Search behavior is becoming much harder to describe with the phrase “type a keyword into Google.”
People can now search with text, photographs, screenshots, uploaded images, video, files, and conversational follow ups. Google’s latest Search Console update makes that change measurable for the first time.
On September 24, Google introduced dedicated reporting for web multimodal search, covering traffic from Google Lens, Circle to Search, uploaded images, and Chrome’s image search feature. Site owners can now isolate this activity inside Search Console performance reports. Google for Developers
For an ai seo agency, this creates a broader SEO question. A page can no longer be evaluated only on how well its text matches a keyword. Images, products, entities, structured information, original evidence, and the relationship between visual and written content increasingly contribute to how a page can be found.
Search is becoming multimodal by default
Google’s current AI Search experience already accepts more than typed text.
At Google I/O 2026, the company described a redesigned Search box capable of reasoning across text, images, files, video, and browser tabs. AI Mode has also passed one billion monthly users. blog.google
That means a search journey might begin with:
- a photograph of a product
- a screenshot of software
- an image of a room
- a product label
- a diagram
- a visual comparison
The user may then ask a conversational question about what they are seeing.
SEO therefore needs to account for the information contained inside the page as a whole.
Images need to carry meaning
For years, image optimization was often reduced to: - file compression
- descriptive filenames
- alt text
- image sitemaps
Those practices still matter.
The bigger opportunity is making images genuinely useful.
Consider a page about enterprise CRM implementation.
A generic stock photograph of people in a meeting contributes very little.
A visual showing:
Lead capture → qualification → opportunity → proposal → customer onboarding
contains information that supports the subject of the page.
The same applies to: - product comparison graphics
- annotated screenshots
- diagrams
- original charts
- process visuals
- before and after examples
- technical architecture illustrations
Google’s new reporting means teams can begin seeing whether those visual assets contribute to actual search visibility rather than treating image SEO as an isolated activity. Google for Developers
AI Search creates another visibility surface
Google’s generative AI Search reporting is also now available worldwide.
Search Console can show impressions for pages appearing across AI experiences, including information about the pages being surfaced and the countries where those appearances occur. blog.google
This creates several different visibility questions:
Traditional Search:
Does the page rank?
AI Search:
Does the page contribute useful information to generated answers?
Multimodal Search:
Can visual content help Google identify and retrieve the page?
These are related, yet they should be measured separately.
A single ranking report cannot explain all three.
Original information becomes more valuable
Google has also spent 2026 increasing visibility for original reporting and first hand sources.
Its Search updates include Preferred Sources in AI Search and broader use of “Highly Cited” labels for influential original coverage. blog.google
That should influence SEO content planning.
A page containing information available on hundreds of other sites gives a search system few reasons to prefer that particular source.
Stronger content includes: - proprietary data
- expert observations
- original screenshots
- first hand examples
- implementation experience
- product testing
- unique visual explanations
- research
- real customer questions
The same principle applies to visual content.
An original diagram explaining a difficult process has more informational value than another generic illustration generated only to fill space.
AI visibility cannot be reduced to one score
A growing number of SEO tools now provide AI visibility scores.
Interest is growing quickly. Ahrefs data cited by Search Engine Journal showed US searches for “AI search tracking” up 184 percent year over year, while searches for “AI rank tracking” increased 175 percent. Search Engine Journal
These metrics are useful when interpreted carefully.
A brand mention, a source citation, a linked citation, an AI impression, and a website visit are different events.
A stronger measurement model separates: - AI mentions
- source citations
- linked citations
- AI Search impressions
- multimodal impressions
- organic visits
- branded searches
- leads
- conversions
That makes the data actionable.
Technical SEO still matters
AI Search has created plenty of new terminology.
The underlying website still needs to be technically sound.
A recent audit of 50 major websites found that many brands had improved basic AI crawler accessibility while still leaving significant gaps in the signals machines need to interpret their content accurately. Search Engine Journal
Teams should still care about: - crawlability
- internal linking
- semantic HTML
- structured data
- canonicalization
- page speed
- accessible images
- clear entities
- descriptive headings
- indexation
AI Search does not eliminate technical SEO.
It gives technical SEO more places to matter.
The SEO audit is becoming broader
A useful search audit in late 2026 should now ask:
- Can Google crawl and index the page correctly?
- Does the page satisfy conventional search intent?
- Does it contain information worth citing in an AI response?
- Are its images informative and understandable?
- Could the page surface through visual search?
- Does it provide original evidence or perspective?
- Can AI systems understand the important entities and relationships?
- Does visibility eventually produce useful business outcomes?
This is a much more complete definition of SEO performance.
Search is becoming conversational, visual, and multimodal.
The websites gaining the most from that shift will likely be those that provide information in forms humans and machines can both understand.