An AI engine may cite a source that appears in its retrieval for a given question. In Citevio's engine panel measurement on 8 August 2026, citevio.com appeared in 5 of 16 buyer-intent agency-selection questions on ChatGPT and Perplexity and did not appear in 11. The measurement records outcomes, not a universal explanation for why a source was selected.
We publish the boundaries of our own measurement rather than presenting a formula. The question set was about choosing an AI-visibility agency, not about finding a dentist or choosing a dental practice.
What did we measure?
In Citevio's engine panel measurement on 8 August 2026, we ran 16 buyer-intent agency-selection questions on ChatGPT and Perplexity and checked whether citevio.com, our own domain, appeared. It appeared in 5 of 16 questions and did not appear in 11. In those 11 questions, citevio.com was cited 0 times.
| What we counted | Value |
|---|---|
| Buyer-intent agency-selection questions run | 16 |
| Questions where citevio.com appeared | 5 |
| Questions where citevio.com did not appear | 11 |
| Times citevio.com was cited in those 11 questions | 0 |
The zero is an outcome from our own domain, citevio.com. It is not evidence about what will happen in another question set or on another date.
This is one measurement window, not a trend. A domain appearing here does not mean it will appear reliably. Gemini was not measured because its run returned quota errors, so no Gemini outcome is claimed.
What kinds of sources can an AI engine cite?
We do not publish individual domains or agencies from this measurement because the underlying responses contain unverified third-party claims. We discuss only broad source categories, such as directory sites, review platforms, agency blogs and local news.
Before source selection can happen, crawler access is a separate, measurable gate. Citevio's national study read 6,497 readable US dental practice robots.txt files on 22 July 2026 and tested 13 AI and data crawlers.
For a separate numbered inventory of readable on-page information, see the dental website checklist for AI search.
Those categories do not explain why an engine selected a source. Citevio did not test ranking signals, backlink profiles, schema markup or other possible contributors to citation selection.
Do AI Overviews cite small practice sites?
Citevio has not published an AI Overviews citation measurement; the section below describes documented mechanics, not our own data.
Google documents that its AI features use the usual Google Search technical requirements and can show supporting links from eligible indexed pages. That is a documented mechanism, not a percentage from Citevio's data. For our methodology, see how we measure AI visibility.
What does this measurement tell you, and what does it not tell you?
Citevio's engine panel measurement on 8 August 2026 records outcomes from ChatGPT and Perplexity in one measurement window of 16 buyer-intent agency-selection questions. It does not tell us why a source was selected, and it does not generalize to patient-facing dental questions or to a specific dental practice.
This is a snapshot of which sources two engines cited on one day, in one question set. It is not a ranking factor list, and we have not tested what changes a citation.
What did Perplexity do differently in our measurements?
How Perplexity chooses which local business to name is not something we know, and Perplexity does not publish it. What can be reported is behaviour rather than mechanism: across Citevio's measurements Perplexity named local dental practices more often than ChatGPT did, and it changed its answer more often when the same question was asked again. Those two observations belong together, because the second one takes most of the shine off the first.
Three separate measurements point the same way, and each carries its own denominator rather than being pooled into one figure.
| What was measured | ChatGPT | Perplexity |
|---|---|---|
| Practices named 527-practice study, 6–26 June 2026 | 3 of the 415 it could read for (0.7%) | 18 of the 297 it could read for (6.1%) |
| Stability across two runs 16 buyer-intent questions, 64 answers, August 2026 | Stable — the same headline answer on both runs of the flagship question | Unstable — the headline answer changed between run one and run two |
| One fixed query, run twice per engine evening of 15 August 2026, 12 of 12 API calls completed | 0 of 2 runs named us | 2 of 2 runs named us |
The third row is one evening's photograph rather than a trend, and it was measured through the engines' APIs rather than their consumer apps. It is included because it is the smallest and most repeatable of the three, not because two runs settle anything.
