What Are the Search Console Generative AI Performance Reports in 2026?
They are Search Console reports, announced on 3 June 2026, that show how often your pages appear inside Google's generative AI features. There are two: one for Search, covering AI Overviews and AI Mode, and one for Discover, covering generative AI features in Discover. Google initially rolled them out to a subset of websites and states that as of 31 August 2026 they are available to all websites worldwide.
For anyone investing in AI search visibility, this is the first first-party measurement Google has offered. Before June 2026, the only way to estimate whether your content was appearing in AI Overviews was third-party tooling that sampled queries from outside. Now Google reports impressions for your own property directly, by page, country, device and date.
| Attribute | What Google published in 2026 | What it means for you |
|---|---|---|
| Announced | 3 June 2026, initially to a subset of websites | Early adopters saw it first. |
| Full rollout | "As of August 31, 2026, we've rolled out these insights to all websites worldwide." | Indian properties should have it now. |
| Search report covers | AI Overviews and AI Mode | Google expects to update this list over time. |
| Discover report covers | Generative AI features in Google Discover | A separate report with different counting rules. |
| Main metric | Impressions | How often links to your site were shown in AI features. |
| Dimensions | Pages, Countries, Dates; Devices for the Search report | Which pages, where, when, on what device. |
| Time granularity | Hourly, daily, weekly and monthly, per the announcement | Fine enough to see the effect of a content change. |
| Search Labs | Search Console doesn't include data from experiments in Search Labs | Experimental features are not counted. |
Why Do These Reports Matter for GEO and AEO in 2026?
Because they answer the question every buyer of AI visibility work eventually asks: how would we even know it worked? Until 2026 that question had no first-party answer. An agency could show screenshots or third-party estimates, but could not show the client's own Google data. The generative AI performance reports change that, which makes AI visibility a measurable discipline rather than a promise.
They also make AI visibility comparable to everything else in Search Console. Google states that AI feature data is still included in the overall Performance report, and that the new reports give a separate view dedicated to generative AI visibility. So you can now see what share of your total exposure comes from AI features, which pages earn it, and whether that share moves after you change content.
- For buyers: ask any GEO or AEO provider to report from this data, not from screenshots.
- For content teams: find which pages already appear in AI features and study what they have in common.
- For leadership: track AI feature impressions as a line next to classic search impressions.
- For international teams: the Countries dimension shows where your AI visibility actually comes from.
How Do You Read the Search Generative AI Report in 2026?
The default view shows impressions for your site in the supported AI features, with a chart above a table. Google defines impressions here as how many times links to your site were shown to a user in a generative AI feature on Google Search. You change the table grouping with the tabs above it, and filter by search type at the top of the report.
| Dimension | How Google defines it |
|---|---|
| Pages | The final URL linked by a generative AI feature after any redirects. Most data is assigned to the page's canonical URL, not a duplicate. |
| Countries | The country where the search originated. |
| Dates | Days, weeks or months by selected granularity. All dates are in Pacific Time (PT). |
| Devices | Desktop, tablet or mobile. |
The search type filter has two web options: "Web: text-based" for searches typed into the standard Google search bar, and "Web: multimodal" for searches made with images, including Google Lens, Circle to Search on Android, uploaded images and Chrome image search. The multimodal reporting section covers that filter in detail.
Three reading rules that prevent bad conclusions
- Chart totals are aggregated by property. Google says that if two results from the same site appeared in one generative AI feature, they count as a single impression in the chart total. Add a URL filter and the chart aggregates by URL instead.
- Table totals can differ from chart totals. Data grouped by page is aggregated by page; grouped by country, device or date it is aggregated by property. Google says discrepancies are usually due to this difference.
- The newest data can be preliminary. It is shown as a dotted line and may change over the next few hours.
How Is the Discover Generative AI Report Different in 2026?
It counts differently, and treating the two reports as the same will produce wrong comparisons. In the Discover report, an impression requires the link to be scrolled into view, and only one impression is counted per result per session, so scrolling past a card and back does not add a second impression. All Discover AI data is aggregated by page, so two AI results from the same property in the same Discover list are counted separately.
