Image SEO: The Optimisation Most Teams Do Last and Regret

Image SEO is the practice of optimising visual assets so search engines can index them, understand their relevance, and surface them in both standard and image search results. Done properly, it adds a measurable traffic channel, strengthens page relevance signals, and improves Core Web Vitals scores that affect how your pages rank overall.

Most teams treat it as a checkbox at the end of a content sprint. That is the wrong order. The decisions you make when you choose, name, compress, and tag an image affect page speed, crawl efficiency, and ranking potential in ways that are genuinely difficult to reverse after publication at scale.

Why Image SEO Gets Deprioritised and Why That Is a Commercial Mistake

I have sat in enough quarterly reviews to know how this plays out. The content team is under pressure to publish. The brief gets written, the article goes live, someone drops in a stock image, adds a vague alt tag like “marketing strategy”, and moves on. It happens across every agency I have run and most client-side teams I have worked with.

The problem is not laziness. It is sequencing. Image optimisation feels like a finishing step, so it gets treated like one. But the choices made at upload time, file format, file name, dimensions, alt text, compress or not compress, create a technical baseline that affects every crawl of that page from that point forward.

When I was running iProspect and we were scaling the team from around 20 people to close to 100, one of the things I noticed was that the accounts with the most consistent organic growth were also the ones with the most disciplined content operations. Not the most creative ones. The ones where someone had built a repeatable process for every element of on-page work, including images. It was not glamorous. It was operationally sound. There is a difference.

Image search also drives real traffic. Depending on your category, a meaningful share of discovery journeys start in Google Images. For e-commerce, interiors, food, fashion, and anything visually led, that share is significant. Ignoring it is not a neutral choice. It is leaving a traffic channel unmanned.

If you want the full picture on how image optimisation fits into a broader organic strategy, the Complete SEO Strategy hub covers the architecture of how all these signals connect.

What Google Actually Does With Your Images

Google cannot see images the way a human can. It reads signals around the image and uses those to infer what the image depicts and whether it is relevant to a given query. Understanding this changes how you approach every element of image optimisation.

The signals Google uses to understand an image include the file name, the alt attribute, the surrounding text content, the page title, any structured data referencing the image, and the image’s position within the page structure. Google also uses its own machine learning models to analyse image content directly, but those models work better when the surrounding metadata confirms what they detect.

Think of it this way. If Google’s vision model detects a photograph that looks like a kitchen, and the file is named “img_4872.jpg” with an alt tag of “image”, it has very little confidence about which kitchen queries this image should appear for. If the file is named “open-plan-kitchen-island-white-cabinets.jpg” with an alt tag describing the specific scene, and the surrounding text discusses kitchen renovation costs, Google has multiple confirming signals. That image becomes rankable for specific queries. The other one does not.

Google also uses images as a quality signal for the page itself. A page with well-optimised, contextually relevant images reads as more authoritative than one with generic stock photography that has no connection to the content. This is one of the subtler ways image quality feeds into E-E-A-T assessment, even if it is rarely discussed in those terms.

File Names: The Optimisation That Happens Before the Page Loads

File naming is the most consistently neglected part of image SEO. Most images arrive from designers, photographers, or stock libraries with names like “shutterstock_1234567890.jpg” or “DSC_0047.jpg”. If you upload them with those names, you have already missed the first optimisation opportunity.

The file name is included in the image URL, which Google crawls. It is one of the earliest signals processed when a crawler encounters an image. Rename your images before upload using descriptive, hyphen-separated phrases that reflect the image content and its relevance to the page topic.

A few principles worth following. Use hyphens, not underscores, as word separators. Google treats hyphens as word separators and underscores as word joiners, so “red-running-shoes.jpg” is read as three separate words, while “red_running_shoes.jpg” is read as one compound string. Keep names concise but descriptive, three to six words is usually sufficient. Do not keyword-stuff. A file named “buy-red-running-shoes-cheap-discount-sale.jpg” is not useful to anyone and signals manipulation rather than relevance.

If you are working with a large content archive, auditing and renaming every historical image is a significant project. Prioritise your highest-traffic pages and your most commercially important content first. The effort compounds over time as those pages accumulate more crawls with better signals.

Alt Text: Writing It for Relevance, Not Compliance

Alt text exists primarily to describe images for users who cannot see them, whether because of a visual impairment, a slow connection, or a browser that has blocked image loading. That accessibility function is the right reason to write good alt text. The SEO benefit is a consequence of doing that well, not a separate goal.

