Ecommerce SEO Audit: a 40-Point Checklist
How to conduct an effective ecommerce SEO audit when your site is mostly templates and product data. Forty checks grouped by area, the tools we use to run them, plus a catalog section that tests AI readiness.
By Alberto Barberis, Founder and CEO, AndromedAI · Updated October 2026
The short answer
An ecommerce SEO audit is a structured review of an online store that checks whether search engines and AI agents can crawl and index its product and category pages, and whether they understand them. It covers crawling, indexing, site architecture, faceted navigation, product pages, category pages, structured data, Merchant Center feeds, Core Web Vitals, plus the completeness of the product data itself.
Ecommerce SEO audit in 30 seconds
An ecommerce SEO audit tells you whether Google and AI agents can reach every product and category page you sell, and whether they understand what's on it. Store problems repeat across thousands of URLs, so you audit by template and data field.
- 01
Most ecommerce SEO problems are catalog problems. One broken template or empty attribute repeats on every product that uses it.
- 02
Start with indexing. A product Google hasn't indexed can't rank, and no AI answer will cite it.
- 03
Filters and variants create more URLs on a store than on any other kind of site, and pagination adds more. Controlling them is most of the technical work.
- 04
Product data is the audit area that has grown fastest, because every channel that sells for you, from Google Shopping to ChatGPT, reads the same attributes and identifiers.
- 05
An audit is only useful if it ends in a ranked fix list with an owner per fix. Score 40 checks pass, partial or fail, and repeat every quarter.
What an ecommerce SEO audit covers, and why catalogs need their own
It reviews seven areas, from crawling and indexing to catalog data quality. A generic site audit stops after the first few and never looks at the product data Shopping and AI agents depend on.
increase in traffic to US retail sites from generative AI tools, 2025 holiday season
AI assistants are a real discovery channelSource: Adobebetter conversion for AI referrals than other traffic, 2025 holiday season
AI shoppers arrive ready to buySource: Adobe for Businessmore clicks on average for retailers who added correct GTINs
One data field, measurable liftSource: Google Merchant Center Helpof mobile ecommerce sites have mediocre or worse product page UX
Product pages hold most audit findingsSource: Baymard Instituteof online holiday transactions happened on a smartphone in 2025
Audit the mobile template firstSource: AdobeWhy a catalog site needs a different audit
A store is a few templates multiplied by thousands of SKUs. Forget an <a href> in the category template and every product below it drops out of the crawl. A filter that writes ?color=blue&size=m and ?size=m&color=blue as two URLs doubles the crawl. An empty material field in the PIM is missing from the page and the markup, and from the feed.
We'd ignore the health score most crawlers show first. It counts URLs with issues and says nothing about how many products Google indexed, which is the number we check first.
Google says AI Overviews and AI Mode have no extra requirements beyond being indexed and snippet-eligible (Google Search Central). So technical checks make you eligible. Product data decides whether yours gets recommended.
| Area | Generic SEO site audit | Ecommerce SEO audit |
|---|---|---|
| Crawling | Broken links, robots.txt, sitemap | Plus faceted URLs, sort parameters, internal search, pagination |
| Duplicates | Duplicate titles and pages | Variant URLs, shared manufacturer copy, filter combinations |
| Content | Keyword targeting on articles | Product titles, specs, reviews, returns info, demand-led category copy |
| Structured data | Organization, Article | Product, Offer, ProductGroup, shipping/returns policies, matched to the feed |
| Feeds | Not covered | Merchant Center diagnostics, ChatGPT product feed |
| AI readiness | Rarely covered | Attribute completeness, identifiers, intent coverage, shopping metadata |
Discovery problems
Pages that can't be crawled or aren't indexed. They block everything else, so they're fixed first.
Relevance problems
Indexed pages that don't use shoppers' words. They cap rankings and Shopping reach.
Readiness problems
Products whose data is too thin for rich results or for AI agents to compare.
The strategy behind each area lives in our ecommerce SEO guide. This page is the audit itself.
How to conduct an effective ecommerce SEO audit
Set a baseline, crawl the site and hold the crawl up against Google's index. Then review each template, sample the catalog data and rank every failed check by impact and effort. A first audit of a mid-size store takes one to three weeks; repeat audits take days.
- 1
Set the baseline
Export 12 months of organic clicks and revenue by page type. Count sellable SKUs and live categories: that's how many URLs should be indexed.
- 2
Crawl the site
Run Screaming Frog, Sitebulb or a similar crawler with JavaScript rendering switched on. Save status codes, canonicals, directives, depth, internal links.
- 3
Compare with the index
In Search Console under Indexing > Pages, put the indexed count next to your product count and read every exclusion reason.
