Alt Text Is Not Dead. Lazy Image SEO Is
A strange thing happens whenever AI gets better at doing something. A group of marketers immediately decides the old practice is dead.
AI can write, so content strategy is dead. AI can summarise pages, so information architecture is dead. AI can see images, so alt text is dead.
The funeral industry must be thrilled.
Like everything else, image recognition has improved. AI systems can identify objects, read text in images, describe scenes, and make useful guesses about what is happening in a visual. But that does change image SEO, and it does not remove the need for image SEO.
It only removes the excuse for treating image SEO as a field where someone types “sustainable skincare product eco green organic natural” into an alt attribute and calls it a day.
The tool has spoken. It has not necessarily said anything useful yet.
AI can recognise an image, but it may still misunderstand the point.
There is a difference between recognising what appears in an image and understanding why that image matters on a page. For example:
A visual system may recognise: “A bronze object in a display case.”
That is useful, but it’s also wildly insufficient if the page is about a Roman brooch, a museum collection, metalwork techniques, trade routes, conservation work, or public access to cultural heritage.
A product image may be recognised as:
“A bottle on a beige background.”
Helpful. Also useless if the commercial and search value sits in the refill mechanism, packaging material, certification, ingredient transparency, product use case, or sustainability claim.
An NGO project photo may be recognised as:
“People planting trees.”
Fine. But the actual meaning may be tied to a wetland restoration project, a local community programme, a biodiversity target, a grant-funded outcome, or evidence of impact.
This brings us to the obvious – AI can describe the pixels, but the SEO still has to describe the relevance.
Discover How SEO Can Improve Your Online Visibility
Request a Consultation Today!
Google did not throw alt text into the sea.
Google’s image SEO documentation still says it uses page content, captions, image titles, filenames, and alt text to understand images. It also says Google uses alt text together with computer vision algorithms and page content to understand the subject matter of an image.
That last part matters.
Google’s own image SEO documentation says it uses “alt text along with computer vision algorithms and the contents of the page” to understand an image’s subject matter. It also says Google extracts image information from page content, including captions and image titles.
Computer vision did not replace text-based context, but joined it.
The most important attribute for providing more metadata about an image is alt text, while warning against stuffing alt attributes with keywords because it creates a bad user experience and may be treated as spam.
So the useful position is that the alt text is one part of the evidence layer around an image.
Alt text was never supposed to be an SEO dumping ground.
The phrase “alt tag” is already a small warning sign. Technically, it is an alt attribute. The web survives the terminology crime, but the misunderstanding often follows.
Alt text exists first for accessibility.
W3C’s guidance makes the decision contextual. If an image contains text that is not otherwise present, the alt text should include that text. If an image is used as a link or button, the alt text should communicate the action or destination. If an image contributes meaning, the alt text should convey that meaning. If the image is purely decorative, an empty alt attribute may be the right answer. For complex graphics, the information should be available elsewhere on the page.
The correct alt text depends on the role of the image. The same image can need different alt text in different contexts. A photo of a museum object on a collection page may need object type, material, date, and collection context. The same photo used in a decorative banner may need empty alt text if the nearby heading already carries the meaning.
The bad habit: treating every image like the same SEO task.
Many websites handle images in one of four bad ways.
First, they leave important images with no useful text alternative.
Second, they stuff keywords into alt text because someone read a blog post in 2014 and never recovered.
Third, they use the same generic alt text across hundreds of images.
Fourth, they put important information inside the image and nowhere else, then hope search engines, screen readers, AI systems, and users will politely solve the problem for them.
This is how you get an infographic with the alt text “chart”.
For small teams, the problem is worse because image work often sits in the grey swamp between content, design, development, accessibility, SEO, and whoever last touched the CMS. Ownership becomes fog. Fog becomes backlog. Backlog becomes “we’ll clean this up later,” which is where good intentions go to develop a smell.
Image SEO now means more visual context.
A useful image SEO system answers several questions.
Can the image be discovered and indexed?
Google recommends using standard HTML image elements, because it can find images in the src attribute of an <img> element, while it does not index CSS images.
Is the image near relevant text?
Google says page content and metadata can strongly influence where images appear in search, and it recommends placing images near relevant text on pages that match the image subject.