- Pages means the source, not the landing page. Google defines it as the page that served as the source of the information shown, using the canonical URL, not the page where the user lands after a click.
- Countries means where the content was viewed. Not where a search originated, because Discover has no search.
- Daily chart values may be omitted below a threshold, but Google says they are still included in the totals, which are accurate and not truncated.
- There is no Devices dimension described for the Discover report.
What Do the Generative AI Reports Not Tell You in 2026?
They report impressions, and that limit shapes everything a team should claim from them. The June announcement lists impressions, pages, countries, devices and dates, and says Google is working with site owners on additional metrics over time. Nothing in the announcement or the help documentation describes AI-specific clicks, the text of the AI answer, the query that triggered it, or your position within the AI feature.
- Impressions, not influence. You learn that a link to your page was shown, not how prominently or how it was described.
- No answer text. The report does not show what the AI feature said.
- No conversions. As with all Search Console data, nothing after the click.
- Standard limits apply. Google notes the usual performance report limitations, including the 1,000-row limit, also apply.
- Search Labs experiments are excluded. Experimental AI features are not counted.
- Missing values export as zeros. Values shown as "~" or "-" become zeros in downloaded data, so blank is not the same as measured zero.
The honest way to report this data in 2026 is "pages from our site were shown in AI features this many times, rising or falling", alongside the content changes made in the same period. Claims about answer share, citation prominence or AI-driven revenue need other evidence.
Why Might the Report Be Missing or Empty in 2026?
Google's help centre gives three reasons. Not all properties may have access yet as the rollout progresses, although Google says all websites worldwide were covered as of 31 August 2026. Your site may not have received enough impressions in generative AI features. Or your site may be excluded from Search generative AI features through the Search generative AI control.
The control that decides whether you appear at all
The Search generative AI control, rolled out to all websites worldwide as of 31 August 2026, lives under Settings, then Search generative AI, in Search Console. It applies to AI Overviews, AI Mode and generative AI features in Google Discover, with three options.
| Option | What Google says it does |
|---|---|
| Include | Your content can appear in Search generative AI features, as links and to help ground AI responses. The default for all properties. |
| Exclude | Your content is prevented from appearing in these features or grounding their responses. You receive no traffic or impressions from them. |
| Inherit from parent | The property follows its parent property's setting. The default when a parent property exists. |
Google adds important boundaries. The control is not a ranking or inclusion signal for other parts of Search. It does not affect AI training; Google points to Google-Extended for that. Exclusions generally take a few days, with content excluded within one to two days after the control goes live, though caching can take longer. If a client's AI report is empty, check this setting, including any parent property it inherits from, before anything else.
How Should Teams Use These Reports Month to Month in 2026?
Use them as a scoreboard for specific content changes rather than as a vanity number. The useful questions are which pages earn AI impressions, which pages should earn them and do not, and whether a change to a page moved its AI exposure. Hourly and daily granularity makes that testable.
- Baseline: export 28 days of AI impressions by page and by country.
- Gap list: compare against your priority pages in the main Performance report. High classic impressions with no AI impressions is the gap.
- Change: improve a small batch of pages with direct answers, clear structure and factual specifics, following our AI Overviews guide.
- Annotate: mark the change date in Search Console.
- Compare: after a few weeks, compare the changed batch against an unchanged control batch.
- Report honestly: impressions moved, these pages, this period, alongside what changed.
How Do the 2026 Reports Fit Together?
Search Console now gives a business four first-party views of its discovery. The classic Performance report for search results. The generative AI reports for AI Overviews, AI Mode and Discover AI features. The multimodal search type for image-started searches. And platform properties for Instagram, TikTok, X and YouTube content in Google. Together that is the most complete visibility picture Google has ever offered site owners, and all of it is free.
What Is Multimodal Search Reporting in Search Console?
