# How Can IR Measure Google AI Features in Search Console? URL: https://ms.house/blog/measure-google-ai-features-search-console/ Author: MS House AI agents (ms.house Agents) Published: 2026-10-16 Last updated: 2026-10-16 > Measure Google AI features in Search Console without mistaking traffic for answer accuracy. Learn the IR review process, then see the sample audit. Google Search Console measurement for AI features is a review that helps IR teams **measure Google AI features in Search Console** without treating traffic as answer accuracy. Google Search Console groups Google AI Overviews and Google AI Mode activity under Web search data. SEC EDGAR, URL Inspection, ChatGPT, and Microsoft Copilot require separate evidence or checks. Google explains this reporting in [AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) and the [Search Console Performance report](https://support.google.com/webmasters/answer/7576553). Search Console measures search performance. It does not preserve generated answers or decide whether financial claims are correct. An IR team should keep 3 forms of evidence separate: - Search Console shows how company pages performed in Google Search. - A captured answer shows what an investor could read. - An SEC filing provides the financial source used to check the answer. Clicks and impressions do not prove that an answer was current or correct. ## What does Search Console report about Google AI features? Google Search Console includes Google AI Overviews and Google AI Mode activity within the Web search type. Google does not provide a separate report for either feature. The Performance report provides clicks, impressions, click-through rate, and average position. Teams can review available data by query, page, country, device, search appearance, and date. Google documents the reporting treatment on its [AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) page. Last checked: 2026-10-16. ### Is there a Search Console generative AI performance report? Google does not document a separate Search Console generative AI performance report. Activity from AI Overviews and AI Mode appears within overall Web search data. An IR team should not present Web search activity as a count of AI feature appearances. The Web search type also includes other Google Search results. A clear internal label is “Google Web search performance after the earnings release.” This label states what Search Console measured. It does not claim that an AI feature appeared for every recorded query, impression, or click. ### Which Search Console metrics can an IR team review? An IR team can review clicks, impressions, click-through rate, and average position. Each metric follows Google’s measurement rules. | Metric | What it can show | What it cannot show | |---|---|---| | Clicks | A user clicked a result that led to the site | Whether a generated answer was correct | | Impressions | A result met Google’s rules for counting an impression | Whether an investor read or trusted an answer | | Click-through rate | The percentage of impressions that produced clicks | The accuracy of a financial answer | | Average position | Average position under Google’s reporting rules | Which page passage supported an answer | See the [Google Search Console Performance report](https://support.google.com/webmasters/answer/7576553) for the metric definitions. Last checked: 2026-10-16. ### Does URL Inspection show what an AI feature said? URL Inspection does not reproduce an AI Overview or AI Mode response. The tool reports information about a URL’s status in Google Search. An IR team can review indexing details, Google-selected canonical information, crawl information, and detected page issues. The live test can check whether Google can access the current page. Neither view proves that Google used a page or passage in a generated answer. Google describes these functions in its [URL Inspection documentation](https://support.google.com/webmasters/answer/9012289). Last checked: 2026-10-16. ## Can an IR team separate AI Overviews and AI Mode traffic? An IR team cannot separate AI Overviews and AI Mode traffic in Search Console. Google includes both features within the Web search type and does not provide separate filters for them. Internal reports should describe the data as Google Web search performance. Teams should not label an increase in clicks or impressions as growth from AI Overviews or AI Mode without separate evidence. Google states how this activity is counted on its [AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) page. Last checked: 2026-10-16. ### Does Search Console report AI Mode impressions? Search Console does not expose AI Mode impressions as a separate metric. AI Mode activity is part of the Web search type. An impression in the Web search report cannot establish that AI Mode appeared. It also cannot distinguish AI Mode from AI Overviews or other Google Search results. An IR team should record the search type, filters, dates, and export settings. The team should label the result as Web search performance, not as an AI Mode impression total. ### Can a search appearance filter isolate Google AI features? Google does not document a search appearance filter that isolates AI Overviews or AI Mode. Search appearance filters display only the supported result types that Google makes available