SEO is becoming multimodal because the query is becoming multimodal. People can now search with a camera, a screenshot, an uploaded image, text, voice or combinations of them. That means visual content is no longer decoration around the searchable page. It can become part of the evidence.
Today Google added multimodal search reporting to Search Console.
That sounds like a nerdy analytics update.
It is actually a clue about where search is going.
The query is no longer only a string of words.
The query is becoming an object.
For decades, almost everything in SEO started with typed language.
A person searched “best trail shoes Arizona.”
Google returned pages.
We built an entire industry around matching words to documents.
Now the query can be a shoe in somebody’s hand.
It can be a house they drove past. A plant. A product label. A screenshot. A room. A chart. A paragraph inside a PDF. A camera pointed at something they do not know how to describe.
Then the person can ask a follow-up in plain language.
That is not a small interface update. It changes what counts as searchable information.
Google is finally exposing some of it.
Google’s new Search Console filter includes searches from Lens, Circle to Search, image uploads to Google Search and Chrome’s “Search this image” feature.
The data is being added to both standard Search performance reporting and Google’s newer generative AI reporting.
For the first time, site owners can start separating some of this behavior instead of watching it disappear into one giant search bucket.
Measurement usually arrives after behavior.
This is one of those moments.
Most websites still treat images like decoration.
Hero image. Stock photo. Gallery. Done.
That is not enough in a world where the machine can look at the image as part of the query and part of the answer.
If you sell a physical product, show the real product clearly.
If you are a real-estate professional, original neighborhood photography can carry location evidence stock photography never will.
If you publish research, charts should have meaningful captions and surrounding explanation.
If you are documenting a system, diagrams should actually explain the system.
The visual layer should contain information.
No, this does not mean stuffing alt text.
This is where the SEO industry usually manages to turn something interesting into a checklist.
Somebody will call this “visual GEO” and tell you to jam keywords into every image description.
Please don’t.
Alt text exists primarily for accessibility. Keep it useful and honest.
The bigger opportunity is much more human: create original visuals that are specific enough to teach both the person and the machine something real.
Search is becoming a perception layer.
Google’s 2026 Search box can work across text, images, files, video and browser tabs. Lens and Circle to Search already let people start with what they see instead of what they can name.
That moves search closer to perception.
The system is trying to understand the world the user is pointing at, not simply the sentence they typed.
That is a massive expansion of intent.
What changes for SEO, AEO and GEO?
For SEO, image and media quality matter because visual discovery can now create measurable search exposure.
For AEO, text still needs to explain what the visual means so the system can build a useful answer around it.
For GEO, the image, page, author, entity and surrounding evidence should reinforce the same reality instead of contradicting one another.
Again: same problem, three lenses.
The websites that win will be easier to understand in more than one format.
Text and links still matter. The difference is that machines can now read far more than the words around them.
They can parse the copy. Inspect the image. Read the chart. understand the entity. Compare the source. Follow the citation.
So I would stop thinking about content as “blog posts.”
Build evidence in multiple formats.
Make every format say something worth understanding.
Google Search Central: Web multimodal Search performance reporting