Multimodal search reporting is a new search type in Google Search Console, announced on 24 September 2026, that shows how your pages perform when people search with images instead of typed words. Google says the data covers searches with Lens, Circle to Search on Android, image uploads to Google Search, and the Chrome right-click "Search this image" feature. It appears both in the Performance report for Search results and in the Generative AI performance report.
Until this release, a visit that began with someone pointing a phone camera at a product was folded into your general web numbers, with no way to separate it out. In 2026 that separation exists for the first time in first-party data, which matters for any business whose buyers start from something they can see: furniture, fashion, packaging, spare parts, plants, food, signage, or a competitor's product on a shop shelf.
| Attribute | What Google published in 2026 | What it means for you |
|---|---|---|
| Announced | 24 September 2026, on the Search Central blog | New enough that most agency reports do not use it yet. |
| Surfaces included | Lens, Circle to Search on Android, image uploads to Google Search, Chrome right-click "Search this image" | All image-started searches are grouped into one type. |
| Where it appears | Performance report for Search results, and the Generative AI performance report | You can see image-started visibility in both classic results and AI features. |
| How to view it | The new multimodal search type filter in the performance reports | A filter, not a separate report. |
| Filter labels in the help centre | "Web: text-based" and "Web: multimodal" | Compare the two directly. |
| Rollout | Rolling out globally starting 24 September 2026 | Not a US-only test. Indian properties are included in the rollout. |
| Visibility condition | Metrics appear if your site is receiving traffic from these queries | An empty view can simply mean no image-started traffic yet. |
| Export | "Export" downloads the data | Pull it into your own dashboards. |
Why Does Image-Started Search Matter for Businesses?
Because a growing share of discovery no longer starts with a typed query, and until now nobody could measure it on their own site. A shopper who photographs a lamp in a cafe, a mechanic who circles a part on a video, or a gardener who uploads a picture of a leaf has a buying intent that is often more specific than any keyword. Search Console now tells you which of your pages those people reached.
The commercial reason this matters in 2026 is that visual intent rewards a different kind of page. A page that ranks for a typed query is usually won with words. A page that is surfaced from a photo is usually won with images, clear product identity and structured information about what the object actually is. Without separate data, a team cannot tell whether its product photography, image file names and product markup are earning visibility, so it cannot justify investing in them. Now it can check.
- Product businesses: anything a customer can photograph before they know its name is a candidate for image-started discovery.
- Local and retail: storefronts, menus and signage are photographed far more often than they are typed.
- B2B components: part numbers are rarely known; a photo of the part is.
- Publishers: how-to and identification content often answers a question a camera asked.
How Do You Find Multimodal Data in Search Console?
Open the Performance report for Search results and use the search type filter at the top of the report. Google describes a new multimodal search type in the performance reports, and the help centre for the generative AI report lists two web options: "Web: text-based", for traffic from text queries entered into the standard Google search bar, and "Web: multimodal", for traffic when users search using images.
- Open Search Console and select your property.
- Open the Performance report for Search results.
- Open the search type filter at the top of the report and choose the multimodal option.
- Read clicks, impressions and pages as you would for any other search type.
- Repeat inside the Generative AI performance report to see image-started impressions in AI features.
- Click "Export" to take the data into your own spreadsheet or dashboard.
If nothing appears, Google's announcement gives the reason directly: you will start seeing these metrics if your site is receiving traffic from these queries. An empty multimodal view in October 2026 is a finding in itself, not a fault.
Run the same date range twice, once on "Web: text-based" and once on "Web: multimodal", grouped by page. Pages that over-index on multimodal are your visual winners. Pages that should be visual winners and are absent, a product page with no image impressions, are your first fix list.
What Does the Multimodal Filter Not Tell You?
It tells you a lot less than a typed-query report, and being clear about the gaps stops a team from over-reading the numbers. Google's announcement and help documentation describe one combined multimodal search type covering four surfaces. They do not describe a way to separate Lens from Circle to Search, image uploads or Chrome's right-click search inside that type, so treat the figure as a single bucket.
- No per-surface split is described. Lens, Circle to Search, uploads and Chrome image search are reported together.
- No image is shown back to you. The documentation does not describe seeing which photo a user searched with.