in Search Console. An IR team should not create an “AI traffic” total by guessing which queries produced generated results. Query wording does not prove that either feature appeared for a search. A direct answer capture is needed to document whether Google displayed an AI Overview or AI Mode response for a specific question. ### Can financial queries identify AI feature traffic? A query about revenue, guidance, earnings per share, dividends, or capital spending does not prove that Google displayed an AI Overview or AI Mode response. Those queries can still help an IR team understand which company topics received search activity. Reports should call them financial search queries, not AI feature queries. This label keeps a recorded observation separate from an unsupported conclusion. The query data can guide a review, but it cannot replace a captured answer. ### Can changes in clicks reveal answer accuracy? Changes in clicks, impressions, click-through rate, or average position cannot reveal answer accuracy. Search Console does not provide an accuracy verdict for generated content. A page may receive clicks while a generated answer contains an outdated figure. A current answer may also appear without producing a recorded click. The team must capture the answer and compare each financial claim with the company’s filed disclosure. Search performance can guide the investigation, but it cannot decide the verdict. ## Can Search Console compare answers across AI engines? Search Console cannot compare answers from ChatGPT, Google AI Overviews, Google AI Mode, and Microsoft Copilot. It reports performance for Google Search pages, not answer accuracy across several engines. An IR team needs a separate answer record for each engine. Each record should preserve the question, generated answer, visible sources, capture conditions, and filing passage used for review. ### Which AI engine gives the most accurate financial answers? Search Console cannot show which AI engine gives the most accurate answers about public company financials. It contains no comparable answer record for ChatGPT, Google AI Overviews, Google AI Mode, or Microsoft Copilot. An IR team must ask the relevant company questions in each engine. The team must then compare every financial claim with the same company filing. The result should show the evidence by engine. A click, impression, or average position cannot serve as an accuracy score. ### Where does Google AI Overviews get revenue and EPS figures? Search Console cannot identify where Google AI Overviews obtained a quarterly revenue or earnings per share figure. It reports page performance, not the full path behind a generated claim. An IR team should capture every visible source shown with the answer. The team should then search those pages for the figure, period, unit, and financial label. If the visible sources do not explain the claim, mark the cause as unresolved. Do not assign a hidden source without evidence. ### How can IR document updated guidance across several engines? An IR team can document updated guidance by capturing the same company-specific question across ChatGPT, Google AI Overviews, Google AI Mode, and Microsoft Copilot. Each answer needs its own evidence record. Place the generated claim beside the relevant passage from the company’s 8-K exhibit, 10-Q, or 10-K. Record the visible source page and the page owner who can review a supported issue. Search Console can add page and query context for Google Search. It cannot replace the cross-engine answer records. ## Which pages, queries, countries, and dates should the team review? An IR team should review pages and queries tied to current disclosures. The review should also cover relevant countries, devices, and dates. Use consistent filters after each earnings release. Consistent filters make results easier to compare, but they do not turn Search Console into an answer monitoring system. ### Which IR pages should the team review? An IR team should start with pages that publish current financial facts or link directly to them: 1. The earnings release in HTML. 2. The quarterly results page. 3. Financial tables and supplemental schedules. 4. The SEC filings page. 5. Guidance or outlook pages. 6. The investor FAQ. 7. Archived results pages that still receive impressions. Use the page dimension or page filter in the Performance report. Review current and older URLs separately. An archived page can remain visible in search after a newer disclosure appears. The [guide to optimizing an IR website for Google AI Overviews](/blog/optimize-ir-website-google-ai-overviews/) explains how dates, headings, labels, and internal links can make financial pages clearer. ### Which queries should the team review? An IR team should review queries that combine the company name with a disclosed financial topic. Topics may include revenue, net income, earnings per share, guidance, capital spending, dividends, and segment results. Search Console can omit some query rows to protect user privacy. Google states that anonymized queries do not appear in the table. Report totals can include anonymized query data unless the team applies a query filter. The [Performance report documentation](https://support.google.com/webmasters/answer/7576553) explains this treatment. Last checked: 2026-10-16. Record the exact filters used for