- 4
Review the templates
Run checks 14 to 34 on five product and five category pages from your top categories. Mobile first.
- 5
Sample the catalog
Export 100 to 200 products and measure fill rates for
gtin,brand,color,size,materialandproduct_type(checks 35 to 40). - 6
Score and rank
Mark each check pass, partial or fail. Rank the failures by how many URLs they touch and the revenue on those URLs, then by effort.
- 7
Fix and re-test
Give every fix an owner. Re-crawl after each release and watch indexing and Merchant Center for 4 to 8 weeks.
Teams skip step three. In our experience it's where most findings come from: the gap between what your site links to and what Google kept.
Tools you need
| Tool | What it shows | Checks |
|---|---|---|
| Google Search Console | Page indexing report, Core Web Vitals, Merchant listings, Product snippets, URL Inspection | 1-7, 25-33 |
| Site crawler | Status codes, canonicals, depth, orphans, duplicate titles | 3-24 |
| Google Merchant Center | Disapprovals, warnings, feed versus page mismatches | 29, 35-39 |
| PageSpeed Insights | Field data for LCP, INP, CLS | 31-33 |
| Rich Results Test | Product and merchant listing markup | 25-30 |
| Catalog export | Attribute fill rates and description quality | 35-40 |
A spreadsheet with one row per check is enough for scoring.
How to prioritize audit findings
| Priority | Typical finding | Fix when |
|---|---|---|
| 1. Blocking | Products not indexed, noindex or robots.txt on product paths, disapproved feed items | This sprint |
| 2. Template-wide | Filter URL bloat, canonical errors, missing Product markup, slow LCP | Next 30 days |
| 3. Data-wide | Empty attributes, missing GTINs, manufacturer copy, vague titles | 60 to 90 days, by revenue tier |
| 4. Page-level | Thin individual categories, missing alt text, weak internal links | Ongoing backlog |
Many audit templates open with PageSpeed scores. We'd put them behind indexing every time: a slow product page can still rank, an unindexed one can't.
A check that fails on one template fails on every product built from it. Fix templates and data fields before you touch individual pages.
Audit
What is broken, and where?
40 checks scored by template and data field
Fix
What do we change first?
Ranked fix list with owners
Publish
Is the fix live everywhere?
Page and markup updated together with the feeds
Measure
Did it move?
Indexed products, clicks, Shopping reach, AI referrals
Run all 40 checks quarterly; check indexing and Merchant Center weekly.
Checks 1 to 13: crawling, indexing and site architecture
These checks confirm the pages you want found can be crawled and indexed, in a clean link structure. An unindexed page fails every later check by default.
Crawling and indexing (checks 1 to 7)
- 1. Indexed products match sellable products
In the Page indexing report, switch the filter to your product sitemap. 20,000 live products and 9,000 indexed means a discovery problem comes before any content work.
- 2. Exclusion reasons are understood
"Discovered, currently not indexed" usually means weak linking or crawl capacity. "Crawled, currently not indexed" points to thin or duplicate pages. A "Soft 404" is often an empty category.
- 3. robots.txt blocks only low-value URLs
Disallow cart, checkout, account, internal search, sort parameters. Keep product and category paths open, image and JS files too. Allow Googlebot and OAI-SearchBot.
- 4. XML sitemaps list only canonical 200 URLs
Split product and category sitemaps, keep each file under 50,000 URLs or 50MB, and make
lastmodaccurate. Nothing redirected, noindexed or filtered. - 5. Status codes are clean
No internal links pointing at redirect chains or 404s. Discontinued products redirect to the closest replacement or category, never all to the homepage.
- 6. Canonicals are consistent
Each product and category page canonicalizes to itself, and that URL matches the sitemap and your internal links. On Shopify, internal links should use
/products/ridge, not the collection path. - 7. Key content survives rendering
In URL Inspection, the rendered HTML shows the title, price, description, specs and related-product links without a click.
Google names these exclusion reasons in its Page indexing report documentation and caps each sitemap at 50,000 URLs or 50MB (Google). On the AI side, OpenAI documents that sites blocking OAI-SearchBot do not appear in ChatGPT search answers, while GPTBot only controls model training (OpenAI). We still find stores blocking both.
Site architecture, facets and URLs (checks 8 to 13)
- 8. Navigation uses crawlable links
Menus and breadcrumbs use
<a href>. A click handler on adivhides pages from crawlers. - 9. Every product is linked from a category
No product reachable only through search or the sitemap. Compare crawl and catalog export to find orphans.
- 10. Revenue pages sit close to the homepage
Top categories within two clicks, top products within three. Link bestsellers from the homepage.