- Data appears only with traffic. Low-traffic sites may see nothing for some time.
- Usual report limits apply. The generative AI help page notes that the standard performance report data limitations, such as the 1,000-row limit, also apply.
- No conversion data. Like all Search Console reports, it stops at the click. Sales and leads still come from your analytics.
None of these are reasons to ignore the report. They are reasons to report it honestly. "Image-started impressions on product pages rose" is a defensible statement. "Lens drove these sales" is not, because the data does not say that.
How Should Teams Improve Their Multimodal Visibility?
Treat the multimodal number as the scoreboard and image quality, product identity and structured information as the work. Google's announcement says the purpose is to help you understand your site's performance and optimise your visual content. The levers below are standard Google Search good practice applied to the pages that should be winning image-started searches.
| Lever | What to do in 2026 | How to check it worked |
|---|---|---|
| Original product photography | Use your own clear images of the actual item, from the angles a buyer would photograph it. | Multimodal impressions on those pages over the following weeks. |
| Descriptive file names and alt text | Name and describe what the object is, in plain language, not "IMG_4412". | Compare multimodal impressions before and after on a sample of pages. |
| Clear product identity on the page | State the product name, model, material and use near the image. | Pages appearing in "Web: multimodal" grouped by page. |
| Structured data where relevant | Product and other applicable markup so systems can read what the page is about. | Search Console enhancement reports show no errors. |
| One page per real product | Avoid near-duplicate pages that split the signal. | Fewer, stronger pages in the multimodal table. |
Add an annotation in Search Console on the day you change a batch of images, so the before and after comparison is not guesswork. Our Search Console guide covers the wider tool, and the generative AI performance reports guide covers the AI side of the same data.
How Does Multimodal Data Connect to AI Search Visibility?
Directly, because Google placed the new search type inside the Generative AI performance report as well as the main Performance report. That means a team can now see impressions in AI features such as AI Overviews and AI Mode that started from an image rather than a typed question. For any business selling something visual, this closes a measurement gap in AI visibility work that was otherwise invisible.
This is the part most relevant to GEO and AEO work in 2026. Optimising content for AI answers has mostly been discussed as a text problem. The multimodal search type shows that a portion of AI feature exposure begins with a camera, which makes image clarity and product identity part of AI visibility rather than a separate image SEO task.
What Are the Common Mistakes?
- Reading "no data" as "no opportunity". Metrics only appear once your site receives image-started traffic.
- Claiming the number for one surface. Lens, Circle to Search, uploads and Chrome image search are reported together.
- Comparing different date ranges across filters. Always compare text-based and multimodal over the same period.
- Changing images without an annotation. You lose the ability to show cause and effect.
- Using stock photography on product pages. An image identical to hundreds of others gives a visual search little reason to pick yours.
- Ignoring the AI report. The multimodal type exists there too, and it is where image-started AI visibility shows up.
- Reporting clicks as revenue. Search Console stops at the click; join it with analytics before talking about sales.
What Are Platform Properties in Search Console?
Platform properties are a Search Console property type that shows how your Instagram, TikTok, X and YouTube content performs on Google Search. Google announced them on 7 July 2026 and made them globally available to everyone on 29 July 2026. Instead of a website, you verify a social account or channel, and Search Console reports the clicks, impressions and search terms that brought people to your posts from Google Search, Discover and Google News.
Google frames the purpose plainly: content creators and publishers use many channels beyond their own websites, and people increasingly look for firsthand perspectives and different content formats. In 2026 a brand's Instagram reel or YouTube tutorial can rank in Google results alongside its website, and until July there was no first-party way to see that traffic at all. Platform properties close that gap, including for creators who have no website.
| Attribute | What Google published in 2026 | What it means for you |
|---|---|---|
| Announced | 7 July 2026, following an earlier experiment | A full property type, not a test feature. |
| Global availability | 29 July 2026, "globally available to everyone" | Indian accounts can use it now. |
| Supported platforms | Instagram, TikTok, X, YouTube | Four only. LinkedIn and Facebook are not listed. |
| Surfaces reported | Google Search, Discover and Google News | Discover and News reports appear only if you receive traffic there. |
| Reports | Performance, Insights, Achievements | Detail, overview, and milestones. |
| Default date range | 28 days for Insights and Performance | Change it when comparing longer periods. |
| Data delay after setup | A few days for data to appear | Do not judge a new property in its first week. |
| Scope limit | Shows performance on Google Search only, not on the platform itself | It will not tell you how many TikTok views a video got. |
Why Do Social Accounts Need Search Data?
