Foot Locker Image SEO Case Study: 228% Increase for Google Image Thumbnails

Most image SEO advice stops at alt text and file naming. This case study is useful precisely because it is about none of that. Foot Locker's images were crawlable, correctly labelled and served from a working CDN. They simply were not big enough to be worth indexing, and that one property was enough to keep image thumbnails off their listings while competitors collected them.

The finding, documented by Brodie Clark, is that Foot Locker's category pages carried product images with very small dimensions, 67x67 pixels, in the server rendered HTML. Google fetched those files. It largely declined to index them. The team raised the minimum to 200x200, which Clark notes was "the highest they were prepared to increase without impacting page size", and the thumbnail presence changed materially. The change went in on 20 July 2023.

Why a small image gets crawled and then dropped

Crawling and indexing are separate decisions for images just as they are for pages. Googlebot will happily fetch a 67x67 JPEG. Whether that file then enters the image index depends on whether it is judged useful, and a thumbnail that cannot be rendered at result size without falling apart is not useful.

Google does not publish a hard pixel minimum, and you should be suspicious of any article that quotes one as though it were documented. What Google's own image guidance does say is that images should be high quality and large enough to be displayed well. The practical read from this case study is a floor, not a formula: 67x67 was clearly under it, 200x200 cleared it for this retailer's templates.

Three distinctions decide whether your images are actually eligible, and teams routinely confuse them:

• Intrinsic file dimensions. The real pixel dimensions of the file the server returns. This is what matters most.
• HTML width and height attributes. What the markup declares. These should match the intrinsic size and are also what protects you from layout shift.
• CSS display size. The box the browser paints the image into. This is not the signal. An image displayed at 400px wide via CSS while the underlying file is 67px wide is still a 67px image.

Foot Locker's problem lived in the first two. The category template was generating genuinely small derivatives, and the markup declared them honestly. Anyone auditing by screenshot would have seen normal looking product tiles and missed it entirely.

How to audit your own templates for this

The audit is fast if you look at the delivered HTML rather than the rendered page. Fetch a category page as Googlebot would receive it and inspect the image nodes directly.

1. Pull the raw HTML with curl and read the <img> tags on a listing template. Do this for the server response, not the DevTools Elements panel, which shows the post JavaScript DOM.
2. Compare declared attributes with the real file. Request the image URL itself and check its actual dimensions. A CDN that resizes on a query string, for example ?w=67 or /67x67/ in the path, is the usual culprit.
3. Check the srcset candidates. If you serve responsive images, confirm the smallest candidate is not the one Google is most likely to take, and that a genuinely large candidate exists. A sizes attribute that resolves tiny on the crawler's viewport can starve the selection.
4. Confirm the images are reachable. Images blocked by robots.txt, including on a separate CDN hostname, cannot be indexed at all. Our disallowed image check covers how that failure presents.
5. Verify lazy loading is not hiding them. Native loading="lazy" with a real src is safe. A placeholder in src with the true URL parked in data-src and no noscript fallback is not.
6. Spot check indexation by searching for a distinctive product image in Google Images with a site: filter on your domain. Absence across a whole template is the signal you are looking for, not the absence of one file.

Reading the 228 percent honestly

This is where practitioners get burned when they take a case study to a stakeholder. The reported figure is a 228 percent increase in estimated organic visits per month to pages that now have image thumbnails, moving from roughly 60,000 to roughly 195,000. That is a measure of how much of Foot Locker's ranking footprint became thumbnail bearing. It is not a measurement that organic traffic rose 228 percent.

Two caveats follow from that, and both are worth stating in your own deck before someone else does:

• The traffic estimate is third party. Clark's analysis uses publicly accessible index data, benchmarked against JD Sports and Hype DC. Estimated visit figures from any index tool are directional, not audited analytics.
• Thumbnail presence is a visibility change, not a guaranteed click change. A thumbnail plausibly improves click through rate on a crowded mobile result, and Foot Locker overtook Hype DC on thumbnail presence in the Australian market, but attributing a specific CTR delta would require split testing that this study does not claim to have run.

Framed correctly, it is still a strong result. A template level change with a bounded performance cost moved a large share of a retailer's ranking inventory from ineligible to eligible. That is exactly the kind of intervention worth prioritising, because it is one engineering ticket applied across thousands of URLs.

Turning this into a repeatable standard

The durable lesson is to make image dimensions a template standard with an owner, rather than an incident you rediscover every eighteen months. Practical steps:

• Set a documented minimum for every image role in your design system: listing tile, product hero, editorial inline, logo. Write the number down and put it in the component definition.
• Add a crawl assertion. Configure your crawler to extract <img> width and height and flag any page where a content image declares under your minimum. This turns a one off audit into a regression test.
• Balance against page weight deliberately. Foot Locker's team picked the largest size they would accept, which is the right instinct. Serve modern formats and correct compression so the dimension increase does not become a Core Web Vitals regression.
• Keep an image sitemap or image entries in your existing sitemaps so discovery does not depend solely on in page links.

If you want the wider checklist that this sits inside, our complete image SEO guide covers formats, compression and markup, image SEO beyond alt text covers the technical delivery layer, and Google Images SEO covers how selection and ranking work once your files are actually eligible.

Frequently asked questions

The wider point is that image SEO failures are usually infrastructural. Nobody chose to serve 67 pixel product images; a resizing rule somewhere did, and no report surfaced it. Put a minimum in your design system, assert it in your crawl, and this stays fixed.

Source: https://brodieclark.com/foot-locker-image-seo/