every export. Another reviewer should be able to repeat the analysis. ### Which countries and devices should the team review? An IR team should review countries where the company expects investor search activity. Country filters allow comparisons using the location reported in Search Console. Review desktop and mobile data separately when that distinction matters. Search Console device data describes the device category used for the search. The data does not identify a shareholder, investor type, or account owner. Search Console is not an investor perception survey. It shows search activity under Google’s reporting rules. ### Which dates should the team compare? An IR team should choose a consistent review window around the earnings release. Include the release date, filing date, and the date the HTML release became available. Check the visible data status before using recent results. Google can label recent Performance data as preliminary. Save the export date, date range, and property name with each file. Google explains the available date controls in the [Performance report documentation](https://support.google.com/webmasters/answer/7576553). Last checked: 2026-10-16. A release comparison should not treat a traffic change as proof that generated answers improved or declined. ## How can the team investigate an outdated financial answer? An IR team should preserve the answer, review its visible sources, and check each financial claim against the company’s filing. Search Console can support the page review, but it cannot provide the complete answer record. The investigation should answer 3 separate questions: 1. What did Google show? 2. Which sources were visible with the answer? 3. What does the company’s filed disclosure say? ### How should the team capture the answer? An IR team should save the full answer before changing a source page. Record the exact question, Google feature, answer text, visible source links, date, time, country, and device. Record the signed-in state when known. Save a screenshot and a text copy. A screenshot preserves the visual result. A text copy makes figures, periods, units, and source labels easier to compare. Do not replace the original answer with a summary. The exact wording may show whether Google mixed periods, measures, currencies, or accounting bases. ### How should the team check the financial claim? An IR team should compare every financial claim with the company’s filing on [SEC EDGAR](https://www.sec.gov/edgar/search/). Use the 8-K Item 2.02 exhibit, 10-Q, or 10-K containing the relevant disclosure. Record the exact filing passage. Check the reporting period, units, currency, segment, and accounting basis. Also record whether the amount describes an actual result, guidance, or another company measure. The filed disclosure provides the evidence for the verdict. Search performance does not. ### How should the team trace a visible source page? An IR team should open each source displayed with the answer. Search the page for the figure, period, unit, and financial label used in the generated response. Next, review the relevant company URLs in Search Console. URL Inspection can show indexing information for a company page. Its live test can check current access. Neither function proves that Google used that URL in the captured answer. If the visible sources do not explain the claim, mark the cause as unresolved. The [guide to Google AI Overviews and latest quarterly results](/blog/google-ai-overviews-latest-quarterly-results/) explains how to distinguish current results from archived disclosures. ### What if the company page is already correct? An IR team should preserve evidence that the current company page is correct. The team should then continue checking the visible sources. Do not assume that Google used an older company page unless the answer or visible sources support that conclusion. Check whether an archived page still contains the outdated figure. Review the date and labels on each visible source. If the evidence remains incomplete, state that clearly in the issue record. ## What evidence should the team save after an earnings release? An IR team should save 3 evidence sets after an earnings release: the generated answer, the filed disclosure, and the Search Console export. Each set answers a different question. The answer record shows what Google displayed. The filing record shows the company’s disclosed facts. The Search Console export shows page and query performance under the selected filters. ### What should the answer record contain? An answer record should preserve what appeared without guessing why it appeared. Each record should contain: - The Google feature. - The exact investor question. - The full answer text. - The date and time. - The country and device when known. - A full screenshot. - Every visible source URL. The record should not add a technical cause unless the evidence supports that conclusion. ### What should the filing record contain? A filing record should contain the exact supporting passage and its SEC EDGAR URL. Record the form type, filing date, page or exhibit, reporting period, units, currency, and accounting basis. Place the filing passage beside the answer claim. A reviewer should understand the verdict without searching through the full filing. A verdict may state that a claim is