I have judged enough Effie submissions to know that the best marketing work tends to solve a real problem and create commercial value as a by-product. Alt text is a small example of the same principle. Write it to genuinely describe the image for someone who cannot see it, and you will naturally produce text that is specific, contextual, and relevant. That is exactly what Google wants.

Practically, this means describing what is in the image, not what you want to rank for. “Woman using a laptop at a standing desk in a home office” is useful alt text. “Remote working productivity tips” is not a description of an image, it is a keyword phrase wearing alt text clothing. Google can tell the difference, and screen readers definitely can.

A few specific guidance points. Decorative images that add no informational value, dividers, background textures, purely aesthetic icons, should have empty alt attributes (alt=””) rather than descriptive text. This tells screen readers to skip them and prevents cluttering the accessibility experience. For complex images like charts or infographics, the alt text can be supplemented with a longer description in the surrounding text or a caption. For product images, include the product name, key variant details like colour or size, and any relevant context.

Keep alt text under 125 characters where possible. Screen readers typically truncate at that point, and longer alt text tends to become padded rather than more descriptive.

File Format and Compression: Where Performance and Quality Meet

Image file size is one of the more direct connections between image decisions and page ranking. Large, uncompressed images slow page load times, which affects Largest Contentful Paint (LCP), one of the Core Web Vitals metrics that feeds into Google’s page experience signals. This is not a theoretical connection. It is a direct mechanical relationship.

The format question has become much clearer over the past few years. WebP delivers substantially smaller file sizes than JPEG or PNG at equivalent visual quality. AVIF goes further still, with even better compression ratios, though browser support is slightly less universal. For most content sites in 2026, serving WebP as the default and falling back to JPEG for older browsers is the right call. There is no meaningful visual quality argument for continuing to serve uncompressed JPEG at scale.

Compression tools have also improved significantly. Lossless compression removes redundant data without affecting visual quality. Lossy compression reduces quality slightly in exchange for much smaller files. For most web use cases, a mild lossy compression at 75 to 85 percent quality produces images that are visually indistinguishable from originals at a fraction of the file size. Tools like Squoosh, ShortPixel, and Imagify make this accessible without specialist knowledge.

Serving correctly sized images is equally important. If the maximum display size of an image on your site is 800px wide, serving a 2400px source file is wasteful. The browser downloads the full file and then scales it down. Responsive image attributes, specifically srcset and sizes, allow you to serve different image sizes to different viewport widths. This is standard practice for any site with meaningful mobile traffic, which is most sites.

Lazy loading is worth implementing for images below the fold. The loading=”lazy” attribute tells browsers not to load an image until it is about to enter the viewport. This reduces initial page load time and bandwidth consumption without any visible effect on the user experience. It is one of those technical decisions that is easy to implement and has a measurable impact on page speed scores.

Structured Data for Images: When Metadata Creates Ranking Eligibility

Structured data does not directly improve image rankings in the traditional sense. What it does is make your images eligible for rich results that standard optimisation cannot achieve. For certain content types, that is a meaningful distinction.

Product schema with an image property makes your product images eligible to appear in Google Shopping surfaces and product-rich results. Recipe schema with image markup makes your recipe images eligible for the rich recipe cards that appear prominently in both standard and image search. Article schema with a featured image property helps Google associate the correct image with your article in Discover and news surfaces.

The image URL referenced in your structured data must match the image that actually appears on the page. Google validates this. If the schema references an image that is not present or accessible, the structured data is ignored. Similarly, the image must meet minimum dimension requirements for rich result eligibility, typically at least 1200px wide for article images in Discover.

For e-commerce teams, getting product image structured data right is one of the higher-leverage technical investments available. I have seen product catalogues where a structured data audit and correction exercise delivered a measurable lift in organic product visibility within a few months of implementation. The work is not complicated, but it requires someone to actually do it systematically rather than assuming it is handled.

Image Sitemaps: Helping Google Find What It Might Otherwise Miss

Google can discover most images by crawling your pages normally. But images loaded via JavaScript, images in iframes, or images on pages with thin crawl budgets may not be discovered reliably. An image sitemap addresses this by explicitly listing the images you want indexed along with relevant metadata.

You can either create a dedicated image sitemap or extend your existing XML sitemap with image tags. The image sitemap format allows you to include the image URL, a title, a caption, a geographic location if relevant, and a licence URL. Google’s documentation is specific about what it accepts, and the format is straightforward to implement if you have CMS access or developer support.