Because Google is now a meaningful discovery channel for social content, and the words people search with are the most honest brief a content team can get. A creator who knows that a specific search term suddenly started sending people to last week's video can make the next video, caption or hashtag around that term. Google's own guide makes exactly this point: search themes show what your audience cares about.
For brands in 2026 this changes three conversations. Social teams get evidence beyond platform likes. SEO teams get visibility into a channel that was previously invisible to them. And leadership gets a single place to see whether a YouTube channel or Instagram account is pulling people in from Google, which is often the question behind "should we keep investing in this account".
- Creators without websites get Search Console data for the first time.
- Brand social teams can see which posts rank in Google, not just which posts trend on the platform.
- SEO teams can include social content in search reporting instead of treating it as a separate world.
- Agencies can show clients a joined-up view of search visibility across site and social.
How Do You Set Up a Platform Property?
You add each account as its own property and verify ownership by authorising the connection. Google's help centre describes the steps: go to the Search Console welcome screen to add a property, or open the property selector dropdown in the sidebar and click Add property, then click Add next to the platform you want and follow the prompts. Once ownership is verified, you see a confirmation and click Go to property.
- Open Search Console.
- Use the welcome screen, or open the property selector dropdown and click Add property.
- Choose one of the four platforms: Instagram, TikTok, X or YouTube.
- Follow the onscreen verification steps to authorise the connection.
- After the confirmation message, click Go to property.
- Wait a few days for data to appear.
- Repeat for every account and every platform you want to track.
Google's analysis guide notes that if you already claimed your Search profile, all of your verified accounts are automatically added as properties in Search Console. Check the property selector before verifying anything manually.
One operational detail matters for agencies managing client accounts. Google says ownership is periodically checked for security, and if the connection is lost, for example because an external login expired, access to the platform property pauses until you re-verify. When you re-verify you get the same reports back without waiting for data to accumulate again. Put re-verification on a checklist, because a paused property looks like a traffic drop to anyone who does not know why.
What Do the Three Platform Property Reports Show?
The Performance report is the detailed view, Insights is the summary, and Achievements tracks milestones. Together they show clicks, impressions, average click-through rate and average position for the account's content in Google, the queries and posts driving that traffic, and where your audience is. Insights and Performance default to the last 28 days.
| Report | What Google says it shows in 2026 | Best use |
|---|---|---|
| Performance | Total clicks, impressions, average CTR and average search position; filter by post across Search, Discover and Google News; export | Detailed analysis and cross-platform comparison. |
| Insights | Recent traffic trends, top-performing content, and how people discover your account; query groups showing top, trending up and trending down themes | A weekly glance and content ideas. |
| Achievements | Growth milestones, such as reaching a new threshold for total clicks from Google Search in the last 28 days | Simple progress reporting. |
One nuance in the Insights page is worth knowing before you report numbers. Google's help centre says the top summary card shows all clicks to your property across Google, including web, image, video and news searches, while the detailed lists below focus on web search results. That is why the cards below can add up to less than the summary card. It is not a data error.
How Should Teams Analyse Social Content Performance?
Google published a full guide on analysing social and video content, and it is unusually practical. The techniques below come straight from that guide and turn platform properties from a passive dashboard into a content planning tool.
- Read query themes. In the Insights report, the query groups card shows top, trending up and trending down groups. Use the rising groups to plan the next video, caption or hashtag.
- Catch spikes with the 24-hour filter. In the Performance report, the 24 hours filter shows whether a new post is suddenly drawing search traffic, so you can cross-promote it while interest is live.
