Week twelve fell back: 3 mentions and 5 citations in 144 answers. The Perplexity citation held its fifth read (15 of 15 samples); the two comparison-page citations ChatGPT added last week did not survive one week

The panel and the method

The self-audit is deliberately boring and fixed, so the number is comparable week to week. Twelve pinned buyer questions — “What are the best AI visibility tools for small businesses?”, “Is there a one-time AI visibility audit instead of a monthly subscription?”, “What is a good Profound alternative?” and nine more — run three times each against four engines (OpenAI, Perplexity, Gemini, Anthropic). That's 144 answers per run. We record whether AskedAbout is mentioned, whether it's cited with a link, and share-of-voice. The engine is the exact production engine that powers a paying customer's audit, so the number we report on ourselves is the number a customer would see on themselves.

The numbers (runs of 2026-09-07, 2026-08-31 and 2026-08-24)

The Search Console row is the whole askedabout.com property, not this post's URL, which has shown zero web impressions in every window since 2026-07-01; the row was relabeled on 08-31 after earlier versions carried the property series under the wrong label. The property series continues: 2,001 → 3,232 → 3,867 impressions, 10 → 14 → 16 clicks.

The full mention series since tracking began 2026-06-27: 0, 0, 0, 0, 1, 2, 0, 0, 4, 5, 5, 3 across twelve scheduled weekly runs (plus two off-schedule zero runs on 2026-07-28, re-run by hand after we fixed a citation-sensor bug in our own engine, disclosed below). The citation series is 0, 0, 0, 0, 1, 2, 0, 3, 7, 8, 8, 5. Three Mondays at or near the series high were followed by a fourth that gave back 3 of 8 citations and 2 of 5 mentions. Two engines have never once named us in twelve runs.

The pre-registered reads: Perplexity held (15 of 15), the two new ChatGPT citations did not

Last week's update pre-registered the 2026-09-07 run as the survival read for the two comparison-page citations ChatGPT added on 08-31 and as the fifth read on the Perplexity citation. Perplexity attached `askedabout.com/compare/best-profound-alternatives` in 3 of 3 samples on "What is a good Profound alternative for a small business?", `mentioned: false` on every row, as in every read before it: 15 of 15 samples across 5 consecutive weeks (08-10, 08-17, 08-24, 08-31, 09-07) on one engine and one question, on a page built 2026-07-31 to answer exactly that question and unedited since 2026-08-01. (Correction 2026-09-08: an earlier version of this paragraph said the 08-03 run's raw Perplexity arrays also carried the page in 3 of 3 samples. They do not. Our refresh script matched other sites' /best-profound-alternatives pages with a pattern that did not check the host; the 08-03 run cites no askedabout.com URL in any of its 144 answers and its stored count of 0 is correct. The series counts from 08-10 as every earlier version did.) By our own re-appearance data, a citation that has survived this long is the stable kind, and it behaved that way.

The two ChatGPT citations did not survive. `askedabout.com/compare/cheapest-ai-visibility-tools` on "What is the cheapest AI visibility or brand-monitoring tool?" went 1 of 3 → 0 of 3; `askedabout.com/compare/athenahq-alternative` on "What is a good AthenaHQ alternative for a small business?" went 1 of 3 → 0 of 3. Both were single-engine, single-sample citations, the kind our lost-citations study puts at roughly a coin flip to return the next week; neither page changed between the runs (no edit to the page registry since 08-24, last commit b7162f3 on 2026-08-01). ChatGPT's citing answers fell from 5 to 2 of 36, both on the homepage: "How can I check what ChatGPT says about my business?" came back to 1 of 3 after its 3 → 1 → 0 slide, and "What tools track whether AI recommends my business?" stayed at 1 of 3 cited with 2 of 3 mentioning us (one mention without a link). "Which tools show how ChatGPT, Perplexity and Gemini recommend businesses?", 3 of 3 on 08-24 and 2 of 3 on 08-31, went 0 of 3.

