What is Image Alt Text?
Image alt text is a short written description of an image, set on the HTML alt attribute (for example, alt="Black leather Chelsea boot, side profile"). It is invisible on the page but read aloud by screen readers, shown when an image fails to load, and parsed by AI engines and search crawlers. WCAG 2.1 Success Criterion 1.1.1 requires a text alternative for all non-text content ā meaning every meaningful image needs alt text.
AI engines like ChatGPT, Perplexity, and Google AI Overviews increasingly cite multi-modal content. Most do not run vision on every image ā they read the alt text, captions, and surrounding HTML to decide what an image shows and whether to cite the page. Pages where text, images, and structured data all reinforce the same topic get cited far more often. This metric is part of the Technical Foundation pillar in your GEO-Score.
Why Alt Text Matters for AI Search
Alt text is one of the few signals that pulls triple duty: it is a legal accessibility requirement, a confirmed image search ranking factor, and the primary text input AI engines use to interpret your visuals. Skipping it costs you on all three fronts simultaneously.
Accessibility is the baseline (and the law)
WCAG 2.1 Level A requires a text alternative for all non-text content. The WebAIM Million 2025 study found 18.5% of all home page images still lack alt text ā and missing alt text is the most common WCAG violation cited in the 3,117 ADA Title III website accessibility lawsuits filed in 2025.
AI engines cite multi-modal pages more often
Wellows analyzed 15,847 AI Overview results across 63 industries and found pages combining text, images, video, and structured data see 156% higher selection rates ā with full multi-modal plus schema delivering up to 317% more citations. Alt text is what turns an image into a citable signal.
Image search and visual citations need text
Google Lens processes more than 12 billion visual queries per month and is growing 30% per year. Descriptive alt text is the single highest-impact image SEO action ā but a Backlinko study of 65,388 visual searches found only 11.4% of ranking images had alt text matching the query.
What the Research Says
Pages that combine text, images, video, and structured data see 156% higher selection rates in AI Overviews, with full multi-modal plus schema integration delivering up to 317% more citations. Multi-modal integration showed a 0.92 correlation coefficient ā the biggest ranking shift of 2025.
ā Wellows, Google AI Overviews Ranking Factors Study (15,847 results, 63 industries), 2026
18.5% of all home page images had missing alternative text ā about 11 missing alt attributes per page on average. A further 11% of images with alt text used questionable values like "image", "graphic", a file name, or duplicates of adjacent text.
ā WebAIM Million 2025, Accessibility Audit of 1,000,000 Home Pages
Only 11.4% of all Google Lens result images contain alt text terms that match the keyword someone just searched for. Contextually rich alt text can improve visual search discovery by up to 35%, particularly for e-commerce and product-focused sites.
ā Backlinko, Google Lens Study of 65,388 Visual Searches, 2025
Real Examples: Bad vs. Good
Alt text quality is binary in practice ā it is either useful to a screen reader and AI parser, or it is noise. These three real-world scenarios show how small wording changes flip an image from invisible to citable.
Example 1: E-commerce product image (Chelsea boot)
<img src="image1.jpg" alt="image1.jpg">
Why this fails: The file name as alt text gives no information about the product. Screen readers will read "image one dot jay peg". Google Image Search has no signal to match the query "black Chelsea boot". AI engines see the image but cannot describe or cite it.
<img src="chelsea-boot-black.jpg" alt="Black leather Chelsea boot, side profile, men's size 10, with elastic gusset and pull tab">
Why this works: A blind shopper hears the exact product. Google can match the alt text to color, style, and size queries. ChatGPT and Perplexity can cite this page when asked "best black Chelsea boots for men". Same image ā totally different outcome.
Example 2: Blog hero image (article on remote work)
<img src="hero-banner.png">
Why this fails: With no alt attribute at all, many screen readers fall back to reading the file name ("hero banner dot p n g"). It violates WCAG 1.1.1 Level A. AI crawlers cannot place the image in context, and the page misses the multi-modal citation boost Wellows measured.