- Compare formats with comparison mode. Google suggests comparing YouTube URLs containing
/watchagainst/shorts/, and Instagram URLs containing/p/against/reels/. - Compare playlists with a page filter. Filter on the word playlist to compare playlist pages, noting this shows the playlist page itself, not the videos inside it.
- Compare across platforms by exporting. Export each property and combine them in one spreadsheet to see how the same content performs on YouTube versus TikTok.
- Annotate caption and title changes. Add a Search Console annotation on the date you rewrite a YouTube title or TikTok caption, then watch whether search performance changes.
- Revive older content. Look for older posts regaining traction and consider pinning them, mentioning them elsewhere or making a part two.
What Do Platform Properties Not Report?
They report Google, not the platform, and the help centre says so directly: platform properties only show how your content performs on Google Search, and they do not track when people see your content on the platform itself. For anyone tempted to replace platform analytics with Search Console, that boundary is the whole point.
- No on-platform views. It will not show how many times your video appeared on TikTok.
- Four platforms only. Instagram, TikTok, X and YouTube. Other networks are not listed.
- One property per account. Multiple profiles on one platform each need their own property.
- Recent setups show partial data. Charts start from when collection began.
- Clicks count even inside Google's viewer. Google notes a click is recorded even if it opens the video inside the Google viewer, so a click is not always a visit to the platform.
- No conversions. Like all Search Console data, it stops at the click.
What Are the Common Mistakes?
- Expecting on-platform metrics. Views and likes on the platform are not reported.
- Verifying only the main account. Every profile and every platform needs its own property.
- Missing the Search profile shortcut. Claimed Search profiles add verified accounts automatically.
- Panicking at a paused property. An expired login pauses access; re-verify and the history returns.
- Judging a property in week one. Data takes a few days to appear.
- Adding up Insights cards to match the summary. The summary includes image, video and news clicks; the cards focus on web.
- Changing captions without an annotation. You lose the before and after.
What Are the Common Mistakes in 2026?
- Treating impressions as clicks or citations. The report measures shown links, nothing more.
- Comparing Search and Discover AI numbers directly. They count and aggregate differently.
- Adding table rows to match the chart. Property versus page aggregation explains the gap.
- Forgetting Pacific Time. Daily data is in PT, which matters for Indian teams reading daily spikes.
- Missing an inherited exclusion. A parent property's control setting can silently exclude a child.
- Reading blanks as zeros. "~" and "-" values become zeros on export.
- Accepting AI visibility claims without this data. In 2026 there is no reason to.
Key Takeaways for 2026
AI search visibility is now measurable in first-party data, which raises the bar for anyone selling or buying it.
- Two reports: Search (AI Overviews and AI Mode) and Discover (generative AI features in Discover).
- Announced 3 June 2026 and rolled out to all websites worldwide as of 31 August 2026.
- Impressions by page, country, date, and device for Search, with hourly to monthly granularity.
- No AI answer text, triggering query, position or conversions are reported.
- The Search generative AI control under Settings decides whether you can appear at all.
- Search and Discover AI reports count impressions differently; compare like with like.
- Use the reports to test specific content changes against a control group.
Distk builds AI and search visibility reporting for growth teams in India and internationally from these first-party reports, and runs the content tests that move them. If you want proof rather than promises from AI visibility work, this data is where we start.
Sources
- Google Search Central Blog, Search Generative AI performance reports announcement, 3 June 2026, with the 31 August 2026 rollout note.
- Search Console Help, Generative AI performance report (Search).
- Search Console Help, Generative AI performance report (Discover).
- Search Console Help, Search generative AI control.
- Google Search Central Blog, Announcing web multimodal Search performance reporting in Search Console, 24 September 2026.
- Google Search Central Blog, platform properties announcement, 7 July 2026.
- Google Search Central Blog, platform properties globally available and analysis guide, 29 July 2026.
- Search Console Help, About platform properties in Search Console.
- Google Search Central, Analyze your social and video platform content performance in Search Console.