Knowledge base v2.13.0 · Community edition · data through 2 October 2026
Day 3 · Friday 2 October 2026 · 14:20
Nik Vujic (Founder, Get Stuff Digital) on Search Console data through BigQuery and an LLM agent, named from his self-introduction, which matches the official speaker list; then a community speaker who was not named, on AI visibility and brand familiarity. Other talks of this block were not captured.
A community slide (Workflow 1) said the first step is to apply the filters you need, even natively in Search Console, for example branded versus non-branded queries, with clicks and impressions.
“Apply the filters you need, even natively in GSC. For example branded vs non-branded, with clicks and impressions.”
Wording checked against the slide or recording
Speaker Nik VujicEvidence slide photo, transcript
A community slide said the Search Console interface export is limited to 1,000 rows.
“The UI export is limited to 1,000 rows.”
Wording checked against the slide or recording
Speaker Nik VujicEvidence slide photo
Used byrequirement DEV-MON-08
A community slide said the Search Console bulk data export to BigQuery gives the most complete query dataset available, with anonymized queries excluded.
“The bulk export to BigQuery gives the most complete query dataset available - anonymized queries excluded.”
Wording checked against the slide or recording
Speaker Nik VujicEvidence slide photo
Used byrequirement DEV-MON-08glossary term Search Console bulk data export
The community workflow's third step feeds the exported Search Console dataset into the agency's LLM agent, which queries the dataset and returns the evidence rather than a sample.
Speaker Nik VujicEvidence slide photo, transcript
Nik Vujic's slide said no agent is needed to start: pull the Search Console data, import it into any LLM and query it in conversation; his agency built its agent to have everything in one place.
“No agent needed to start: pull the data and import it into the LLM of your choice, and talk with it.”
Wording checked against the slide or recording
Speaker Nik VujicEvidence slide photo, transcript
In a community lightning talk, agency founder Nik Vujic presented how his agency feeds Search Console and GA4 data to an LLM agent to support decisions for its clients.
Speaker Nik VujicEvidence transcript
Nik Vujic called Google Search Console his favourite SEO tool, used alongside GA4, and said third-party SEO tools give only estimates, although they have some use cases.
Speaker Nik VujicEvidence transcript
Nik Vujic said it is super important to export all of a site's Search Console data, and his agency's workflow starts from Search Console's export to BigQuery.
Speaker Nik VujicEvidence transcript
Used byrequirement DEV-MON-08glossary term Search Console bulk data export
Nik Vujic recommended Looker Studio as a second way to get past Search Console's data limits.
Speaker Nik VujicEvidence transcript
Nik Vujic said his agency uses the Search Console data fed to an LLM in five use cases, the first being cross-comparing and following up different articles.
Speaker Nik VujicEvidence transcript
Nik Vujic said that after a page is published, the queries it appears for are the first thing to show up in Search Console, while clicks and impressions come much later.
Speaker Nik VujicEvidence transcript
To report content refreshes on large sites, Nik Vujic's agency compares date ranges and has an LLM give the exact number of queries the content improved on, then reports that number to the client.
Speaker Nik VujicEvidence transcript
Nik Vujic said Google loves fresh, quality content, as a reason for large sites to refresh their content.
Speaker Nik VujicEvidence transcript
Nik Vujic said analysing Search Console query data with an LLM also helped his agency spot developer mistakes, such as unwanted pages that got indexed, which can also be found in Search Console itself.
Speaker Nik VujicEvidence transcript
Nik Vujic said some keyword cannibalization is fine because it is logical, but for bad cannibalization his agency groups all of an article's Search Console queries and then decides whether to redirect the pages or rewrite and split the content.
Speaker Nik VujicEvidence transcript
Nik Vujic advised starting every SEO experiment with a hypothesis, for example a bulk change to one page element across a group of pages, and using an LLM to report the group's Search Console data, for example click-through rate, for the test.
Speaker Nik VujicEvidence transcript
Nik Vujic called the GA4 API combined with Looker Studio the only LLM traffic tracker that actually works.
Speaker Nik VujicEvidence transcript
Nik Vujic said GA4 setups that separate out traffic from LLMs do not capture visits from AI Overviews and AI Mode.
Speaker Nik VujicEvidence transcript
Nik Vujic said Google Tag Manager events can be used to follow real traffic arriving from different LLMs.