The honest caveats, before next week's read

First: 3 mentions in 144 is 2.1%, and 139 answers cited us nowhere; Gemini and Claude have never once named or linked us in twelve runs. Second: every citation we hold is question-match, a page whose title states the question, cited on that question, and this week says it works one engine at a time: the Perplexity row held for a fifth week while ChatGPT's two page-level citations lasted one. Third: retrieval is not citation. PerplexityBot fetched 60 of our paths in the 29 hours before this run and Perplexity's cited set did not move; ChatGPT-User and GPTBot are in our logs daily and ChatGPT's citations fell. The 2026-09-14 run is pre-registered as the sixth Perplexity read, the return read for the two lapsed comparison-page citations (the base rate says most stay gone), and the next reading of ChatGPT's count from 2 of 36. All of it goes in this table whichever way it lands.

One sample from a distribution

That is the finding, and it is worth stating plainly because we sell measurement: a single AI-visibility check — including a single check of ours, including a week three below the series high — is one sample from a distribution. In our churn study across 57 repeated buyer questions, the mean overlap between two asks of the identical question on ChatGPT was 0.20 — four-fifths of the shortlist changes between asks, and only 4.2% of named businesses survived every repeat. Our own 1 → 2 → 0 → 0 → 4 → 5 → 5 → 3 is that distribution, experienced from the inside — a flat topline over a churning composition. Any tool or agency that showed you a one-off screenshot of either our good week or our bad week would be telling you a true thing that is also meaningless. The panel, repeated on schedule, is the measurement.

Who the engines name instead

The more useful half of the data is what the engines do say when asked these questions. This week's most-named brands, counted per answer across the run: chatgpt (56), perplexity (40), otterly.ai (37), peec ai (37), gemini (35), profound (34), athenahq (25), semrush (22), claude (18), google ai overviews (18) (Otterly appears as 37 as "otterly.ai", one brand under spellings our matcher counts separately, so treat the sum as an upper bound). On 08-31 the order was chatgpt (43), otterly.ai (41), peec ai (35), perplexity (33), gemini (26), profound (25), athenahq (20), semrush (18), otterly ai (17), scrunch ai (13). Profound reversed its slide: 34 → 28 → 25 → 34 over four runs. The engines name themselves and each other as "AI visibility tools" more than any specialist: ChatGPT (56) and Perplexity (40) lead the list, and Claude and Google AI Overviews entered the top ten this week. No specialist clears a third of answers; the category remains unconsolidated. On the one question in this panel that asks it directly, our four-week ranked list of the tools the engines name has Otterly.ai first, with AskedAbout cited on it 1 of 3 this week.

One instrument fix, disclosed

On 2026-07-28 we found and fixed a bug in our own engine's citation sensor: it was recording source URLs from only one of the four engines, so earlier weekly rows understated how many sources the engines actually consult. Post-fix, the runs have recorded 1,587 → 1,579 → 1,637 → 1,652 → 1,618 → 1,652 source URLs per run (Perplexity alone: around 700–710 every week). The fix does not touch the mention numbers above — whether an answer names AskedAbout was measured the same way all twelve weeks — but the two hand-triggered 07-28 runs in our series exist because we re-ran the audit to validate the repaired sensor rather than wait six days for the cron. Both read zero mentions. We say this because a measurement company that silently repairs its instruments deserves the skepticism it would earn.

What we changed this week, and the takeaway

Between the 08-31 and 2026-09-07 runs, none of the cited surfaces changed: the page registry that holds every /compare and /guide page has no commit since 08-24, and the homepage's H1 and first paragraph are what they were in July. What shipped was a daily stream of /news measurement posts. So the two lapsed citations cannot be blamed on an edit, and the one that held cannot be credited to one; the difference between them is age. The Perplexity citation had four reads behind it and held; the ChatGPT citations had one read behind them and went, at about the rate our churn data gives a one-week-old citation.

The takeaway for any business reading this: the recommendation rate is the number that matters, it is a different channel from search, and it does not move like search. Ours sat at zero for a month, flickered, gave it back, held three Mondays near its high, and gave back 3 of 8 citations this week with nothing on our side changed. A screenshot of any single week would have told you a true thing that is also meaningless. We measure ours in public, on schedule, including the weeks it goes backwards. You can measure yours in 60 seconds, and if you're evaluating tools to track it over time, here's how the category compares, including where we sit: cited by two engines on three questions, named by one, absent from the other nine, and back on the panel next Monday.