Direct answer: open Web: multimodal in Search Console to find the landing pages Google surfaces for image-led searches. Compare those pages with text-based search, spot products and guides gaining visual discovery, and investigate where impressions fail to turn into clicks. Use the report to choose a page to improve, not just another metric to watch.
What you can actually learn from multimodal data
- Which pages attract visual discovery: find the products, categories, and guides appearing in image-led searches.
- Whether visual search finds different destinations: compare the landing-page mix with text-based search over the same period.
- Which pages deserve a closer look: identify unexpected impressions on a model-identification guide, a product variant, or a category page.
- Where visibility stops short of a click: use impressions and CTR to select pages for a relevance review.
Your first task: pick one important page with multimodal impressions, inspect the visitor problem it solves, and write down one improvement to test.
What changed in September 2026
Google announced multimodal reporting on September 24, 2026. It covers searches initiated through Lens, Circle to Search on Android, uploaded images, and Chrome image search. The announcement includes Search results and Generative AI performance reports, with a global rollout starting that day. Metrics appear when a property receives traffic from these queries.
The practical opportunity is a clearer view of image-led discovery. A page that looks ordinary in your text-search review may be an important destination for people searching with a picture.
Multimodal SEO is not the same as image SEO
Web: multimodal describes image-led web searches; Image describes traffic from the Google Images tab. Google's Performance report help page lists them as separate search types. Use that distinction to keep your analysis focused.
| View | What it measures | Useful question |
|---|---|---|
| Web: text-based | Web searches initiated with text | Which pages do people find by typing a query? |
| Web: multimodal | Image-led web searches | Which pages appear when the search starts with a picture? |
| Image | Traffic from the Google Images tab | Which pages receive visits through image results? |
Multimodal SEO is a useful shorthand for serving those visual-search tasks, not an official Google ranking system or a separate optimization switch. Technical image SEO still supports discoverability. The additional question is whether the receiving page helps someone identify, compare, understand, or buy what they saw.
Keep the technical work in our image SEO implementation guide. For broader Google/Bing AI reporting, use the AI performance guide.
How to analyze the report in one focused session
- Choose your property and period. Start with a complete recent window, such as the previous 28 days. Record the date range and active filters.
- Select Web: multimodal. Open the Performance report and use the search-type filter. Keep Search results and Generative AI report exports separate.
- Review Pages. Sort by impressions to find visual exposure, then inspect clicks and CTR. Shortlist commercially important pages and surprising destinations.
- Compare with text-based search. Use the same property and dates. Export separate views if that makes the comparison clearer. Look for differences in the page mix, not just the largest traffic total.
- Read the shortlisted pages on mobile. Check whether the visible product, model information, explanation, and next action match the job a visual-search visitor needs done.
- Choose one action per page. Add the hypothesis, planned change, and next review date to the template below.
Do not begin by editing every image. First find a page where the report and the visitor experience point to a useful change. If you have very little data, extend the observation window before turning small differences into priorities.
Example: three chair pages, three different decisions
Fictional example: all URLs and figures below are invented to explain the analysis. They are not FloxoLab results or a Search Console export.
| Page | Multimodal clicks | Impressions | CTR | What to inspect |
|---|---|---|---|---|
/products/red-chair | 42 | 1,300 | 3.2% | Product image and model details |
/guides/chair-models | 18 | 900 | 2.0% | Identification intent |
/category/chairs | 2 | 1,100 | 0.2% | Possible intent mismatch |
The category page is worth inspecting because it has visual exposure but few clicks. That does not tell you to replace its image. A visitor trying to identify one chair may need a specific model page, while the category offers a broad collection. Check the page's promise and the available search context before choosing the fix.
The guide suggests another opportunity: help readers distinguish similar models, then link to relevant products. The product page deserves a review of model name, dimensions, material, color, and availability so a visitor can confirm the match.
These CTR values are not benchmarks. Compare like-for-like periods and page purposes; use the numbers to select an inspection, not to declare a winner.
What to change on the landing page
- Product pages: make the pictured variant easy to confirm. Use precise model and color information, useful dimensions, and a clear route to purchase or inquiry.
- Category pages: help visitors narrow the collection. Make product names and differences understandable; link to the exact products a visitor might be trying to identify.
- Identification and how-to guides: answer the visual question early. Show distinguishing features or the relevant step, then guide the reader to the next task.
- Pages receiving clicks: check the mobile experience and the next action. Finding the right page should lead to a useful answer, not a search for hidden details.
If the image itself is inaccessible, misleading, or poorly implemented, fix it. Otherwise, start with the missing answer on the page. Record one meaningful change so the next review has a clear question to answer.
Three conclusions the report does not support
- A dedicated Lens/Circle split: the combined multimodal view does not by itself establish a breakdown for each input surface.
- Business outcomes: search clicks do not establish leads or sales. Check those in your analytics and business records.
- Causation: higher clicks after an edit do not isolate its effect from seasonality, ranking changes, or a changed offer.
Keep the same report, dates, and filters when comparing results. Google's help page also explains differences in chart/table aggregation. For tagged arriving visits, use our separate Gemini referral tracking guide.
Download the multimodal SEO review template
Open the CSV in Excel or import it into Google Sheets. It contains column headings and no fabricated account data. Add one row per page, report, and date range.
Columns: Date range, Property, Report, Search type, Page, Clicks, Impressions, CTR, Position, Page purpose, Hypothesis, Change, Review date.
Copy the reported values into the first fields. In Page purpose, state the job the visitor should complete. In Hypothesis, explain the possible gap. In Change, record the edit you actually make. Set a Review date and use a comparable reporting window for the follow-up.
Leave unreported values blank. Keep the original export alongside this working sheet so facts and interpretation remain easy to separate.
Can Search Console show Google Lens traffic?
Yes. Google announced image-led web-search reporting in Search Console on September 24, 2026, under Web: multimodal. Covered inputs include Google Lens, Circle to Search on Android, uploaded images, and Chrome image search. This is separate from the Image search type for Google Images traffic. The multimodal view combines these inputs; it does not automatically give a Lens-only total.
What should an SEO specialist do first?
Open the multimodal Pages view, shortlist important pages with impressions, and compare their purpose with the task a visual-search visitor would need to complete. Pick one gap to fix and record the baseline before editing.
Does low multimodal CTR mean the image is bad?
No. It selects a page for investigation. The page may serve a different task, present an unclear product match, or need a better explanation. Inspect the destination and context before deciding to change an image.
Turn visual discovery into a useful landing page. FloxoLab connects reporting, technical image SEO, and page improvements into one practical action plan. Explore the technical SEO audit.
Sources and verification
Google's announcement and Performance help page linked above were checked on October 10, 2026. The fictional table and CSV review framework are FloxoLab teaching tools; they contain no client measurements. The cover is conceptual artwork.