How to optimize your visual content for AI SEO

Optimizing your visual content for AI SEO helps your brand get discovered and recommended by AI systems, like LLMs, agents, and AI search your potential customers might already be using to learn about you.

Today’s internet looks different than it did a few years ago. Now, your digital content isn’t only consumed by humans — your audience has expanded to include search engines, algorithms, and AI.

AI search is rapidly growing and projected to overtake traditional search. People are using AI to search for and learn about your company, so if you’re not taking steps to optimize your content, you’re missing out on potential customers and a larger footprint.

In this post, we’ll cover what AI SEO is, some foundational best practices for optimizing and implementing your visual content, and a couple examples to put it into practice.

Check out the video below for the full walkthrough on AI SEO for visual content.

What is AI SEO?

There are lots of similar terms floating around that mean essentially the same thing: 

• AEO = answer engine optimization
• GEO = generative engine optimization
• AIO = AI optimization
• LLMO = large language model optimization

The industry hasn’t aligned yet around one term or another. In this post, we’ll use the terms “AI SEO” and “AI optimization” interchangeably.

How does AI SEO differ from traditional SEO?

AI SEO, as the term suggests, has a lot of overlap with traditional SEO, but there are some key differences. Think of traditional SEO as the foundation for good web content and AI SEO as the new means of discovery. Traditional SEO best practices like E-E-A-T signals that build credibility and authority are still super important.

There’s also a lot of overlap between AI SEO and accessibility best practices. That’s because screen readers and AI use similar signals to interpret your content. Check out our video on accessibility tips for more info, because optimizing your content for accessibility will also give you a leg up with AI.

Why does AI SEO matter for visual content?

So why does AI SEO matter for visual content? Think of it this way: Both the people and AI that consume your content love visuals.

Grab and engage human users

For people, visuals grab users’ attention, keep them on the page for longer, and build your brand’s trust and authority. Those qualities that make your site more engaging for people also lead to better SEO.

Provide original content for AI to recommend and cite

For AI, visuals are highly recommendable in AI search. Original content like custom visuals or primary research is like catnip for AI search — you want to be the one who owns the original content, not just a broadcaster of secondary information. Visuals also increasingly appear in multimodal search — meaning queries or responses that incorporate a combination of text, images, video, or other modes of content.

The types of visuals AI tends to recommend are meaningful, content-rich visuals like charts, infographics, interactives, diagrams — essentially, all the cool stuff we create here at Lemonly — not just decorative fluff that fills space. So investing in visual content can be a great way to boost your AI SEO strategy.

Start by checking your website’s technical foundation

Before we get into AI SEO best practices for your visual content, we highly recommend beefing up your website’s technical foundation. If your website isn’t already set up for success with traditional or AI SEO, there’s only so much that optimizing your content can do.

Checklist: AI SEO technical foundation

Here’s a checklist we recommend for auditing your website’s technical foundation for AI SEO. These steps apply to your website as a whole — making sure that AI systems can access and crawl your website to learn about who you are.

• robots.txt - A basic allowlist that tells various web crawlers, including those for AI training and search, whether they can access your site.
• LLMs.txt - An AI-targeted summary of your website designed to explain what your website is, how to navigate it, and important context to know.
• Sitemap - Both the overall strategy around how content is organized on your site and the actual XML sitemap that documents it.
• Schema markup - JSON-LD structured data used to describe your site/content, build a knowledge graph, and create rich search snippets.
• Markdown - Plain-text versions of your web pages that can be automatically served as a more token-efficient option to AI crawlers.
• RSS feed - Helpful for signaling when fresh content is published on your site and ready to be indexed/crawled.

These technical foundation steps apply to your website as a whole, so it's worth investing some effort here before tackling specific content optimizations. Start by making sure your house is structurally sound, then pick out the furniture. In the rest of this post, we’ll talk about specific best practices for optimizing the visual content on individual pages of your site.

3 key principles for visual content AI SEO

There are three key principles to keep in mind for optimizing your visual content for AI:

1. Think with an answer mindset
2. Make content machine-consumable
3. Build signals for relevance and authority

Think with an answer mindset

First, think with an answer mindset. This has to do with how you strategize and structure your content. 

An “answer mindset” means framing your content around the user questions it answers. AI search programs — sometimes called “answer engines” — commonly provide answers to a user’s questions by pulling concise, citation-ready snippets from across the web. You want your content to be cited in these answers.

