Guides
Ecommerce SEO audit: the checklist that fits a store
A store is not a blog with products. The URL set changes with stock, filters multiply pages nobody asked for, and the same description is syndicated to every retailer. This checklist audits the templates that generate pages, then verifies the fix on the URLs that earn demand.
By the Visdenda research team · Reviewed by the Visdenda editorial team
Published 2026-11-18 · Updated 2026-11-18 · 9 min read
Why a store audit is different
Three properties make ecommerce URLs harder than content URLs. The set is generated by templates, so one fault repeats hundreds of times. The page count moves with stock and filters, so last month's crawl is not this month's site. And the copy is often syndicated from the manufacturer, so the page can be technically perfect and still duplicate every other retailer selling the same item.
The audit therefore starts with the templates that generate URLs, not with a sorted list of severity. A canonical fault on a product template is one fix; the same fault on 800 products is one fix with 800 affected URLs.
What crawlers may fetch — and what they should not
- Cart, checkout, account and search. Keep them out of the index with robots.txt or noindex, and make sure the rule is live: a blocked page that still appears in the sitemap contradicts itself.
- Facets. Decide which filter combinations have demand and which are crawl traps. Canonical, robots and internal links must agree on the same answer — a filter page that is linked, indexed and canonicalised elsewhere is a three-way conflict.
- Pagination. Google no longer uses rel=next/prev as an indexing signal. Infinite scroll needs real links to the next page, and page 2+ should stay indexable when the products are.
- Parameters. Sorting and tracking parameters should not create new canonical URLs; check what the crawl sees rather than what the template intends.
The sitemap indexability check is the fastest way to catch the contradiction: a URL submitted in the sitemap while noindexed, robots-blocked or canonicalised elsewhere.
Audit templates, not URLs
Group the crawl by the template that produced each URL: product, category, brand, filter, search, CMS page. Then read findings per group. A title pattern that repeats across a template is a code fix; a one-off missing H1 on a hand-made landing page is editorial. Treating them the same is how an audit turns into a long list nobody finishes.
The cloud crawler keeps the page set stable between runs so the template groups stay comparable; a compare after the fix shows added, fixed and regressed findings per URL rather than a fresh snapshot.
Product data and structured data must agree
Product markup is a claim about the page: price, currency, availability, condition and reviews. When the marked-up price differs from the visible one, the rich result is at risk and the page is telling two stories. Audit the markup against the rendered page, then keep the template honest when prices change on a schedule.
- Product schema. One Product per page, offers with price, priceCurrency and availability; aggregateRating only when real reviews exist.
- Variants. Either one page per variant with its own offer, or a parent page whose offer describes the default — not a mix of both.
- Breadcrumbs. Category paths in markup should match the visible breadcrumb; a mismatch is a small but repeated inconsistency.
The product structured data explainer lists the fields that fail most often; the JSON-LD validator checks one URL in a run.
Duplicates, thin pages and stock
- Syndicated descriptions. The manufacturer copy is on every retailer. Add the specification, fit or comparison only you can write, or expect the category page to outrank the product.
- Thin filter pages. A facet with three products and no copy is not a landing page; either enrich it or keep it out of the index.
- Out of stock. Keep the URL and report availability while it is temporary; 410 or 301 when it is permanent. Re-check the template after a stock cycle, because that is when 404s appear.
- Duplicate titles and metas. Generated titles that differ only by SKU are near-duplicates; a template should vary by product attribute that people actually search.
The 12-check ecommerce SEO audit
| Check | Evidence that confirms it |
|---|---|
| Cart, checkout and account excluded | robots.txt or noindex live on the URL |
| Search and internal-result pages excluded | No indexable search URLs in the crawl |
| Facet policy is one answer | Canonical, robots and internal links agree |
| Pagination keeps products reachable | A plain <a> path to page 2+ |
| One canonical per product family | Self-canonical or parent canonical, not both |
| Titles differ by attribute | Template output across a 20-product sample |
| Descriptions are not syndicated | Unique copy on the money products |
| Product schema matches the page | Marked-up price and availability equal visible |
| Out-of-stock handled by policy | 200 with availability, or 410/301 if gone |
| Sitemap lists canonical, indexable URLs | Cross-check each entry against headers |
| Internal links reach money pages | Depth and orphan check on the crawl graph |
| Index coverage matches the catalogue | Search Console pages report, not a guess |
Verify after the fix
A template fix touches hundreds of URLs, so the verification is a re-crawl of the same page set, not a spot check. Compare the findings run over run, then read the search side: did the affected template earn more clicks at the same average position, or did impressions move without clicks? A change that only moved impressions is not a win yet.
If the fix was a title, meta, H1 or structured-data change, the verification method applies: a complete window, a baseline and same-site controls. Fewer than 100 impressions in either window returns insufficient — a no-conclusion is a result, not a failure.
Frequently asked questions
What is an ecommerce SEO audit?
It is a crawl-and-evidence review of the templates that generate a store: product, category, filter, brand and search pages. It checks what crawlers may fetch, which URL each template canonicalises to, whether product data matches the page, and whether the fixes changed search or traffic. It is not a keyword report and not a revenue forecast.
How many pages should I crawl?
Enough to cover every template with a real sample: the top categories, a spread of products including variants and out-of-stock items, plus one example of each filter family. A fixed 500-URL cap is a desktop-tool limit, not an audit design; the page set should follow the templates you need to judge.
Should product variants be canonicalised to the parent?
Only when the variant has no demand of its own. If people search the colour or size, a self-canonical variant with unique copy can earn the query; if they do not, canonicalising variants to the parent prevents near-duplicate pages. Decide per family, then check the canonical the crawl actually sees.
Do out-of-stock products hurt SEO?
Only when the URL is handled badly. A temporarily out-of-stock product should keep its URL and return 200 with availability in the product data; a permanently discontinued one should 410 or 301 to the closest alternative. Mass 404s during a stock-out are the mistake, not the stock-out itself.