Speaker Nik VujicEvidence transcript
Nik Vujic's agency puts Search Console data, LLM referral traffic from GA4 and the data from Search Console's generative AI report into one LLM prompt to compare a client's performance in classic search and in AI features.
Speaker Nik VujicEvidence transcript
Nik Vujic said he hopes Search Console's generative AI report will get query data.
Speaker Nik VujicEvidence transcript
Nik Vujic said his agency uses the combined data to test whether good SEO is good GEO, and that the answer is 'it depends': for one client it found pages that performed better in generative AI features and LLMs than in standard search.
Speaker Nik VujicEvidence transcript
A community speaker described a business whose visibility and average rankings for many important, high-value commercial keywords improved over a year while its clicks and impressions did not rise with them.
Speaker not identifiedEvidence transcript
A community speaker argued that SEOs who panic when rankings rise but clicks do not are relying on a metric, rankings and visibility, that no longer stands up to scrutiny.
“we rely on a metric that no longer stands up to scrutiny”
Speaker not identifiedEvidence transcript
A community speaker said most people in the room had probably seen clicks decline over the past couple of years because AI now takes part of the demand by summarising information from websites, a shift that hit informational queries first.
Speaker not identifiedEvidence transcript
A community speaker said the loss of clicks on informational queries to AI summaries probably happened for a good reason, with benefits for the user journey (the word 'benefits' is an uncertain reading of the recording).
Speaker not identifiedEvidence transcript
In data a community speaker tracks for particular brands, local search terms have started to lose clicks too as AI features appear on more results pages.
Speaker not identifiedEvidence transcript
A community speaker reported, calling it worrying, that AI features have started to appear more often on results pages for commercial search terms too, in data the speaker tracks for particular brands.
Speaker not identifiedEvidence transcript
A community speaker said the problem SEOs face is not ranking but positioning SEO differently and measuring and showing its value to stakeholders while AI features take away clicks, some of which were probably not relevant anyway.
Speaker not identifiedEvidence transcript
Key stakeholders such as heads of department just want to see metrics and numbers, a community speaker said, which makes the way SEO reports its value the question to solve.
Speaker not identifiedEvidence transcript
A community speaker said digital marketing has been spoiled by about 20 years of plentiful attribution and that SEOs should relearn the mindset of marketers who worked with very little of it, while still using the performance data that remains.
Speaker not identifiedEvidence transcript
A community speaker advised SEOs to focus, as marketers did before detailed attribution, on downstream impact: the effect their work has on other channels.
Speaker not identifiedEvidence transcript
Used byglossary term Downstream impact
A community speaker cited a recent Similarweb study as showing that being recommended in AI-driven services multiplies a brand's chance of getting traffic downstream; a Similarweb study reported in June 2026, probably the one meant, found brands recommended by ChatGPT 2.5 times more likely to get a site visit within 7 days (US desktop data, finance, travel and beauty).
Speaker not identifiedEvidence transcript
Traffic that follows an AI recommendation need not arrive through organic search but may come direct, through paid search or through another channel, a community speaker said, predicting the effect will show in data in the coming months and years.
Speaker not identifiedEvidence transcript
Used byglossary term Downstream impact
A community speaker concluded that rankings and clicks no longer equal real business outcomes.
“rankings and clicks don't really equal real outcomes anymore”
Speaker not identifiedEvidence transcript
A community speaker ran a controlled test with a brand that had seen organic clicks drop: for a set period the only activity was work to raise the brand's AI visibility, with no external campaigns and no budget changes.
Speaker not identifiedEvidence transcript
A community speaker reported that a higher rate of recommendation in AI answers brought one brand a really significant uplift in inquiries during a controlled test, without giving a figure.
Speaker not identifiedEvidence transcript
A community speaker said the brand in the controlled test had a record month, the kind of real business value that business owners care about more than a dip in traffic.
Speaker not identifiedEvidence transcript
During a community speaker's controlled test, the brand's paid search click-through rate rose by 36% while the only activity was work to raise its AI visibility.
Speaker not identifiedEvidence transcript
In a community speaker's controlled test the brand's AI visibility was tracked across the main AI engines and was described as really strong early in the experiment (whether as a starting level or a gain was not said).
Speaker not identifiedEvidence transcript
In the same controlled test the brand's top-of-page share in paid search dropped, so a slight dip in paid click-through rate would have been expected rather than the 36% rise, a community speaker said.
