0 · The ladder
No blended score — a five-rung ladder per engine, made of observed states: Invisible (not in the answer, not in the sources) → Read (your site in the engine's sources, you unnamed) → Named (in the answer) → Recommended (a repeating pick) → Nameworthy (recommended and holding it, month over month). Progress is rungs climbed. The top rung can't be bought or faked — it only exists under repeated measurement. And yes: it's the state we named the company after. Become nameworthy isn't a slogan here; it's the top of a ladder you can watch yourself climb.
| Engine | Rung this month | Movement |
|---|---|---|
| ChatGPT | Nameworthy — the custody #1, six weeks and counting | ▲ up from Recommended |
| Google AI Mode | Recommended | ▬ holding |
| Gemini | Recommended | ▬ holding |
| Google AI Overviews | Named | ▼ decay event — re-hardened, retake pending |
| Claude | Named | ▬ holding |
| Perplexity | Read — retrieved often, quoted rarely | ▬ two pages rebuilt answer-first this month |
1 · Where you stand, per engine
Twenty-five client-approved frozen questions across six engines. The 10 highest-intent run three times each on Gemini, Perplexity, AI Overviews and AI Mode and twice on the ChatGPT and Claude apps, where every session is driven by hand; the 15 supporting questions run once on the four automated engines. Reported separately, because the engines genuinely disagree — this month Firm A is a top pick on one surface and absent on another, which is normal, and exactly what a blended score would hide.
| Engine | Won (#1) | Named | Absent | vs last month | What's driving it |
|---|---|---|---|---|---|
| ChatGPT | 5 | 9 | 11 | ▲ +2 named | Directory profile depth; the custody page is now quoted directly |
| Google AI Mode | 3 | 9 | 13 | ▲ +1 won | Own-site content mapped to question intent; strongest surface |
| Gemini | 2 | 13 | 10 | ▬ flat | Entity/GBP presence; credential corroboration still thin off-site |
| Google AI Overviews | 1 | 3 | 21 | ▼ −1 named | One decay event (see §3) — re-hardened, retake pending |
| Claude | 0 | 2 | 23 | ▬ flat | Review volume below this market's card threshold; program running |
| Perplexity | 0 | 3 | 22 | ▬ flat | Retrieved often, quoted rarely — two pages rebuilt answer-first this month |
220 scored observations behind this table. The complete run log — every answer, every citation, every timestamp — ships with the scorecard.
2 · Citation durability: weeks held
Positions decay on their own — in our cross-panel measurement, an undefended #1 lasts roughly one to two days (first-pick hold rates by engine: ChatGPT 50%, Perplexity 52%, Gemini and AI Mode 37%, AI Overviews 16%, cross-day on identical questions). That baseline is the point of comparison for the only durability number that matters: how long your positions hold while defended.
| Position | Engine | Weeks held | Status |
|---|---|---|---|
| Custody question · top pick | ChatGPT | ≥6 weeks | Still held (censored — streak continues) |
| Fee/cost question · named | Google AI Mode | ≥5 weeks | Still held |
| Consultation question · named | Gemini | 4 weeks | Decayed week 4 → re-hardened → retaken (new streak, week 1) |
| Process question · cited | Google AI Overviews | 3 weeks | Decay event this month — see §3 |
The Hold view
The same four positions, drawn as the client sees them month over month — each cell is one weekly check from the run log. This is what “holding the line” looks like on paper: streaks, the decay events, and the re-hardening that turns a slip into a one-week dent instead of a lost quarter.
For scale: unmanaged, the expected lifetime of a #1 is one to two days, not weeks — the full durability baseline is published here.
3 · Decay report & re-hardening
One decay event this month: the AI Overviews citation on the process question rotated out when the engine re-weighted toward a competitor page with fresher statute references. Action taken: the client page's statute citations updated and a dated revision published; retake expected at next re-measurement. This is the monthly work — the engines re-roll constantly, and a slipped position that gets re-hardened within the month rarely becomes a lost quarter.
4 · Competitor movement
| Competitor | Movement this month | Read |
|---|---|---|
| Firm B | ▲ gained 2 named slots (Gemini) | New attorney-profile corroboration went live; watching whether it carries to ChatGPT |
| Firm C | ▼ lost its Perplexity citation | Their cited page now 404s — their loss is a retake opportunity on two questions |
| Firm D | ▬ unchanged | Still holds the generic "best" question on volume signals |
5 · Fix assets delivered this month
- Cost-and-fees page rebuild — answer-first, with the concrete figures engines can quote. Delivered paste-ready; published by your team on the 12th.
- Free-consultation page — the stated offer, on an indexed page, in liftable language. Published on the 19th.
- Review-program brief — this month's themes and ask-language for the front desk, aimed at the review-ranked surfaces.
What this sample is
A real client-format scorecard populated from one of our measurement panels, anonymized (Firm A–D), with the panel's repeated measures presented in the monthly format. Durability baselines are from our cross-panel archive. Nothing here is a projection or a promise — it's the shape of the evidence you get: per-engine, never blended, run log included, decay quantified. The measurement protocol is public.