Applying an answer mindset to strategizing and structuring your content might look like using real FAQs as headings (you might notice a few of those in this post) or laying out the content in chunks that are easy to skim and ready to be cited by AI.

Make content machine-consumable

Second, make content machine-consumable. This applies to how you build and publish your content. 

Basically it means providing content in machine-friendly formats, often using live text or code in place of or along with certain types of visuals. AI is really good at reading text and code, but not as great with images or videos that are more resource-intensive to process. So essentially we need to translate certain types of visuals into a language that AI can understand more easily.

A good general rule is to make sure there’s more than one way to consume any visual content on a page — especially using live text, code, or both to supplement images and video.

Build signals for relevance and authority

Third, build signals for relevance and authority. This applies in two ways:

The first is about the metadata and surrounding content that goes with your visual. It means providing multiple signals that reinforce the meaning and topical relevance of the content. It might feel repetitive, but it’s actually strategic reinforcement. It’s kind of like sending a series of signals that say: 

1. “There’s a thing on this page.”
2. “Here’s that thing — this is its type, format, and meaning.”
3. “Here’s the exact content of that thing.”

Remember: Those signals can come in multiple formats and don’t just come from the visual itself, but all of the stuff around the visual that describes and supports it. The stronger your metadata, the better your visibility.

The second aspect of authority signals is about how your content is shared. Visuals are prime content for sharing and backlinks. Getting mentioned on other channels and websites makes your content more likely to get picked up by AI — it signals that you’re a trusted authority on this topic. 

You can do everything right on your own site, but it’s just as important to make sure other people are talking about you, too. Visuals are a great way to get people talking about your brand.

5 practical tips to optimize your visual content for AI search

Let’s dive into five tips for optimizing your visual content publishing for AI SEO. Not all of these will apply to every visual, but the more tactics you can stack that fit your content, the better off you’ll be.

1. Filename

First is the filename. Make it something meaningful, readable, and keyword-aligned. Not a random string or an internal naming convention. AI uses filenames to help understand what an image is, kind of like the title on the spine of a book on your bookshelf.

2. Alt text

Second is alt text. Alt text is primarily an accessibility feature that’s invisible to human users most of the time, but it also acts as a key signal for AI about what an image is. 

For every non-decorative image, include a concise description of the meaning or purpose of the image, not just its contents — describe the purpose, not the pixels. Keep it to about the length of one sentence. And remember, you never have to include “Image of” in your alt text — that’s implied since alt text is inherently an image attribute.

3. Captions

Third is captions, which, depending on the content type, could be either an image caption or video captions.

Image captions

Image captions are often shown as small text or an expandable toggle beside an image on a web page. Alt text and image captions have similar functions, but whereas alt text is invisible and read by machines like screen readers and AI, captions are visible and can be read by humans. Use a caption to describe the image and connect it to the larger narrative of your content. 

Captions can also be longer than alt text if needed. For example, you might use a caption to describe a chart in more detail, with 40–80 words about the main insight or takeaways.

Note that captions are optional depending on the image type. They’re generally best for smaller images that capture a single idea — not, for instance, a full infographic that covers multiple ideas. For classic vertical infographics, skip the caption in favor of the other methods we’ll cover next.

Video captions

Video captions are different from image captions — usually these are closed captions that can be toggled in the video player. We always recommend including captions (usually in the form of an SRT file) with videos for overall user experience, accessibility, and AI SEO.

4. Content sandwich

Now we’re getting to the more complex steps. Fourth is what we call a “content sandwich” — basically the surrounding content before and after your visual. 

The content sandwich has three parts:

1. Intro content that sets up the visual
2. The visual itself (often an infographic or video, but could be other types of visuals)
3. Transcript of any written or spoken content in the visual (usually nested in an accordion)

Intro content

Your intro content should have at least a heading and 1–2 paragraphs of text, but could be longer depending on your page layout and content strategy. The intro content should set up the visual, summarize what it entails, and establish any other important context.

We generally prefer to publish visuals on pages that are as content-rich as possible — not just on a page all by itself. A content-rich approach might look like putting an infographic in-line in a blog post or another web page where there’s lots of surrounding content that builds credibility, reinforces topical relevance, and provides a comprehensive answer to a user question or keyword. But if you are publishing a visual on a standalone page, the content sandwich and intro content matter even more.