Speaker not identifiedEvidence transcript
A community speaker advised SEOs to work with paid search teams to understand changes such as a rise in paid click-through rate and what impact SEO work may have had, since those teams may not credit SEO for it.
Speaker not identifiedEvidence transcript
A community speaker cited a study from about 10 years ago, repeated since, as finding that 82% of people clicked on a brand they already knew regardless of its position; the matching source is Red C's eye-tracking study of shopping-type searches, reported by Econsultancy in October 2018 (about eight years before the event).
Speaker not identifiedEvidence transcript
A community speaker suggested that people who meet a brand at several touchpoints, including in AI services, become familiar with it and are then more likely to choose it in a generic search, which can lift other metrics.
Speaker not identifiedEvidence transcript
A community speaker said brand familiarity should be built through AI-driven services and earned media as a whole, by improving content and visibility across services ('earned media' is an uncertain reading of the recording).
Speaker not identifiedEvidence transcript
A community speaker said the customer journey that starts in AI services still exists and only the direct attribution is missing, so SEOs should look at wider metrics they have not tracked before.
Speaker not identifiedEvidence transcript
A community speaker's simplified journey: an AI recommendation makes a visit more likely; the person may check the brand on social media or see it while scrolling, search the brand name, and later click it in a generic search because they know it.
Speaker not identifiedEvidence transcript
When a brand bids on its own name, people who search the brand after an AI recommendation may arrive through paid search rather than organic, so organic reporting loses that visit, a community speaker noted.
Speaker not identifiedEvidence transcript
Clicks are down for a lot of businesses but rarely tell the whole story, a community speaker summarised, advising SEOs to ask the paid search team for Google Ads click-through rate reports.
Speaker not identifiedEvidence transcript
A community speaker advised SEOs to report conversion rates and traffic by channel, not only organic clicks.
Speaker not identifiedEvidence transcript
For the brand in a community speaker's controlled test, organic sessions in GA4 roughly halved while direct sessions stayed fairly stable (an uncertain reading; a change of about 2% was mentioned, direction unclear in the recording).
Speaker not identifiedEvidence transcript
A community speaker advised comparing the relative performance of channels and choosing new metrics to report, rather than judging SEO on organic sessions alone.
Speaker not identifiedEvidence transcript
A fully controlled test of AI visibility was possible for one brand only because it ran no other marketing activity, and is not viable for most brands, a community speaker said.
Speaker not identifiedEvidence transcript
A community speaker recommended picking areas of a business where testing is possible and measuring how AI visibility affects conversion rates and revenue there.
Speaker not identifiedEvidence transcript
Google's launch post for Search Console's branded queries filter (20 November 2025) says the branded versus non-branded split is made by an internal AI-assisted system, not by a regular expression, and that some queries may occasionally be misidentified.
Publisher Search Central blog (20 November 2025, updated 11 March 2026)
Used byglossary term Branded queries filter
A Search Central blog post on Search Console data limits (October 2022) says the interface exports at most 1,000 rows, while the Search Analytics API and the Looker Studio connector return up to 50,000 rows per day per site per search type.
Publisher Search Central blog (19 October 2022)
Used byrequirement DEV-MON-08
Google's blog post on Search Console data limits says anonymized queries are always left out of the report tables but counted in chart totals unless you filter by query, and are left out whenever a filter is applied, so filtered rows do not add up to the chart totals.
Publisher Search Central blog (19 October 2022)
Used byrequirement DEV-MON-08
Google's AI features guide says traffic from sites appearing in AI features such as AI Overviews and AI Mode is included in the overall search traffic in Search Console and reported in the Performance report within the Web search type.
Publisher Google Search Central
Used byrequirement DEV-MON-07
Search Console's branded queries filter works only for top-level properties, not URL-path or subdomain properties, and only for sites with enough queries and impressions; Google made it available to all eligible sites on 11 March 2026.
Publisher Search Central blog (20 November 2025, updated 11 March 2026)
Used byglossary term Branded queries filter
In an August 2025 post, Google's Head of Search wrote that total organic click volume from Google Search to websites had been relatively stable year-over-year and that Google was sending slightly more quality clicks (clicks where users do not quickly click back) than a year earlier; the post gave no figures.
