Search is no longer just text. People point their phone cameras at products, ask questions about images, search with screenshots, and watch video answers. AI systems increasingly understand images, video, and audio alongside text, combining them in multimodal answers. For businesses, this creates new opportunities to be discovered through visual content, and new risks if visual assets are poorly optimised.
This guide explains how multimodal AI search works and how to optimise images and video for visibility.
What Is Multimodal Search?
Multimodal search allows users to search using, and receive answers containing, multiple types of media: text, images, video, and audio. Examples include:
- Searching with a photo using visual search tools such as Google Lens
- Asking an AI assistant questions about an uploaded image
- Receiving AI answers that include images or video results
- Voice queries answered with spoken and visual responses
- Circling or highlighting part of a screen to search for it
Multimodal AI models can interpret visual content, recognise objects, read text in images, and connect visuals with textual information.
Why Multimodal SEO Matters
- Visual queries are growing: users increasingly search with cameras and screenshots
- AI answers include media: images and videos appear alongside generated text
- Products are discovered visually: shoppers identify items from photos
- Video answers how-to questions: users often prefer watching demonstrations
- Context improves understanding: well-described visuals help AI understand your pages
For the broader shift, see our guide to how AI search works.
Image Optimisation for AI Search
Use Original, High-Quality Images
Original photos of your products, team, premises, and work carry more value than generic stock photos. They demonstrate real experience and are unique to your brand.
Write Descriptive Alt Text
Alt text describes images for screen readers and helps machines understand content. Describe what the image shows and its relevance.
Weak: “image1” or “SEO”
Strong: “Dashboard showing AI referral traffic from ChatGPT and Perplexity over six months”
Use Descriptive File Names
Name files meaningfully before uploading, such as “stainless-steel-french-press-1-litre.jpg” rather than “IMG_4832.jpg.”
Add Captions Where Helpful
Captions are widely read and provide additional context close to the image.
Place Images Near Relevant Text
Surrounding text helps systems interpret images. Place images next to the content they illustrate.
Use Structured Data
Product, Article, Recipe, and other schema types include image properties. Make sure images referenced in schema are high quality and crawlable.
Optimise Performance
Compress images, use modern formats such as WebP or AVIF, set dimensions, and lazy-load below-the-fold images, but not the main hero image.
Keep Text Out of Images Where Possible
Important information should exist as HTML text. If images contain text, such as infographics, also provide the key points in text.
Video Optimisation for AI Search
Host Strategically
YouTube videos often appear in search and AI answers. Embedding videos on your site as well connects them with your content. Consider where your audience searches.
Provide Transcripts
Transcripts make spoken content accessible to text-based systems and users. Publish them on the page or provide accurate captions.
Use Chapters and Timestamps
Chapters help users and search systems jump to specific moments, which can appear as key moments in search results.
Write Descriptive Titles and Descriptions
Titles and descriptions should clearly state what the video covers, using the language people search with.
Add VideoObject Schema
VideoObject schema includes name, description, thumbnail URL, upload date, duration, and content URL. It helps search engines understand and display videos.
Create Custom Thumbnails
Clear, relevant thumbnails improve click-through and help convey content.
Match Video to Intent
How-to, demonstration, review, and explainer videos align well with common search and AI queries.
Multimodal Optimisation Checklist
| Asset | Key optimisations |
|---|---|
| Images | Original, descriptive alt text, file names, captions, compression, schema |
| Infographics | Text summaries of key points, descriptive alt text |
| Product photos | Multiple angles, consistent quality, Product schema |
| Videos | Transcripts, chapters, titles, descriptions, VideoObject schema |
| Audio and podcasts | Transcripts, show notes, episode pages |
Common Mistakes
- Using generic stock images for everything
- Missing or keyword-stuffed alt text
- Embedding videos without transcripts or descriptions
- Placing important information only inside images
- Uploading uncompressed, slow-loading media
- Blocking image or video files in robots.txt
An Example
Picture a furniture retailer selling a handmade oak dining table. Its product page has one small, compressed stock-style photo named “product-final.jpg” with no alt text. A shopper who photographs a similar table in a café and searches visually is unlikely to find it. By adding several original photos from different angles, descriptive file names and alt text mentioning the wood, size, and style, a short assembly video with a transcript, and Product schema with image properties, the retailer makes the table far easier for visual and AI search systems to recognise and surface.
Measuring Multimodal Performance
- Image and video impressions in Google Search Console, using search type filters
- YouTube analytics for search traffic sources
- Engagement on pages with video
- AI answer testing for queries where visuals appear
For wider measurement, see our guide to AI search visibility.
Final Thoughts
Multimodal SEO ensures your images and videos work as hard as your text. Use original visuals, describe them clearly, provide transcripts and structured data, and keep performance fast. As AI search becomes more visual, well-optimised media will help your brand appear in more places and more formats. For a complete review, see our AI SEO checklist and consider an AI SEO audit.
Frequently Asked Questions
What is multimodal SEO?
Multimodal SEO is optimising text, images, video, and audio so search engines and AI systems can understand and surface them in visual searches, multimedia results, and AI-generated answers.
How does alt text help AI search?
Alt text describes image content in words, helping search engines and AI systems understand what an image shows and how it relates to the page, while also improving accessibility.
Should I use stock photos or original images?
Original images are better where possible. They demonstrate real experience, are unique to your brand, and provide more meaningful visual information than generic stock photos used across many sites.
Do videos need transcripts for SEO?
Transcripts make spoken content readable by text-based systems and accessible to users. They help search engines and AI understand video content and can surface relevant information in answers.
What is VideoObject schema?
VideoObject schema is structured data describing a video, including title, description, thumbnail, upload date, duration, and content URL, helping search engines understand and display your video content.
How do I measure image and video search performance?
Use Google Search Console with image and video search type filters, YouTube analytics for search sources, engagement metrics on video pages, and AI answer testing for visual queries.
