Reference · 2026-10-09 · status page · no measurement has run yet
AIEO Measurement
Three instruments can show whether AI answers cite this site. What each one counts, what it leaves out, and where each one stands.
The answer, in one paragraph. Three instruments can show whether AI answers cite machinebehavior.io: the Generative AI performance report in Google Search Console, the AI Performance report in Bing Webmaster Tools, and a probe that sends a fixed list of questions to Perplexity's API and logs the sources in each answer. On 2026-10-09 none of the three feeds a number to this site. Setting them up is an open decision for the site owner. This page records what each instrument counts and leaves out. It carries no baseline, because no run exists.
Status on 2026-10-09
| Instrument | What it counts | What it leaves out | Status here |
|---|---|---|---|
| Google Search Console, Generative AI performance report | Impressions: how many times links to your site were shown to a user in a generative AI feature on Google Search, in AI Overviews and AI Mode, grouped by page, country, date and device. Open to all sites since 2026-08-31. | The answer text, and the query: no query dimension is available in the report. Data is assigned to the canonical URL, so the three pieces whose canonical is their Substack copy count under substack.com, outside this property. | A Search Console verification record for machinebehavior.io is in DNS (checked 2026-10-09). No reading from the report is recorded or published here. |
| Bing Webmaster Tools, AI Performance | Citations in answers from Microsoft Copilot, AI-generated summaries in Bing and select partner integrations: total citations, cited pages, average cited pages per day, and grounding queries. | In Microsoft's words, the data does not show individual AI answers, exact prompts, or why a specific page was referenced. Grounding queries are grouped phrases: They are not full user questions or prompts.ChatGPT, Perplexity and Claude are outside it. | No reading from the report is recorded or published here. |
| Perplexity API probe | For each question in a fixed list, the answer and its sources: the Agent API returns a message item for the answer text and a search_results item for the sources. | Every other engine, and answers in Perplexity's own app, since the probe calls the API. | Not built. |
Search Console, in more detail
- Google launched the report on 2026-06-03 for a subset of sites. A note on the announcement reads:
As of August 31, 2026, we've rolled out these insights to all websites worldwide.
- The same account holds the Search generative AI control, which sets whether a site's links and content can appear in AI Overviews, AI Mode and generative AI features in Discover. Include is the default for every property. Whether this property's setting differs from the default is visible only inside Search Console and is not checked here.
- Google on outside tools:
No third-party tool has access to our internal ranking or AI systems.
Bing, in more detail
- Microsoft launched AI Performance as a public preview on 2026-02-10.
- Microsoft states that the report counts citations only:
AI Performance does not measure rankings, authority, performance, or importance. It simply shows which content was cited.
- Microsoft adds a warning about the trend view:
Changes in citation trends are observational and do not indicate the impact of any single update, model change, or content modification.
The probe, if it is built
The design below is a plan. Nothing in it has run.
- A fixed question set. Ten to twenty questions in three groups: definitions of terms this site defines, questions about the site and its author, and problems the site's pages answer. The list is dated and hashed before the first run, the same way the experiments hash their predictions, so the questions stay fixed while the answers change.
- A fixed schedule. The same questions once a week, three samples per question, because one answer is one sample.
- One row per answer. Date, model identifier, question id, the cited URLs in order, whether a machinebehavior.io page is cited and which one, whether the page's own definition appears verbatim, paraphrased or not at all, and the other domains cited.
- A public record. The log as a CSV in the repository, next to the site's other run records.
Stefan Coetzee runs one public test against these instruments: a term experiment with a prediction hashed on 2026-10-09, read by hand until the probe is built.
What this page will not publish
- A baseline before a run exists.
- Numbers from third-party tracking tools.
- A claim that a lever on this site caused a citation. The two first-party reports count citations and do not attribute them to a cause.
Sources
All pages fetched on 2026-10-09. Quotes are copied from the fetched text.
- Google Search Central Blog, Introducing Search Generative AI performance reports in Search Console, 2026-06-03.
- Google Search Console Help, Generative AI performance report (Search).
- Google Search Console Help, Search generative AI control.
- Google Search Central, Optimizing your website for generative AI features on Google Search, updated 2026-07-10.
- Microsoft Bing Webmaster Blog, Introducing AI Performance in Bing Webmaster Tools Public Preview, 2026-02-10.
- Bing Webmaster Tools Help, AI Performance.
- Perplexity, Migrate from Sonar to the Agent API.
Changelog
- v1, 2026-10-09: first version.
- v1.1, 2026-10-09: link to the term experiment at /your-next-cto/experiment/ added under the probe design.