Publisher Google blog (6 August 2025)
Google's August 2025 post on AI in Search said traffic is shifting between sites, decreasing for some and increasing for others, and that for some questions where people want a quick answer they may be satisfied with the response and not click further.
Publisher Google blog (6 August 2025)
Google's AI features guide says clicks from search results pages with AI Overviews are higher quality, meaning users are more likely to spend more time on the site.
Publisher Google Search Central
Beyond Search Console, Google's AI features guide suggests tracking conversions and time spent on the site in tools such as Google Analytics to understand the value of traffic from AI features.
Publisher Google Search Central
A May 2025 Search Central blog post advised site owners to look at the overall value of visits from Search rather than focusing too much on clicks, using indicators of conversion such as sales, sign-ups, a more engaged audience or information lookups about the business.
Publisher Search Central blog (21 May 2025)
In the bulk export, 'anonymized queries excluded' means the query text, not the traffic: Google's table reference and query guidelines show those rows kept with an empty query field, often as the single most common 'query'. Tell an LLM to skip empty queries when ranking top queries, but keep them in totals.
Author Ibrahim Anjro
Used byrequirement DEV-MON-08
A query reaches Search Console's Performance report only through an impression, so queries cannot come before impressions; what rises first after a publish or refresh is the number of queries with a few impressions each, before total impressions and clicks grow noticeably.
Author Ibrahim Anjro
Google's ranking systems guide describes 'query deserves freshness' systems that show fresher content where it would be expected, which is narrower than a general preference for fresh content; refresh pages whose queries expect current information, and judge other refreshes by quality.
Author Ibrahim Anjro
Pages that do better in AI answers than in classic results do not refute Google's line that optimising for people is optimising for generative AI Search, since the comparison mixed Google's AI features with third-party LLMs; measure each surface separately, per page, before deciding what to change.
Author Ibrahim Anjro
Because Google says anonymized queries are left out whenever a filter is applied, branded plus non-branded exports should not be expected to add up to the unfiltered chart totals; tell an LLM that sums filtered exports that the gap is expected, not missing data.
Author Ibrahim Anjro
Before feeding a branded versus non-branded split to an LLM, spot-check a sample of queries in each group, since Google's AI-assisted classification can misidentify queries; on URL-path or subdomain properties, where the filter is unavailable, a regex query filter is the fallback.
Author Ibrahim Anjro
Looker Studio lifts Search Console's 1,000-row interface limit but keeps the connector's cap of 50,000 rows per day per site per search type; only the bulk export to BigQuery is free of the daily row limit, so large sites that need every non-anonymized query should use it.
Author Ibrahim Anjro
Used byrequirement DEV-MON-08
The click decline described on stage is what many individual sites report, while Google's published position (August 2025) is that total organic clicks are relatively stable but shifting between sites, with fewer clicks for quick-answer questions; neither side gave figures, so measure the trend per site and per query type rather than assuming a market-wide fall.
Author Ibrahim Anjro
The community speaker's figures (paid search click-through rate up 36%, organic sessions halved, a record month) come from one unpublished test of one unnamed brand with no stated period or baseline; use them as an illustration, not as a benchmark.
Author Ibrahim Anjro
Google's own line on Day 1 was to measure success by metrics that matter to the business, and its Generative AI performance report shows impressions but no clicks; together with the community talk this argues for a report that puts organic next to direct, branded paid search, conversions and revenue.
Author Ibrahim Anjro
For a site's own test of AI visibility, write down the period, the baseline and every other marketing change before starting, and compare a test area with a similar control area; paid click-through rate and direct and branded traffic are the signals the community speaker watched.
Author Ibrahim Anjro
The two studies cited on stage are third-party and narrow (Similarweb, 2026: ChatGPT only, US desktop, three industries; Red C, 2018: eye-tracking on shopping-type searches), so quote them with that scope; Similarweb found that 55.9% of the resulting site traffic came from branded searches, the step after an AI recommendation in the speaker's customer journey.
Author Ibrahim Anjro
In a community lightning talk, agency founder Nik Vujic presented how his agency feeds Search Console and GA4 data to an LLM agent to support decisions for its clients.
A community speaker who is not a developer automated monthly SEO reports with Python in Visual Studio Code, using Claude and ChatGPT as coding partners and Google Cloud for access to the Search Console API.
Google's launch post for Search Console's branded queries filter (20 November 2025) says the branded versus non-branded split is made by an internal AI-assisted system, not by a regular expression, and that some queries may occasionally be misidentified.
