How to Optimise Images for Visual Search & AI & AI Overviews

Visual search enables users to find information using images instead of text, relying on tools such as Google Lens, Pinterest Lens, and Bing Visual Search, with billions of searches conducted through these platforms every month. According to Google, there are over 20 billion searches per month when we talk about Google Lens (Source 1).

As AI technology advances, optimising your images carries more weight than ever; it’s not just about SEO, but also about standing out visually and delivering a better, more engaging user experience that meets rising expectations for instant, visual-driven content.

What is Visual Search?

Visual search is a technology that allows users to perform searches using images instead of words. By uploading a picture or using a camera, users can discover related content, products, or information. Tools like Google Lens, Pinterest Lens, and Bing Visual Search identify elements in an image, such as objects, text, colors, or shapes, and return relevant search results based on what they see.

For example, a user can take a photo of a houseplant and immediately find care tips, similar-looking species, or online retailers that sell it. This search method has become especially popular in e-commerce, travel, and lifestyle content, where visuals play a dominant role in user decision-making.

How to Optimise Images for SEO & AI

When preparing images for your website or platform, focus on a few key areas:

Best File Formats

Select the best image formats for the web, such as WebP or AVIF. These modern formats offer smaller file sizes with high visual quality, which speeds up your site. JPEG is suitable for high-quality photos, while PNG is best suited for graphics that require transparency.

Ideal image sizes and compression tips

Avoid using unnecessarily large images. Resize them for their display context and compress using tools like TinyPNG, Squoosh, or ImageOptim. Always specify the width and height in HTML to prevent layout shifts.

Best File Names

Use descriptive file names. Rename images with meaningful titles, such as delante-logo-in-office.png instead of a generic IMG_1234.png.

Alt Text

Write clear and concise alt text that accurately describes the essential content of the image. Alt text supports accessibility (required in the EU by the European Accessibility Act) and helps search engines understand the content of the image. For comprehensive tips and a checklist on writing great alt tags, see the dedicated section in our How to Improve Both SEO & Website Accessibility on Your Site article.

HTML

Add proper HTML attributes, such as alt, title, and srcset, to enrich both user and AI understanding of your visuals.

Implement structured data, such as the ImageObject schema, which enhances how search engines and AI models interpret and display your images in search results.

Utilise lazy loading to enhance your site’s loading speed. Ensure the images remain discoverable by search engines after they are loaded.

What Does AI “See” and Prioritise in Visual Search?

AI analyses images on multiple levels: it recognises colours, shapes, text embedded within images, and the overall composition. Make main subjects prominent and avoid cluttered backgrounds to improve recognition. Uniqueness matters: original, well-composed visuals consistently outperform generic stock photos, both in terms of SEO and branding.

For e-commerce and product pages, display products from multiple angles and in various scenarios, including classic packshots, lifestyle images, and 360° formats where possible. This variety enhances both the visual appeal and the comprehensiveness with which AI and users understand your offering.

It’s also worth considering embedded metadata, such as EXIF or IPTC, which can carry camera, location, or copyright details. While the direct SEO effect is debated, this data can help with image protection and potentially with local relevance in some contexts. It is always a wise idea to leave a footprint on your content.

How Do Images Affect AI Overviews?

AI-generated overviews (for instance, Google’s AI snippets) draw on both the image and the related text. This is called multimodal analysis.

Place clear captions and descriptive text near images, both AI models and users rely on this context to interpret imagery correctly.

When preparing content, always provide a few sentences of context or a caption directly beside the image to maximise its use in AI-driven summaries and enhance accessibility.

Are AI-Generated Images OK for SEO?

AI-generated images can provide unique, never-before-seen visuals, which is a significant plus. Still, indiscriminate use or relying on bland, easily recognisable AI-generated stock can reduce brand coherence and, in the future, may be downgraded by search algorithms. 

For best results, use AI-generated images with intention, ensure they align with your brand’s style, are relevant, and genuinely add value beyond what’s available in stock libraries.

I’d say that the use of AI-generated images depends mainly on the purpose, context, and the value they provide to users. For e-commerce, it’s especially important to follow Google Merchant Center guidelines on AI-generated content. In this context, AI-generated images may be acceptable, but only if they accurately represent the product, aren’t misleading, and meet standards outlined by Google, such as proper metadata.

Outside of e-commerce, the situation is more flexible, but the same core SEO principle applies: the image must be helpful, relevant, and enhance the user experience. The industry also plays a big role. For example, the travel sector typically performs better with authentic, real-world photos that build trust and emotional connection. In contrast, in tech or conceptual services, high-quality AI-generated visuals can be perfectly acceptable and even effective.

In short, it’s not about whether an image is AI-generated or not; it’s about whether it serves the user and aligns with platform-specific standards and requirements.

Be informed that Google is taking steps to reduce the flood of AI-generated content on its services. Recently, a July 2025 update was pushed to YouTube’s channel monetisation policies to lower the visibility and revenue of low-effort, mass-produced, AI-generated content. This might happen on a broader scale. But, as long as you put your human brain and heart into the content you create, you should be fine.

Common Mistakes to Avoid in Image Optimisation

Instead of focusing just on tools, be mindful of frequent mistakes that can weaken your visual SEO efforts:

• Using non-descriptive file names or omitting alt text can hurt accessibility and discoverability.
• Overcompressing images results in visible artefacts or loss of detail.
• Relying on generic stock imagery makes your site less distinctive to users and AI.
• Ignoring the surrounding context or omitting captions results in images losing relevance in AI-generated summaries.
• Failing to consider accessibility requirements (such as the European Accessibility Act) for global audiences.
• Neglecting structured data or using incorrect HTML causes search engines to misinterpret images.

In Summary

Image optimisation has evolved from being a simple technical requirement to a core strategy in both traditional SEO and new AI-driven search. Invest in unique, brand-aligned visuals, use effective metadata and accessibility practices, and reinforce images with nearby context. By doing so, you’re setting up your content for stronger search visibility, better user engagement, and more relevance in today’s visually focused digital landscape.

Need help with optimising your website’s visuals for better performance in search and AI results? Get in touch with us. At Delante, we combine technical SEO expertise with a creative approach to help your brand get noticed where it matters most.

Sources
1. https://blog.google/products/ads-commerce/google-lens-ai-overviews-ads-marketers/