Does the image have useful metadata?
Short descriptive filenames, captions, titles where appropriate, alt text, and structured data all help create context. Google says filenames provide light clues, and it recommends descriptive filenames rather than generic names such as IMG00023.JPG.
Is the image representative of the page?
For preferred images, Google recommends using relevant, representative, high-resolution images in metadata such as primaryImageOfPage, schema image properties, or og:image, and avoiding generic images such as a logo when a better representative image exists.
Does the important content exist as text?
This is where AI search makes the point sharper. Google’s guidance for AI features says existing SEO fundamentals still matter, including making important content available in textual form, supporting text with high-quality images and videos where appropriate, and ensuring structured data matches visible text.
Just…do not trap your evidence inside a JPEG and act surprised when it behaves like a JPEG.
What this means for AI search and LLM visibility?
The phrase “LLM SEO” attracts nonsense at industrial speed, so keep the claim modest.
AI systems are getting better at interpreting visual content. Some search products also use crawlers, indexes, retrieval systems, and page-level signals. OpenAI’s documentation, for example, says OAI-SearchBot is used to surface websites in ChatGPT’s search features, and that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though they may still appear as navigational links.
Google’s AI features guidance is boring in the most useful way: there are “no additional technical requirements” for AI Overviews or AI Mode. Existing SEO fundamentals still apply, including making important content available in text, supporting that text with high-quality images and videos, and ensuring structured data matches visible page content.
This only means the basics become more expensive to ignore.
If your images are inaccessible, uncrawlable, context-poor, duplicated, generic, slow, or detached from useful text, you are asking machines to infer meaning from weak evidence.
Sometimes they will. Sometimes they will not. Sometimes they will confidently describe the wrong thing, which is very on-brand for the internet.
Better image SEO gives search systems and assistive technologies a clearer job.
Here is the image.
Here is what it shows.
Here is why it matters.
Here is the page context.
Here is the entity, object, product, place, person, project, or data behind it.
Here is the same important information in text, because we are not monsters.
Examples: weak alt text vs useful visual context
For a museum collection image:
Weak:
alt="old object"Better:
alt="Bronze Roman horse brooch displayed on a neutral museum background."Stronger page context:
The surrounding copy explains the object type, approximate date, material, collection name, excavation context where appropriate, and why the object appears on the page.
The alt text does not need to become a dissertation with a tiny saddle. The page can carry the fuller meaning.
For a sustainable product page:
Weak:
alt="eco friendly skincare product natural organic green beauty"Better:
alt="Refillable aluminium moisturiser bottle with recyclable refill pouch."Stronger page context:
The product page explains the packaging material, refill system, certification scope, ingredient sourcing, disposal instructions, and claims that can actually be substantiated.
Adjectives are cheap. Evidence has invoices.
For an NGO impact page:
Weak:
alt="people outside"Better:
alt="Community volunteers planting native trees beside a restored wetland."Stronger page context:
The page names the project, location, date range, partner organisations, measurable outcomes, and how the image supports the impact claim.
The image becomes evidence instead of mood wallpaper.
What should teams fix first?
Do not start by rewriting every image on the site.
That is how image SEO becomes another spreadsheet with a heroic number of rows and no adult supervision.
Start where the cost of misunderstanding is highest.
1. Fix images that carry important information.
Charts, infographics, event posters, diagrams, maps, screenshots, product comparison tables, collection labels, and campaign graphics need attention first.
If the image contains information users need, that information should exist in HTML text nearby or in a useful equivalent format. The alt text can summarise the image, but the page should carry the substance.
A complex chart with “alt=chart” is a shrug in markup.
2. Fix images on commercial or mission-critical pages.
Product pages, service pages, donation pages, event pages, collection pages, case studies, reports, and key landing pages should have meaningful image context.
These pages often carry the business or public-value case. Do not let the visual evidence sit there wearing a filename from a camera roll.
3. Fix repeated generic alt text.
If every product image says “product image,” nobody wins.
Generic alt text creates noise for screen reader users and weak context for search systems. It also tells the entire organisation, quietly but firmly, that nobody owns the details.
4. Remove noisy alt text from decorative images.
Decorative dividers, background textures, repeated logos, and purely visual flourishes often do not need descriptive alt text. W3C guidance supports empty alt attributes for purely decorative images.