- 11. Faceted navigation is controlled
Pick the filters that deserve indexable URLs and block the rest in robots.txt. Fix filter order; empty combinations return 404.
- 12. Pagination is crawlable
Every page has its own URL such as
?page=2, its own canonical and a plain link to the next page. Load more buttons need that link as a fallback. - 13. URLs are clean and stable
Descriptive words, lowercase, no session IDs. Variants get their own URL, with the base URL as canonical.
Google's faceted navigation guidance says to use robots.txt for filter URLs that do not need to rank, or to ensure "the logical order of the filters always stays the same and that no duplicate filters can exist".
The usual advice to canonicalize every filter URL to its parent category is weaker than it sounds. A canonical is a hint, and Google must still crawl each URL to read it. For filters that will never rank, we'd use robots.txt. Our guide to faceted navigation SEO explains how to pick the facets that deserve an indexable URL.
Google's pagination guidance adds that page one should not be the canonical for the whole series. Its URL structure guide recommends the URL without the variant parameter as canonical. Platform defaults often create these duplicates, so check yours: our WooCommerce SEO guide and Magento SEO guide cover the settings behind duplicate category and product URLs.
Checks 14 to 24: product pages and category pages
Product and category pages are where rankings and AI citations are won. These checks test whether each template uses shoppers' words and targets real demand.
Product pages (checks 14 to 20)
- 14. Title tags lead with what the product is
Product type and defining attribute first, brand after: "Women's Waterproof Leather Hiking Boot, Trailmark Ridge". No two products share a title.
- 15. Descriptions are unique
No manufacturer copy shared with other retailers. No blank or one-line descriptions. Spot-check with exact-match searches.
- 16. Specs are in text
Material, dimensions, weight, care: in HTML text or a spec table. Images and PDFs can repeat them.
- 17. Images are optimized
Descriptive file names and
alttext, several angles, an in-scale or on-model shot, WebP or AVIF with width and height set. - 18. Reviews are on the page
Review text is crawlable on the product URL, not trapped in an iframe. The rating sits near the title.
- 19. Shipping and returns are visible
Delivery time and cost appear near the buy button with the return policy.
- 20. Out-of-stock pages stay useful
Temporarily unavailable products keep their URL, show availability and link to alternatives.
Baymard's benchmark of 155+ sites found that 44% do not show the return policy on the product page, though 60% of users look for it there, and 37% provide no in-scale images (Baymard Institute).
On multi-brand retailers, check 15 is the one we see fail most, and how to write product descriptions covers the fix. To score single pages against checks like these, use our product page checklist for AI readiness; the full template is in product page optimization.
Category pages (checks 21 to 24)
- 21. Each category targets one real query
Map every category to a keyword with search volume. Two categories chasing "linen shirts men" split rankings.
- 22. Category copy helps the choice
A short intro above the grid. Below it, a buying guide or FAQ about fit and fabric. No keyword-stuffed blocks.
- 23. High-demand filters have landing pages
Searched combinations such as "waterproof hiking boots women" get a static URL with its own title and H1.
- 24. Breadcrumbs and cross-links connect categories
Visible breadcrumbs with BreadcrumbList markup, plus links between parent and sibling categories.
Building categories around real searches is covered in category page optimization.
Checks 25 to 34: structured data, Merchant Center, performance and international
Markup and feeds tell Google what each product is, what it costs and how it ships. Core Web Vitals and hreflang decide how well each market is served.
Structured data and Merchant Center (checks 25 to 30)
- 25. Every product page has Product and Offer markup
Required:
name,imageand anOfferwith apriceabove zero andpriceCurrency. Watch for a theme and an app both outputting it. - 26. Recommended properties are filled
gtin,brand,sku,color,material,size,availability, plusaggregateRatingwhere reviews exist. - 27. Shipping and returns are marked up
Define the policies once under Organization. Override per Offer only where needed.
- 28. Variants use ProductGroup
ProductGroupwithvariesBy,hasVariantandproductGroupID, and a uniqueskuorgtinper variant. - 29. Merchant Center is clean
No disapprovals, warnings reviewed, and feed price and availability match the landing page.
- 30. Markup is valid and visible
Zero errors in the Rich Results Test and the Merchant listings report. Every marked-up value is visible on the page.
Google lists the properties, including hasMerchantReturnPolicy and shippingDetails, in its merchant listing documentation, and its product variants guide requires a unique ID per variant. See structured data for ecommerce and our Google Merchant Center guide. Before you score check 29, run through the Google Merchant Center requirements checklist for account and policy issues.
Performance and international (checks 31 to 34)
- 31. LCP is 2.5 seconds or less
At the 75th percentile on mobile, for product and category templates. Never lazy-load the main product image.
