Google Search Console is beginning to reveal a strange side effect of conversational search: query reports can include fragments such as "yes," "yes go on," and "yes, pricing." According to an August 6 report from Search Engine Roundtable, these entries can result from follow up questions inside Google AI Mode.
Google's own explanation, quoted in the report, says that when a user asks a follow up question in AI Mode, that interaction is treated as a new query. Impression, position, and click data for the new response are counted against that follow up query. Search Console therefore contains clues about AI Mode behavior even though Google's dedicated generative AI reporting does not expose query data.
For teams already using the Google Search Console guide as the basis of organic reporting, this changes how unusual query strings should be interpreted.
AI Mode is changing what a query looks like
Traditional keyword reporting assumes that a query contains enough language to express intent on its own. "Best accounting software for a small manufacturer" is understandable without any earlier context. "Yes, pricing" is not.
Conversational search breaks that assumption because each message depends on the conversation that came before it. A user can begin with a broad question, ask for a comparison, narrow the choice, ask about price, and confirm a preference without repeating the original subject. If Search Console records each follow up as a new query, marketers can see the later fragments while missing the complete conversation that gives those fragments meaning.
That makes some Search Console data harder to interpret, but it also provides evidence of a deeper shift. Search behavior is moving from isolated queries toward sequences of questions. The reporting interface still exposes those interactions largely as separate rows.
The generative AI report and the main performance report answer different questions
The most important reporting detail is the split between Google's generative AI performance reporting and the main Search Console performance report. Search Engine Roundtable notes that generative AI reports do not provide query data, while the main performance report does. Google also combines AI Overviews and AI Mode activity into general Search Console performance data.
That means marketers should resist two opposite mistakes. First, they should not assume every unusual conversational query is definitely AI Mode. The report itself describes this as an inference. Second, they should not ignore those queries simply because Google does not label them with an AI Mode filter.
The right approach is to treat conversational query patterns as a diagnostic signal. Look for changes in query length, follow up language, pricing questions, comparison language, and sequences that appear to reflect an ongoing dialogue. Then compare those patterns with landing pages, impressions, clicks, and changes in the broader search mix.
This is also why AI search visibility needs a measurement layer beyond classic rank tracking. A brand can participate in an AI generated answer without the interaction looking like a traditional ten blue links search.
Search Console can reveal demand even when attribution is incomplete
Search Console has always been valuable because it shows how real users phrase demand. AI Mode makes that data less tidy, but potentially more revealing.
A fragment such as "yes, pricing" tells you very little as a keyword. As evidence of user behavior, it tells you that pricing can emerge as a follow up after an earlier answer. Similar patterns may appear around implementation, alternatives, compatibility, risk, timelines, features, or local availability.
For content teams, the implication is practical. Do not build every page around one primary keyword and assume the user's information need ends there. Map the next questions a buyer is likely to ask after receiving the first answer. This is close to the logic behind an AEO strategy: organize content around questions, evidence, and useful answers rather than keyword repetition alone.
Do not overfit content to strange query fragments
The appearance of "yes" or "yes go on" in Search Console does not mean companies should create pages optimized for those exact strings. Those phrases are artifacts of conversational context. A page titled "Yes Go On Pricing" would misunderstand what the data represents.
The useful work happens one level above the literal query. Ask what previous question could have produced the follow up. Ask what topic, product, service, or comparison led the user to request pricing. Ask whether your content answers that next stage clearly.
Search Console therefore becomes a prompt discovery tool as much as a keyword tool. It can show marketers where the user's journey continues after an initial answer, even when it cannot reconstruct the entire AI Mode conversation.
Reporting needs more context, not more certainty
The temptation with new AI search data is to create a precise label before the data supports one. That would be a mistake. Google currently gives marketers partial evidence. Some AI Mode activity is included in general performance reporting, follow up interactions can generate unusual query strings, and the dedicated generative AI reports omit queries.
A useful report should state those limits openly. Separate directly observed Search Console data from inferred AI Mode behavior. Track conversational query patterns over time. Keep conventional SEO metrics such as clicks and landing page performance. Add AI visibility monitoring where it is useful, but do not pretend that one system provides complete attribution.
This is the broader measurement challenge of AI search. The interfaces are becoming conversational faster than analytics systems are becoming conversational. Search Console's strange new queries are a small but valuable window into that transition.
Originally published on the Mustard Seed blog.
Top comments (0)