5. Add context around original, evidence-heavy images.
Original photos are valuable because they prove things.
For purpose-driven organisations, this matters. Museums have collections, archives, exhibitions, conservation work, public programming. NGOs have projects, partners, locations, communities, outcomes. Sustainability brands have materials, packaging systems, processes, certifications, and trade-offs.
The image should connect to that evidence.
Otherwise it becomes a nice picture that search systems may understand only at the level of “people in room” or “object on table.”
Technically true. Spiritually useless.
What can wait?
Some image work is useful but rarely urgent.
Renaming every legacy image file across a large site can wait if it risks breaking URLs, cache behaviour, CMS references, or developer trust. New uploads should follow a better naming convention. Old files can be handled when pages are being updated anyway.
Perfecting alt text on low-traffic, low-value decorative pages can wait.
Adding elaborate structured data before the visible page content is accurate can wait. Google’s AI feature guidance says structured data should match visible text on the page. Schema that describes a reality the page does not show is not strategy. It is cosplay with braces.
Bulk AI-generated alt text without human review can also wait in the nearest bin.
AI can help draft alt text. It can describe obvious visual features. It can speed up the first pass. But it may miss the product detail, collection significance, sustainability limitation, or access need that gives the image its actual purpose.
Use AI as an intern with good eyesight.
Do not make it head of accessibility.
A better image SEO workflow
Image SEO should become part of publishing, not a guilt project conducted every two years after a crawl screams at everyone.
For each meaningful image, ask:
- What does this image add to the page?
- What would someone miss if they could not see it?
- Is the important information also available as text?
- Does the alt text describe the relevant meaning rather than the entire visual scene?
- Is the image connected to the page’s entity, product, object, project, data, or claim?
That is enough to make better decisions without turning every CMS upload into a committee meeting with snacks and despair.
For teams with limited time, the workflow can be simple.
When publishing a page, identify the images that matter. Write useful alt text for those. Add captions when interpretation helps. Make sure the image sits near relevant copy. Use descriptive filenames for new files. Choose a representative og:image or schema image. Leave decorative images empty.
Fix the blockers. Improve the assets that carry meaning. Stop feeding the alt-text goblin.
From image description to visual evidence
AI image recognition raises the standard for image SEO because it makes weak context more obvious.
If a machine can already see “a bottle,” your job is not to repeat “bottle” with three adjectives and a keyword dream attached.
Your job is to make the image useful.
For a sustainable brand, that means connecting the visual to proof.
For a museum, that means connecting the object to collection knowledge.
For an NGO, that means connecting the photograph to mission-critical evidence.
For a B2B SaaS company, that means making diagrams, screenshots, workflows, and interface images understandable as part of the buying and implementation process.
The right image can support trust, clarify a claim, reduce ambiguity, or, it can help users decide about their actions. The right image can also help search systems understand what the page contains and it can make content more accessible to people who cannot see the image.
That is the job.
The practical takeaway
Alt text is still important.
It is also too small a box to hold the whole conversation.
Modern image SEO needs accessible alt text, crawlable image implementation, relevant surrounding copy, descriptive filenames, useful captions, structured data where appropriate, strong representative images, and important information available as text.
AI can recognise more than it used to. Good. Now give it something worth recognising.
Lazy image SEO is dead.
Alt text has work to do.
FAQ
Is alt text still important for SEO?
Does AI image recognition make alt text unnecessary?
Does Google still use alt text to understand images?
What makes good alt text?
Should every image have descriptive alt text?
Can I use AI to generate alt text?
What matters for image SEO besides alt text?
Should keywords be included in image alt text?
Is image SEO important for AI search and LLM visibility?
Author
Milica McAreavy is a technical SEO consultant specializing in aiding purpose-driven entities, green brands, and cultural institutions.
She helped shape the digital presence of various NGOs, green startups, and nonprofits, leveraging SEO to champion purpose and sustainability.
And when she’s not busy optimizing websites, you’ll find her traveling, visiting ancient landmarks, singing to cats, and scouting flea markets for old books.
Stay Updated!
Get the latest archSEOlogist posts directly in your inbox!