Google said query groups are not built by matching query text alone: deciding which queries belong in which group goes beyond simple text filtering.
The community workflow's third step feeds the exported Search Console dataset into the agency's LLM agent, which queries the dataset and returns the evidence rather than a sample.
An agency's automated monthly SEO report follows four simple steps: the data comes in, a script processes it, AI summarises it and the result goes onto a dashboard.
Nik Vujic's slide said no agent is needed to start: pull the Search Console data, import it into any LLM and query it in conversation; his agency built its agent to have everything in one place.
An agency's automated monthly SEO report follows four simple steps: the data comes in, a script processes it, AI summarises it and the result goes onto a dashboard.
Google's ranking systems guide describes 'query deserves freshness' systems that show fresher content where it would be expected, which is narrower than a general preference for fresh content; refresh pages whose queries expect current information, and judge other refreshes by quality.
Freshness is a signal for queries that deserve fresh results ('query deserves freshness'): when a breaking event hits a city, such as possible closure of Barcelona's airport, users want really fresh results, not results from two weeks ago.
Nik Vujic said GA4 setups that separate out traffic from LLMs do not capture visits from AI Overviews and AI Mode.
The regular Search Console Performance report still shows all traffic including AI features, while the generative AI view shows only impressions from AI surfaces.
Nik Vujic said he hopes Search Console's generative AI report will get query data.
Google's launch post for the Generative AI performance reports in Search Console (3 June 2026; rolled out to all sites worldwide by 31 August 2026) lists impressions, pages, countries, devices (Search only) and dates, and no click or query metrics.
Nik Vujic said he hopes Search Console's generative AI report will get query data.
Google called AI reporting in Search Console an evolving space and expects the generative AI report to get richer, with more information in the future.
Pages that do better in AI answers than in classic results do not refute Google's line that optimising for people is optimising for generative AI Search, since the comparison mixed Google's AI features with third-party LLMs; measure each surface separately, per page, before deciding what to change.
Optimising for people is optimising for generative AI Search.
Google's AI features guide says traffic from sites appearing in AI features such as AI Overviews and AI Mode is included in the overall search traffic in Search Console and reported in the Performance report within the Web search type.
The regular Search Console Performance report still shows all traffic including AI features, while the generative AI view shows only impressions from AI surfaces.
Before feeding a branded versus non-branded split to an LLM, spot-check a sample of queries in each group, since Google's AI-assisted classification can misidentify queries; on URL-path or subdomain properties, where the filter is unavailable, a regex query filter is the fallback.
Review the filters and regexes that Search Console's AI-powered configuration proposes before trusting the numbers: Google's launch post calls the feature experimental, warns that AI can misinterpret requests, and limits it to the Search results Performance report (not Discover or News) and to configuration, not sorting or exporting.
A community speaker concluded that rankings and clicks no longer equal real business outcomes.
Myth: the old metrics don't work in the AI era. Google's answer: measure success through metrics that matter to your business.
Beyond Search Console, Google's AI features guide suggests tracking conversions and time spent on the site in tools such as Google Analytics to understand the value of traffic from AI features.
Google's guide for generative AI features recommends the Generative AI performance report in Search Console for measuring how content performs in generative AI features on Google Search and Discover.
A May 2025 Search Central blog post advised site owners to look at the overall value of visits from Search rather than focusing too much on clicks, using indicators of conversion such as sales, sign-ups, a more engaged audience or information lookups about the business.
Myth: the old metrics don't work in the AI era. Google's answer: measure success through metrics that matter to your business.
Google's own line on Day 1 was to measure success by metrics that matter to the business, and its Generative AI performance report shows impressions but no clicks; together with the community talk this argues for a report that puts organic next to direct, branded paid search, conversions and revenue.
Myth: the old metrics don't work in the AI era. Google's answer: measure success through metrics that matter to your business.
Google's own line on Day 1 was to measure success by metrics that matter to the business, and its Generative AI performance report shows impressions but no clicks; together with the community talk this argues for a report that puts organic next to direct, branded paid search, conversions and revenue.
Google's launch post for the Generative AI performance reports in Search Console (3 June 2026; rolled out to all sites worldwide by 31 August 2026) lists impressions, pages, countries, devices (Search only) and dates, and no click or query metrics.