AI-era search · The 2026 shift series

Multimodal Search: Customers Now Search with Photos and Videos

By Manas Panda & the SynapseIN teamPublished: July 31, 2026

A woman photographing a flowering plant with her phone in a garden — Scene illustration, AI-generated. It depicts no real person, client, place or event. Why we label images.

In one line: Users increasingly search with images and video - the damage, the rash, the part. Visual content carrying ImageObject and VideoObject schema with honest descriptions gets matched; unlabeled images are invisible.

A note on the stories here: scenarios in this article are composites and teaching illustrations drawn from patterns we see repeatedly — not named client engagements. The technical guidance is real; the characters are devices.

Multimodal Search: Your Customers Are Now Searching with Photos and Videos

A homeowner in San Jose took a photo of a wet spot on their ceiling. They uploaded it to Google's search box and typed: "What is this and how much will it cost to fix?"

Google's AI analyzed the image. It identified the pattern as a likely roof leak. It searched for roofing contractors in San Jose who have content about roof leak repair. It found a contractor with Service schema for "Roof Leak Repair," HowTo schema for "How to Fix a Roof Leak," and before/after photos with ImageObject schema and descriptive alt text.

The AI generated an answer: "This appears to be a roof leak caused by damaged flashing. In the Bay Area, roof leak repair costs $400-$1,200 depending on severity. [Contractor Name] in San Jose specializes in emergency roof leak repair and offers same-day assessment."

A phone camera over a vintage machine part with a glowing result card above it — Scene illustration, AI-generated. It depicts no real person, client, place or event. Why we label images.

The homeowner called the contractor. A $2,500 repair job. From a photo.

This is multimodal search. And it's the biggest change to how people find businesses since mobile search.

What Is Multimodal Search?

At Google I/O 2026, Google announced that the search box now accepts images, videos, files, and Chrome tabs — not just text. Users can search by uploading a photo, a video, or a document and asking a question about it.

This is multimodal search. The AI processes multiple types of input simultaneously — text, images, video, documents — and provides answers that incorporate all of them.

For businesses, this means:

* A patient can upload a photo of their skin condition and ask "who should I see near me?" Google's AI will match the image to dermatology or plastic surgery content, then find clinics with relevant schema and visual content.

* A homeowner can upload a photo of their roof and ask "how much will this cost to fix?" Google's AI will match the image to roofing content, then find contractors with relevant schema and visual content.

* A business owner can upload a cybersecurity audit report and ask "what should I do next?" Google's AI will match the document to cybersecurity services, then find MSPs with relevant schema and content.

What Multimodal Search Requires

1. ImageObject Schema on Every Image

Every image on your website needs ImageObject schema. This tells Google's AI what the image contains.

Without ImageObject schema, your before/after photos, procedure diagrams, and installation photos are invisible to multimodal search. The AI can't match a user's uploaded photo to your content if it doesn't know what your images contain.

2. Descriptive Alt Text

Every image needs descriptive alt text. Not "IMG_9274.jpg" — but "Rhinoplasty before photo, 34-year-old female, Beverly Hills clinic, pre-operative profile view."

The alt text is what Google's AI reads when it's trying to match a user's uploaded image to your content. The more specific and descriptive the alt text, the better the match.

3. Descriptive Filenames

Your image filenames should be descriptive. "rhinoplasty-before-photo-beverly-hills.webp" tells Google's AI what the image contains. "IMG_9274.jpg" tells it nothing.

4. VideoObject Schema on Every Video

If you have procedure videos, installation walkthroughs, or consultation recordings, every video needs VideoObject schema. This tells Google's AI what the video contains, how long it is, and what it's about.

5. Before/After Photos with Context

For medical clinics, dental clinics, and cosmetic surgery centers, before/after photos are your most valuable multimodal asset. But they need:

* ImageObject schema with description

* Descriptive alt text: "Patient before and after dental implant surgery, 52-year-old male, Singapore, 6-month post-operative result"

* Descriptive filenames

* Context on the page: what procedure, what timeline, what results

What This Looks Like in Your Industry

Cosmetic Surgery Center, Beverly Hills

A patient uploads a photo of their nose and asks "what would rhinoplasty do for me?" Google's AI searches for plastic surgery centers with before/after rhinoplasty photos, MedicalProcedure schema, and descriptive alt text. Your "IMG_9274.jpg" with no alt text? Invisible. Your competitor's "rhinoplasty-before-after-beverly-hills.webp" with ImageObject schema? Featured.

Solar Contractor, San Jose

A homeowner uploads a photo of their roof and asks "can I install solar on this?" Google's AI searches for solar contractors with installation photos, HowTo schema, and Product schema. Your photos without schema? Invisible. Your competitor's photos with ImageObject schema and descriptive alt text? Featured.

Dental Clinic, Singapore

A patient uploads a photo of their smile and asks "what would dental implants look like for me?" Google's AI searches for dental clinics with before/after implant photos, MedicalProcedure schema, and descriptive alt text. Your photos without schema? Invisible. Your competitor's structured visual content? Featured.

The Multimodal Checklist

* [ ] ImageObject schema on every image

* [ ] Descriptive alt text on every image (procedure, patient type, location, result)

* [ ] Descriptive filenames (no "IMG_XXXX.jpg")

* [ ] VideoObject schema on every video

* [ ] Before/after photos with context and schema

* [ ] Process photos with HowTo schema

* [ ] Product photos with Product schema

* [ ] Consistent image schema across all pages

Multimodal search is the next frontier. The businesses that optimize their visual content with schema, alt text, and descriptive filenames will be the ones Google's AI matches to user photos and videos. The ones that don't will be invisible to the fastest-growing search modality in 2026.

Is your visual content optimized for multimodal search? Send me your URL. I'll audit your images, videos, schema, and alt text — and show you exactly what multimodal search can and can't see. No follow-up. No sequence. Just the visual audit.

— Manas

SynapseIN.com

#MultimodalSearch #AEO #ImageSEO #TechnicalSEO #SchemaMarkup #DigitalMarketing #VisualSearch

________________

Our honesty standard: statistics here are attributed to published studies where exact, or stated as directional patterns where studies disagree — we do not print precise numbers we cannot source. Google-event claims link to coverage. Our own claims carry evidence here.

Run the 5-check AI Presence Snapshot See your site through machine eyes