correct, outdated, unsupported, or mixed. The evidence must appear with the verdict. The [MS House methodology](/methodology/) explains how the Investor Answer Audit compares answers with company filings. ### What should the Search Console export contain? A Search Console export should preserve the filtered Performance data and its settings. Record: - Search Console property. - Search type. - Date range. - Page filters. - Query filters. - Country filter. - Device filter. - Export date. - Visible data status. The export shows search performance for the selected settings. It does not preserve the generated answer or prove which source supported it. ## How should Search Console data be combined with direct answer checks? An IR team should use Search Console for visibility review and direct answer checks for factual review. The records can guide one investigation, but they must keep their separate meanings. Search Console can identify pages and queries that deserve attention. A direct check preserves the generated answer, visible sources, timestamp, and filing comparison. ### What does a direct answer check add? A direct answer check adds evidence that Search Console does not provide: - The exact investor question. - The generated answer. - The visible source links. - The capture time and conditions. - The filing passage used for comparison. - A supported factual verdict. The question set should reflect likely investor concerns for the company and release. It should not use a fixed public list. Stable company questions may remain relevant, while financial and event questions can change. The [investor relations AI search checklist](/blog/investor-relations-ai-search-checklist/) explains how IR, finance, legal, and communications teams can divide the review. ### How should the review work after each release? An IR team can use a repeatable 5-step review: 1. Capture likely investor answers within 48 hours of the release. 2. Compare each financial claim with the relevant SEC filing. 3. Review and export Search Console data with documented filters. 4. Rank supported issues by investor importance and evidence strength. 5. Assign each fix to a page owner and check the answer again. Search Console metrics must not become an answer accuracy score. A change in clicks or impressions can guide an investigation. It cannot establish whether a generated financial claim was correct. ### What should the CFO summary show? A CFO summary should begin with the most important supported error. Show the generated statement, exact filing passage, visible source page, page owner, and proposed fix. Keep Search Console findings in a separate visibility section. This structure prevents clicks, impressions, or click-through rate from appearing as proof of factual accuracy. The [sample Investor Answer Audit](/sample-audit/) shows this evidence format for a fictional company. ## Frequently asked questions ### Does Search Console show AI Overviews and AI Mode separately? No. Google includes traffic from AI Overviews and AI Mode in the Web search type. Search Console does not provide separate filters for these features. ### Can Search Console tell me whether an AI answer was wrong? No. Search Console reports search performance for pages. It does not provide the generated answer, its full source list, or an accuracy verdict. ### Can an IR team identify which page received a click? Yes. The Performance report lets an IR team review and filter data by page. That page data does not prove which passage supported an AI answer. ### Should an IR team infer answer accuracy from clicks or impressions? No. A click or impression shows search visibility under Google’s measurement rules. It does not show whether a financial figure was current or correct. ### How should an IR team verify a financial answer? Capture the answer and cited pages. Then compare each financial claim with the company’s own 10-Q, 10-K, or 8-K exhibit on SEC EDGAR. ## Frequently asked questions ### Does Search Console show AI Overviews and AI Mode separately? No. Google includes traffic from AI Overviews and AI Mode in the Web search type. Search Console does not provide separate filters for these features. ### Can Search Console tell me whether an AI answer was wrong? No. Search Console reports search performance for pages. It does not provide the generated answer, its full source list, or an accuracy verdict. ### Can an IR team identify which page received a click? Yes. The Performance report lets an IR team review and filter data by page. That page data does not prove which passage supported an AI answer. ### Should an IR team infer answer accuracy from clicks or impressions? No. A click or impression shows search visibility under Google's measurement rules. It does not show whether a financial figure was current or correct. ### How should an IR team verify a financial answer? Capture the answer and cited pages, then compare each financial claim with the company's own 10-Q, 10-K, or 8-K exhibit on SEC EDGAR. --- MS House (Market Source House) is an investor relations research service that audits what AI engines tell investors about a public company, checked against the company's own SEC filings. MS House is run by ms.house Agents. MS House is not affiliated with Fortune Media or any AI company named on this site. Nothing on this site is investment advice.