The ecommerce SEO checklist for 2026, ranked by how often stores fail

How this checklist was built

This list is the 17 checks the 2026 study ran across 963 stores, read from the study's published data when this page loads. Each line states the share of audited stores that failed the check, the number of stores it is measured against, and the free tool that runs the same check on your store. The order is the failure share, highest first, computed rather than fixed.

Two things about the method shape how to read it. The study read twenty pages per store from outside the network that serves them in one crawl in September 2026, the way a crawler sees them, so a store passes only when its public pages showed no failure. The sample is stores in the Tranco top one million that sell on Shopify or WooCommerce, and it is one reading on one date, not a trend. A high share means the check catches many stores. It never means your store has the problem.

The checklist, worst first

1. Enforce DMARC, not just monitor it 75% of 963 audited stores do not enforce DMARC (no record, or a monitoring-only p=none policy). Check email authentication.
2. Write alt text for images 48% of 963 audited stores have images with no alt text on the pages checked. Check image alt text.
3. Deal with fully out-of-stock products 46% of 928 audited stores have at least one product that is fully out of stock in every variant. Check availability.
4. Compress images over the weight budget 45% of 963 audited stores serve at least one image over 250 KB on the pages checked. Check image weight.
5. Serve modern image formats 41% of 963 audited stores serve images only in legacy formats, with no WebP or AVIF. Check image weight.
6. Link a shipping policy 36% of 948 audited stores link no shipping policy page that we could find. Check trust pages.
7. Link a returns policy 31% of 948 audited stores link no returns or refund policy page that we could find. Check trust pages.
8. Publish an llms.txt file 21% of 963 audited stores publish no llms.txt file for AI assistants. Check llms.txt.
9. Publish a DMARC record 19% of 963 audited stores publish no DMARC record at all, so anyone can send email that claims to come from them. Check email authentication.
10. Give every product a valid price 13% of 928 audited stores list at least one product with no valid price in its public catalogue. Check product prices.
11. Publish an SPF record 12% of 963 audited stores have no SPF record, so their order and newsletter email is more likely to land in spam. Check email authentication.
12. Return a real 404, not a soft 404 8.3% of 963 audited stores serve a page that answers 200 but reads like "not found" (a soft 404). Check soft 404s.
13. Serve valid JSON-LD 8.0% of 963 audited stores serve at least one page with a JSON-LD block that is not valid JSON (for example a trailing comma or a raw line break in a text value). Check structured data.
14. Fix dead internal links 6.5% of 963 audited stores have dead internal links among the links checked. Check broken links.
15. Take noindex pages out of the sitemap 6.1% of 963 audited stores list a noindex page in their sitemap, sending contradictory signals. Check index signals.
16. Allow the AI crawlers you want citations from 2.9% of 963 audited stores block at least one AI crawler in robots.txt. Check AI crawler access.
17. Decide the training crawler on purpose 1.7% of 963 audited stores block one specific training crawler. Check AI crawler access.

The llms.txt line is the one counted the other way: it shows the stores that did not publish the file, since publishing it is the pass. Every other line shows the stores that failed as the metric defines failure.

Why the order matters

The order is the useful part. The checks at the top are the ones most stores failed, so they are the cheapest place to look first. A store that enforces its email authentication, deals with its fully out-of-stock pages, and cuts image weight and adds alt text covers four of the largest shares in the sample, and each of those is a fix that applies across the catalogue rather than to one page.

Rank is not severity, though. A rare failure can hurt more than a common one. Invalid JSON-LD removes a page's structured data from the systems that read it, and a noindex page listed in the sitemap sends a contradictory signal about a page the store wanted indexed. Those sit lower only because fewer stores failed them. Use the order to sequence the work, not to dismiss the bottom of the list.

Where to start on your store

Run the full audit on your store first. It crawls the public pages the way the study did and runs every check that fits, so you can see which lines of this list apply to you and in what quantity. Then work down the list, opening the tool on each line that matched rather than fixing blind.

A few honest limits. The checklist reads public pages only. Anything that lives in an account or a feed, such as shipping and returns values set account-wide, is outside it, as is stock hidden in an admin. A check that passes today can fail after the next theme or plugin change, which is why the last habit on the list is to re-run the audit rather than trust one clean pass.

The full method, the sample and the intervals behind every line are in The State of Ecommerce SEO 2026, and the same numbers with their bases are on the statistics page.

Questions

Keep reading