Visual

Next comes the visual itself, often an infographic or video. 

Many static infographics are JPGs or SVGs with flattened text; we prefer SVGs whenever possible for their visual quality and scalability. For some static infographics and most animated infographics, Lemonly uses a custom tool that essentially applies an HTML wrapper around a series of SVG and JSON assets to render the layers and animations properly, along with hidden accessibility content that’s machine-readable. The exact approaches for implementing each of these types can vary from one website to the next, so if you have questions, just let us know.

If you’re publishing a video, you’re probably using a platform like YouTube or Vimeo to host it, then embedding the video player on your web page. Remember: Always upload captions with your video.

Transcript

The bottom part of our content sandwich is a transcript of the written or spoken content in the visual. Often this comes in the form of an expandable accordion labeled with “View transcript” or “View text,” but it doesn’t strictly have to be an accordion. The point is to include the written or spoken content as live text somewhere on the page, in whatever format works best for your website and visual.

Here are a few examples from Lemonly and our clients showing this approach in action:

• Lemonly: Illustrated guide to the dragons of HBO's House of the Dragon
• Wells Foodservice: 3-gallon packaged ice cream for foodservice
• DistributeRx: Prescription access made simple

It’s worth noting that if your infographic is responsive — meaning it’s built entirely out of HTML, with live text within the visual itself — you don’t need to include the transcript since the written content already exists as live text on the page. 

5. Schema markup

Our last tip is more advanced and generally best handled by a web developer, someone who’s comfortable putting code on your website, or maybe your trusted pals at Lemonly. It’s called schema markup — it’s a type of structured data written as JSON-LD that describes your content using a nested set of properties.

What is schema markup for visual AI SEO?

Think of it like a robust tagging system that lets you name and relate many different attributes of your content as metadata. Using schema essentially tells search engines and AI, “hey, there’s something important here” — it helps AI understand your content and can help you get rich snippets in search results.

For infographics and videos, two of our most common content types, we mainly use ImageObject and VideoObject entities with multiple related properties including the publisher, date, topic, description, keywords, chapters, and more. Check out a couple schema examples and templates below — including the example projects further down the page.

Content-level schema vs. site-level or page-level schema

It’s worth noting that here we’re talking about content-level schema that applies to specific pieces of content. You can also have site-level schema that defines your organization’s identity and page-level schema for some of your highest-value pages. You can create structured data for any or all of these levels. You can really get as complex as you want.

Use these tips in combination based on what works for your content

Remember to use these in combination, incorporating as many as apply to your project. Our goal is to give AI multiple signals that reinforce the credibility and topical relevance of your content.

Examples: Visual content AI SEO in action

Let’s look at two examples to put these tips into practice — one infographic and one video example.

Example 1 — Infographic: Who is SUE the T-rex?

We’ll start with this infographic about SUE the T-rex, who was discovered right here in our home state of South Dakota. (Fun fact: This is the very first infographic I wrote as a Lemonly intern back in 2017.) Let’s see what we can do to help its AI SEO potential.

File: Upgrading from JPG to SVG for visual quality

This infographic is currently a JPG. If I want, since I have the working file, I could use an SVG instead for sharper visual quality. But since this infographic is static, not animated or responsive, opting for an SVG is mostly an aesthetic choice with minimal AI SEO impact. 

If I wanted the biggest gains, I’d make it fully responsive in HTML, but let’s say that's outside the scope of what I want to invest in this project for now.

Filename and alt text: Descriptive and keyword-aligned

I’ll start with the filename, giving it a meaningful, human-readable title. I could change the filename locally before uploading the file or right here in my CMS if it’s already uploaded to my website.

• Before: 170706_Lemonly_IG_WhoisSue_Final.jpg
• After: SUE T-Rex Discovery Timeline_South Dakota’s Biggest Dinosaur Fossil_Infographic.svg

Next, I’ll add some concise, descriptive alt text, around the length of one sentence. I can’t capture everything about this infographic in so few words, so I’ll go for the overall purpose of the graphic.

• Alt text: Infographic illustrating facts about SUE the T-Rex fossil and a detailed timeline of the fossil’s history as a living dinosaur, discovery in South Dakota, and battle for its custody.

