Bulk Image Optimizer attempted a preselected sample of 50 well-known US websites. Thirty-eight popular websites produced a usable sample, with 1,250 homepage image records after within-site deduplication. We recorded filenames and alt attributes in a one-day snapshot.
This is an observational sample, not a representative survey or a ranking study. We measured no search rankings, traffic outcomes, image-search visibility, or screen-reader experience. The complete study and its limitations are on Bulk Image Optimizer.
What stood out
- The mean of the 38 site-level shares was 49% descriptive filenames and 22% opaque IDs, hashes, or numbers. These are averages of per-site percentages; a site with more images does not get more weight.
- Three of the 1,250 records used camera-default names. The sample also included many non-camera ID-style filenames.
- The sampled recipe sites were consistent: Serious Eats, Pinch of Yum, and Minimalist Baker each had descriptive filenames for every sampled record; Allrecipes had 94%.
- Opaque names were common on some large platforms in this sample: 97% of sampled Amazon images, 67% of IKEA’s, and all sampled Shopify.com images.
- The mean per-site share with alt text of three or more words was 59%; empty alt attributes averaged 29%, and missing alt averaged 4%. Empty alt can be correct for decorative images. Word count does not measure whether text is useful or accurate.
Each site contributes one share to the mean, regardless of how many images were captured on its homepage.
The 3+ word bucket measures length only. This was not a quality or screen-reader audit.
Site-by-site snapshot
Percentages below are within-site shares. “IDs” includes hashes and numbers. “3+ word alt” describes the number of words in the HTML alt attribute, not its quality. The full record and definitions are in the original study.
| Site | Descriptive filenames | IDs, hashes, or numbers | Alt with 3+ words | Images | Type |
|---|---|---|---|---|---|
| mailchimp.com | 100% | 0% | 0% | 6 | Software |
| minimalistbaker.com | 100% | 0% | 95% | 21 | Publisher |
| pinchofyum.com | 100% | 0% | 87% | 15 | Publisher |
| seriouseats.com | 100% | 0% | 94% | 16 | Publisher |
| allrecipes.com | 94% | 6% | 100% | 34 | Publisher |
| wired.com | 83% | 0% | 38% | 40 | Publisher |
| allbirds.com | 82% | 0% | 53% | 17 | Brand |
| outdoorvoices.com | 77% | 17% | 63% | 86 | Brand |
| brooklinen.com | 75% | 3% | 95% | 40 | Brand |
| casper.com | 74% | 11% | 93% | 27 | Brand |
| bonappetit.com | 69% | 10% | 92% | 39 | Publisher |
| glossier.com | 69% | 8% | 31% | 13 | Brand |
| costco.com | 67% | 0% | 70% | 46 | Retail |
| canva.com | 58% | 29% | 48% | 31 | Software |
| ulta.com | 57% | 32% | 0% | 101 | Retail |
| healthline.com | 56% | 12% | 52% | 25 | Publisher |
| apple.com | 55% | 36% | 64% | 11 | Retail |
| lowes.com | 55% | 36% | 98% | 44 | Retail |
| bombas.com | 46% | 8% | 0% | 13 | Brand |
| theverge.com | 46% | 38% | 100% | 13 | Publisher |
| thespruce.com | 43% | 4% | 91% | 23 | Publisher |
| hubspot.com | 43% | 0% | 43% | 7 | Software |
| cnn.com | 41% | 9% | 62% | 34 | Publisher |
| walmart.com | 41% | 59% | 100% | 17 | Retail |
| rothys.com | 35% | 0% | 5% | 20 | Brand |
| rei.com | 33% | 37% | 100% | 27 | Retail |
| squarespace.com | 33% | 0% | 0% | 18 | Software |
| ikea.com | 32% | 67% | 74% | 76 | Retail |
| kohls.com | 26% | 0% | 26% | 53 | Retail |
| everlane.com | 25% | 0% | 6% | 16 | Brand |
| sephora.com | 23% | 0% | 100% | 13 | Retail |
| nordstrom.com | 10% | 47% | 88% | 68 | Retail |
| nerdwallet.com | 10% | 0% | 80% | 10 | Publisher |
| warbyparker.com | 4% | 83% | 48% | 23 | Brand |
| amazon.com | 1% | 97% | 13% | 102 | Retail |
| buzzfeed.com | 0% | 77% | 95% | 66 | Publisher |
| shopify.com | 0% | 100% | 23% | 30 | Software |
| target.com | 0% | 0% | 0% | 9 | Retail |
How we collected the sample
We chose 50 sites before collection: 20 large retailers, 10 direct-to-consumer brands, 15 publishers, and 5 software companies. Each homepage was loaded once in desktop Chrome at 1440 × 900 on October 25, 2026, between 23:08 and 23:21 UTC. We scrolled to the bottom for lazy-loaded images and recorded raster images displayed at least 100 × 100 pixels. Within each site, records with the same normalized filename stem and alt-text string were collapsed; image pixels were not compared.
Twelve attempts were excluded: eight sites returned HTTP 403 or 429, Away failed to load, and Best Buy, Macy’s, and Nike each yielded fewer than five records. The excluded sites were Best Buy, Home Depot, Macy’s, Wayfair, Etsy, eBay, Nike, Chewy, Away, The New York Times, Smitten Kitchen, and Apartment Therapy. The 38 remaining sites contributed 1,250 included records.
For filename labels, we removed extensions and recognized CDN size suffixes, split names into alphabetic tokens, and excluded common layout and campaign words such as “hero,” “banner,” and “offer.” Two or more qualifying non-generic words counted as descriptive; one counted as one-word. This is our rule-based heuristic, not a Google score. We did not independently hand-label a validation sample and make no classifier-accuracy claim.
Headline and category percentages are unweighted averages of within-site shares. Alt buckets describe whether an HTML attribute is present and its word count; they do not judge relevance, accuracy, or purpose. This purposive, one-day homepage sample is not representative of all websites.
Practical use
For new uploads, a short filename that identifies the image is easy to add before the file goes live. Alt text should describe an image’s purpose; decorative images may correctly use an empty alt attribute. When renaming an image that is already live, update every reference or redirect the old URL to avoid broken images.
Our bulk image renamer can create descriptive filenames and editable alt text in the browser. For broader guidance on formats, sizes, and sitemaps, see the image SEO guide. Bulk Image Optimizer provides browser-based image tools; files stay on the device unless the visitor chooses to create a share link.
Google’s published guidance covers image SEO best practices and URL structure. The full study, including its methodology and limitations, is available here.
Bulk Image Optimizer creates browser-based image tools and practical guides for working with digital images.
Top comments (0)