What none of this supports is the conclusion a reader most wants to draw from it. Naming more practices is not the same as naming your practice, and an engine that changes its answer between two runs is an engine whose earlier answer told you less than it appeared to. A higher rate on one engine is a fact about the sample, not a route in. Whether one engine is genuinely easier to win than the other is a different question with its own page: do AI engines agree with each other about dental practices.
None of these three measurements tested a mechanism. No ranking signal was varied, nothing was changed and re-measured, and no engine documentation describes the selection Perplexity performed. What was recorded is what came back. How the sampling is set up, and what it deliberately cannot show, is on how we measure AI visibility.
Which questions are commonly asked about AI citations?
Do AI Overviews cite small practice sites?
We have not measured this. Citevio has not published an AI Overviews citation measurement, so we cannot say what share of AI Overviews citations point to small practice sites. What is documented is the general mechanism: Google has said AI Overviews draws on its search index and ranking systems rather than a separate citation list, so a page's ordinary search visibility affects whether it can be pulled into an AI Overview at all. That is a documented mechanic, not a number from our own data.
What kinds of sources got cited in the measurement?
We do not publish individual domains or agencies from this measurement because the underlying responses contain unverified third-party claims. We discuss only broad source categories, such as directory sites, review platforms, agency blogs and local news.
Does this measurement explain why those sources were cited?
No. It records outcomes from one measurement window for one question set. It does not test why an engine picked one source over another, and Citevio has not run a controlled test that would answer that question.
Was citevio.com cited in this measurement?
No. In Citevio's engine panel measurement on 8 August 2026, citevio.com was cited 0 times across the 11 buyer-intent agency-selection questions where it did not appear; the full measurement window contained 16 questions on ChatGPT and Perplexity.
How many of the 16 questions did citevio.com actually appear in?
5 of 16. In Citevio's engine panel measurement on 8 August 2026, citevio.com appeared in 5 of the 16 buyer-intent agency-selection questions run on ChatGPT and Perplexity, and did not appear in the other 11.
Did this citation measurement include Gemini?
No. Gemini's run returned quota errors during this measurement, so no Gemini outcome is claimed for the 16 buyer-intent agency-selection questions.
Does Perplexity behave differently from ChatGPT in Citevio's measurements?
Yes, in two ways that cut against each other. Across Citevio's measurements, Perplexity named local dental practices more often than ChatGPT (18 of 297 readable practices in the 527-practice study, against ChatGPT's 3 of 415), but it also changed its answer more often when the same question was asked again — across two runs of the same 16 buyer-intent questions, ChatGPT gave a stable headline answer both times while Perplexity's changed. A higher naming rate on one engine is a fact about that sample, not a reliable route in, because an engine that changes its answer between two runs told you less than its first answer appeared to.
What happened when you ran the same query twice through each engine's API on 15 August 2026?
ChatGPT named us in 0 of 2 runs and Perplexity named us in 2 of 2 runs, in a set of 12 of 12 completed API calls that evening. That's one evening's photograph rather than a trend — it's included because it's the smallest and most repeatable of the three Perplexity-behaviour measurements on this page, not because two runs settle anything.
Does this measurement tell you anything about patient-facing dental search questions?
No. The 16 questions in this measurement were about choosing an AI-visibility agency, not about finding a dentist or choosing a dental practice, so the outcomes don't generalize to patient-facing dental questions or to a specific dental practice.
Can I check the underlying data behind this measurement myself?
Yes. Citevio's dental market research, including this measurement, is published openly under a CC-BY-4.0 licence with a DOI, so the numbers can be checked and reused. Client data is never published.
Citevio's dental market research is published openly under CC-BY-4.0 with a DOI, so the numbers can be checked and reused; client data is never published.
Sources: Google Search Central, AI features and your website; OpenAI, ChatGPT search; Perplexity, What is Perplexity?; Citevio open dataset, Zenodo DOI 10.5281/zenodo.23143214; methodology: how we measure AI visibility.
Want to know if your practice gets cited?
Run a free scan and see what ChatGPT, Perplexity, Gemini and AI Overviews currently say about you.
Run my free AI scanDelivered within 48 hours · no sales call — ever