<img src="remote-team-video-call.png" alt="Distributed engineering team on a video call, with shared screen showing a Kanban board">
Why this works: The alt text describes both subject and context, reinforcing the article's topic. AI engines now have aligned signals (heading, body, image alt) ā exactly the multi-modal pattern that earns 156% more AI Overview citations.
Example 3: Data visualization (quarterly revenue chart)
<img src="q4.png" alt="chart">
Why this fails: WebAIM flagged "chart" as a classic questionable alt value (11% of all alt text on the web is this kind of placeholder). A screen reader user learns nothing. AI engines cannot extract the numbers and will skip the page when summarising revenue trends.
<img src="q4-revenue-bar-chart.png" alt="Bar chart of quarterly revenue 2025: Q1 $4.2M, Q2 $4.8M, Q3 $5.1M, Q4 $6.7M ā 60% year-over-year growth">
Why this works: The chart type is named, the four data points are listed, and the headline takeaway is included. WCAG complex-image guidance is satisfied, and AI engines can cite the actual numbers in answers about revenue trends.
How to Improve Your Image Alt Text
Do NOT Do This
- āOmit the alt attribute entirely ā many screen readers will read the file name aloud, which is worse than nothing
- āUse raw file names like IMG_1234.jpg, screenshot-final-v2.png, or hero.webp as the alt value
- āStart every alt with "Image of" or "Picture of" ā screen readers already announce the element as an image, so it just wastes characters
- āPack alt text with unrelated keywords ā Google explicitly calls this spam in its image SEO documentation and may penalise the page
- āLeave decorative images (dividers, background flourishes) without alt="" ā without an empty alt, screen readers may try to describe meaningless visuals
Do This Instead
- āDescribe what the image shows AND why it is on the page ā content + context, not just labels
- āFor charts, infographics, and screenshots, lead with the type ("Bar chart...", "Screenshot of...", "Diagram showing...")
- āKeep alt text in the 80-125 character range ā long enough to be specific, short enough that screen readers do not truncate
- āUse alt="" (with no space between the quotes) for decorative images so screen readers and AI cleanly skip them
- āMake sure file name, alt text, caption, and surrounding paragraph all describe the same thing ā aligned signals win in AI search
Quick Tips for Better Alt Text
- ā¢Add an alt attribute to every img tag ā descriptive for content images, alt="" for decorative ones. There is no third option.
- ā¢Aim for 80-125 characters. Long enough to be specific, short enough that JAWS and NVDA do not truncate the description.
- ā¢Skip "Image of", "Picture of", "Photo showing". Screen readers announce the role themselves ā those words are pure noise.
- ā¢For charts, infographics, and diagrams, write a short alt PLUS a longer description in the surrounding text or a linked detail page.
- ā¢If an image is also a link, alt text should describe the link's destination, not just the picture. WebAIM found 44% of missing-alt images are links.
- ā¢Run a free GEO-Score Check to spot missing or low-quality alt text across your page in seconds ā fixing these is one of the fastest wins.
Frequently Asked Questions
How long should alt text be?
What is the difference between alt="" and a missing alt attribute?
Do AI engines like ChatGPT actually read alt text?
Does alt text help my Google ranking?
How do I write alt text for charts, graphs, and infographics?
Can I just use AI to auto-generate alt text for my whole site?
Related Metrics to Explore
- AI Optimization
Alt text is one technical AI signal ā see the full set of optimizations that make your page readable to ChatGPT, Perplexity, and AI Overviews.
- Content Structure
Aligned signals win in AI search. Learn how heading hierarchy and image placement work together to make passages extractable.
- Schema Validator
Pair alt text with ImageObject schema to give AI engines structured metadata. Schema plus multi-modal content is the 317% citation lift Wellows measured.
- Comprehensiveness
Multi-modal pages with text, images, and data cover topics more completely. Learn how comprehensiveness compounds with alt text for AI visibility.