Content sandwich: Making the page content-rich

Now we’ll move on to our content sandwich — the surrounding content on the page.

Intro content

I think this intro content can be better. I’m going to use around three paragraphs, since I have three goals in mind: 

1. Introduce the project
2. Summarize the content of the infographic
3. Highlight the creative choices that went into designing it (since this is a page in our website portfolio)

Just for fun, I’ll even add a button to download the infographic to make sharing easy.

Transcript

For the transcript below the infographic, I’ll put all the written content here in an accordion. I can pretty much copy this right from the copy doc, just making sure the heading levels proceed sequentially with the rest of the page content.

That completes my content sandwich, but I could also add more to this page for a more content-rich experience by adding other relevant information and resources, like some text from an old blog post, the moodboard showing the visual inspiration, and links to other resources where folks can learn more. I can make this page as robust as I want — the infographic is still the hero, but all the surrounding content makes the page richer, more engaging, and topically relevant.

Schema markup: Structured metadata geared toward AI search

Last, I’ll add some schema markup to this page with a variety of metadata about the infographic, publisher, topic, keywords, sources, and more. Click below to see the full schema markup.

And there we go! With just a little bit of time and effort, I definitely improved the overall quality of the page and boosted its potential AI visibility.

Example 2 — Video blog post: Accessibility tips for visual content

For one more example, let's look at one of our YouTube videos — just like the blog post you're reading right now. We'll use our video blog about accessibility tips for visual content. This shows how we treat longer-form video content, but the same principles can apply to a shorter explainer video as well.

YouTube publishing steps

Every video we publish includes all of the following when publishing:

• Filename - Change this to be aligned with your topic and keywords before uploading.
• Title - The user-facing video title.
• Description - Usually a couple paragraphs including the hook, summary, and benefit, plus any relevant links and some boilerplate links to Lemonly across the web.
• Chapters - Included in the description for easy navigation (automatically applied to the video player). Chapters are more suited to videos longer than three minutes.
• Hashtags - Good discovery signals. Up to three included in the description will be shown next to the title.
• Tags - Up to 15 keywords related to the video; invisible to the viewer. (YouTube says these play a minimal role in discovery, but hey, it can't hurt!)
• Custom thumbnail - Sometimes we A/B test multiple thumbnails or video titles to see what works.
• Captions - Uploaded as an SRT file.
• Cards - Clickable, in-video banners that appear in the upper-right corner.
• End screen - Clickable tiles that appear at the end of the video to promote other videos, a playlist, or your channel.
• Playlist - A collection of related videos on your channel. Signals that this content is related to other videos in the playlist.

Video embed and schema markup

Once it’s uploaded to YouTube, I’ll embed the video on our web page. We’re putting this video inside a blog post. While I’m setting up the embed, I’ll also add the appropriate VideoObject schema since I’m already in my code editor. 

For a video, we’ll include all the chapters and the full video transcript in the schema, so the structured data is a complete representation of the video content.

Content sandwich: Long-form blog post

For the content sandwich, I’m using a more robust approach on this page. This video is about 15 minutes long, and the transcript is a couple thousand words. I want basically all that content to be live text on the page, so I’ll repurpose the video script into a full blog post with headings, sections, links, and in-line graphics. 

It’s the same content sandwich approach we’ve been talking about, but I adapted the format of the content to fit the video and our content strategy for this page. The point is, you can expand your content sandwich to fit your needs — add whatever toppings and add-ons you want. As long as the core structure is there, you’re doing it right.

How to get started with visual content AI SEO

So that’s an intro to AI SEO for visual content! We covered what AI SEO is, why it matters for visual content, three principles to keep in mind, and tips and examples. 

This is a big and quickly evolving topic — it’s easy to get overwhelmed or feel like you have to do everything all at once. Here are two ways to get started:

1. Review our technical foundations checklist with your web team to make sure your website is set up for success. It’s worth making progress on those site-level steps before going all-in on content-level optimization.
2. Start by optimizing the key visuals on your highest-value pages. You don’t have to optimize every visual on every page right away. You can keep going with new content you’re publishing now.

Your partner for AI-ready visual content

If you have questions or are looking for a partner to help create visual content that’s optimized for AI SEO, reach out to Lemonly — we'd love to help. 

You can also subscribe to our YouTube channel for more content marketing tips from Lemonly.