For large sites with thousands of product images or a significant visual content library, an image sitemap is worth the setup time. For smaller sites where all images are embedded in crawlable HTML pages, it is a lower priority. The question to ask is whether there are images on your site that you want indexed but that Google might not reliably find through normal crawling. If the answer is yes, an image sitemap is the solution.

Submit your image sitemap through Google Search Console and monitor the coverage report. If images are listed in the sitemap but not appearing as indexed, the coverage report will surface the specific reasons, whether that is a crawl error, a noindex directive, or a content policy issue.

Captions, Surrounding Text, and Contextual Relevance

Images do not exist in isolation on a page. Google reads the text around an image as part of its relevance assessment. The paragraph immediately preceding and following an image, the section heading it sits under, and the overall page topic all contribute to how Google interprets what the image is about and what queries it should appear for.

Captions are read by more users than most body copy. Eye-tracking studies have consistently found that image captions have high readership, often higher than the paragraphs surrounding them. Writing descriptive, informative captions is good for users and provides Google with additional contextual signals about the image. A caption does not need to repeat the alt text verbatim. It should add context that the alt text does not cover.

Placing images near the text they illustrate is also a relevance signal. An image of a product placed in a section discussing that product’s features will be associated with those features by Google. The same image placed at the top of the page as a decorative header, disconnected from any specific text, gives Google less to work with.

This is one of those areas where good editorial practice and good SEO practice align completely. Images should be placed where they add meaning to the content, not just where they break up visual monotony. When you follow that editorial principle, the SEO signals take care of themselves.

Common Image SEO Mistakes That Are Worth Auditing For

After working across dozens of client accounts and multiple agency audits, the same mistakes appear with remarkable consistency. None of them are complicated to fix. They accumulate because no one has made image quality a first-class concern in the content process.

Missing alt text is the most common. Not wrong alt text, just absent. CMS platforms vary in how they handle this, and some make it too easy to publish without completing the alt field. Auditing for missing alt text using a crawl tool like Screaming Frog takes about twenty minutes and produces a prioritised list you can work through systematically. There is a good case in the Moz blog on testing beyond title tags for treating these kinds of foundational elements as testable variables rather than assumptions.

Duplicate alt text across multiple images is a close second. If you have twenty product images all tagged with the same alt text, you are not giving Google anything useful to differentiate them. Each image needs alt text that describes that specific image.

Serving oversized images is nearly universal on sites that have been running for more than a few years without a technical audit. Images uploaded at full camera resolution, stock images served at their native dimensions, hero images that are 4MB files. These are page speed problems that also happen to be image SEO problems. A well-structured homepage introduction loses its impact entirely if the hero image takes four seconds to load.

Blocking images in robots.txt is less common than it used to be, but it still happens on sites that have carried forward old crawl directives without reviewing them. If your robots.txt disallows access to your image directories, Google cannot index those images regardless of how well everything else is optimised. Check it.

Finally, using images as the only carrier of important information. Text embedded in images, key data points presented only in a chart image, calls to action that exist only as image files. Google cannot reliably extract text from images, and users on screen readers cannot access that information at all. If the information matters, it needs to exist as actual text on the page.

Building Image Optimisation Into the Content Process

The reason image SEO problems persist is not ignorance. Most content teams know what good looks like. The problem is that image optimisation is treated as an individual task rather than a process step. When it depends on whoever happens to be publishing remembering to do it, it will be inconsistent.

The fix is to build it into the workflow. A content checklist that includes image file naming, compression, alt text, and caption review before publishing takes about three minutes per article and eliminates the most common mistakes entirely. The discipline of consistent content execution is what separates teams that build sustainable organic traffic from those that publish a lot and wonder why results are uneven.

For larger teams, assigning image quality ownership to a specific role, whether that is an SEO manager, a content editor, or a technical lead, creates accountability. Without ownership, it remains a shared responsibility that no one actually owns.

Automation helps at scale. Image compression plugins for WordPress like ShortPixel or Smush handle format conversion and compression automatically on upload. They do not write your alt text or name your files, but they remove the manual burden of compression entirely. That is a reasonable division of labour: automate what can be automated, and build human process around what requires judgement.

Periodic audits matter too. A crawl-based image audit every quarter, checking for missing alt text, oversized files, missing captions on key images, and broken image URLs, keeps the baseline quality high without requiring a full site review every time. The case for regular marketing audits applies equally to technical content quality. You cannot manage what you do not measure, and you cannot improve what you do not audit.

Image SEO sits within a broader set of technical and on-page decisions that collectively determine how well your organic strategy performs. If you want to understand how these pieces connect at the strategy level, the Complete SEO Strategy hub is the right place to see the full picture.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions