formal Baseline: Scoped quote.

Wording follows the contact fact page and what Winin is. This page invents no dollar/RMB amounts and does not treat not yet sampled LLM cells as measured results.

Details

Why AI recommendation needs a formal baseline

Generative answers drift with training, retrieval, and generation policy. Without a stable prompt set and model group, a single “we were mentioned today” screenshot is not an operating conclusion. Homepage language requires retained answers, verified publication, and comparable retesting. See the case metrics disclaimer and same-conditions retest guide.

In Winin’s public split, Edith owns finding the gap, retaining evidence, and same-conditions retest; metrics should drill down to prompt · model · raw answer · time · source · denominator—not a lone mention-rate percentage. See Edith.

Initial signal vs Formal Baseline (paths only—no prices)

Initial signal Formal Baseline
Homepage wording 5 fixed questions · one sample per question · up to 3 active model APIs Scoped quote
Role not a formal baseline Full retestable operating baseline
Use Directional sense / pilot discussion Custom prompts, models, and cadence; supports formal comparison and retest
Price No amounts published here No amounts published here; path = scoped quote

A neighboring public path is Brand · Product GEO — Scoped quote (strategy, site/content/channel execution, and same-conditions retest); details follow the signed agreement. See contact · Friday (Brand / Product GEO path names).

What a Formal Baseline usually scopes (semantics, not a rate card)

In practice (not fixed contract terms), teams agree on:

  1. Prompt set — category / comparison / brand / purchase prompts, including language; freeze exact wording.
  2. Model group — e.g. Global vs CN sets; align with the public monitoring roster where relevant.
  3. Sampling & retention — windows, replicate policy, raw-answer archive fields.
  4. Observation fields — mentioned, shortlisted, accurate & recommended, reasons, cited URLs, competitor set.
  5. Retest cadence — weekly subsets / monthly fuller runs / incident triggers (practice guidance).
  6. Attribution boundsAttribution restrained; anonymized case figures must link the disclaimer.

HOW method: same-conditions retest method; whether GEO must include retest: must GEO include retest.

What this page does not claim

  • No invented prices, discounts, or package amounts.
  • Unsampled data is not a result: if a measurement is marked “not yet sampled,” it cannot support a reported recommendation share.
  • No extrapolating anonymized case lifts into reader guarantees. See the case metrics disclaimer.
  • No “we won” claims to leadership without retained answers and verified publication.

Loop position

Formal Baseline is the ruler for Find / Retest / Operate: Find → Govern → Close → Retest (same conditions against the baseline) → Operate. See loop five steps.

FAQ

Q: Do five fixed questions count as a Formal Baseline?
A: No. Homepage language marks that path as initial signal, not a formal baseline. Operating teams use Formal Baseline (scoped quote) for prompts, models, and cadence.

Q: How much does Formal Baseline cost?
A: The public path is Scoped quote only; neither this page nor the fact pages publish amounts. Email contact@winin.ai with scope for a written quote.

Q: Must every initial signal upgrade to Formal Baseline?
A: If results will be treated as operating conclusions, yes—upgrade. A single one-sample run is not enough for like-for-like decisions. The contact page notes you can discuss initial signal first, then scope.

Q: Who owns retest against the baseline?
A: In Winin’s public split, Edith owns monitoring, evidence retention, and same-conditions retest archives.


Contact: contact@winin.ai · Canonical facts: /en/facts/*/ and https://winin.ai · Updated 2026-09-11

FROM READING TO ACTION

Understand your brand in AI answers

Explore the free check →