- 32. INP is 200 milliseconds or less
Filters, variant selectors and add to cart respond quickly. Start with third-party scripts: review apps, chat widgets, tag manager containers.
- 33. CLS is 0.1 or less
Images and banners reserve their space. Review widgets and promo bars never push the buy button down.
- 34. hreflang is complete for every market
Each localized page lists itself and every alternate, with codes such as
en-gbandit-it, return links andx-default.
A page passes Core Web Vitals only when all three metrics meet the target at the 75th percentile (web.dev). For hreflang, Google warns that "if two pages don't both point to each other, the tags will be ignored" (Google Search Central).
Multi-market stores should localize titles too; a translated size chart in the wrong units still fails. See ecommerce localization.
Checks 35 to 40: catalog data and AI readiness
The last six checks audit the product data itself: completeness, identifiers, keyword and intent coverage, consistency across channels. Google Shopping and ChatGPT both read this data to decide what to recommend.
- 35. Attributes are complete
Fill rates of 95% or more for the attributes shoppers filter on:
color,size,material,gender,age_group,patternplus category specifics. - 36. Every product has identifiers
Correct
gtin,brandandmpneverywhere the product appears: page, markup, feed. No "Generic" brands, no GTIN shared by every size. - 37. Titles and descriptions cover demand keywords
Searched words such as "waterproof", "wide fit" or "for flat feet" appear in the copy, checked against keyword data.
- 38. Use cases and intents are answered
Each product says who it's for and when to use it, and how it compares, so it can match a query like "boots for a rainy city commute".
- 39. Shopping metadata is rich
google_product_category2 to 3 levels deep, your ownproduct_type, plusproduct_highlight,product_detailandquestion_and_answer. - 40. Every channel shows the same data
Title, price, availability and attributes match across the page, the markup, Merchant Center and any ChatGPT product feed.
Most audit checklists stop at check 34. We think that's the biggest blind spot in ecommerce SEO today.
What the data checks look like in practice
Take an illustrative product, the Trailmark Ridge boot. Its page title is "Ridge Boot Brown". The feed has no gtin, material is empty, color reads "BRN-02" and the description is two lines from the manufacturer. A classic technical audit passes it.
A shopper asks an AI assistant for waterproof leather hiking boots for wide feet under $200. The Ridge boot states none of those facts.
ChatGPTGoogle's spec leaves room to fix this: title up to 150 characters, description up to 5,000, up to 100 product_highlight entries and 30 question_and_answer pairs (Merchant Center product data specification). AI-generated titles go in structured_title (Google Merchant Center Help). OpenAI's ChatGPT product feed spec uses the same core fields and limits.
Technical checks make a product eligible to be found; data checks decide whether it gets recommended. That's the core of Agentic Commerce Optimization, because search engines, shopping ads and AI agents all read the same catalog. To track results, see how to measure AI visibility.
Ecommerce SEO audit with AndromedAI
AndromedAI covers the catalog side: it scores product data, then fixes content and attribute failures at scale and publishes to your store and feeds. Technical checks such as 1 to 13 and 31 to 33 stay with your development team.
How the platform maps to the checklist
| Checks | AndromedAI | What it does |
|---|---|---|
| 35-39 | AI Readiness Audit | Scores product pages on the four AI Readiness dimensions and shows which products to fix first |
| 14-16, 35-38 | Optimizer | Rewrites titles and descriptions, bullets, FAQ; extracts attributes; adds use cases and intents |
| 15, 16 | Creator | Creates complete product pages from brand or supplier data, including PDF spec sheets, in 12 languages |
| 21-23 | Category page optimizer | Builds category pages around real search demand |
| 29, 40 | Integrations | Publishes to Shopify, Google Merchant Center, Salesforce Commerce Cloud, Adobe Commerce, Shopware, Akeneo, Plytix, WooCommerce |
The Ridge boot after an Optimizer pass (illustrative scores):
page_title
BeforeRidge Boot Brown
AfterWomen's Waterproof Leather Hiking Boot, Wide Fit, Trailmark Ridge in Walnut Brown
description
BeforeDurable boot for the outdoors. Leather upper.
AfterFull-grain leather hiking boot with a waterproof membrane and a wide 2E fit, built for muddy trails and wet city commutes.
attributes
Beforecolor: BRN-02
Aftercolor: walnut brown; material: full-grain leather; width: wide (2E); waterproof: yes; activity: hiking, commuting
product_highlight
Before(empty)
AfterWaterproof membrane keeps feet dry on wet trails; Wide 2E last for broad feet
Every rewrite follows your Brand Kit (tone of voice, rules, examples, glossary, banned words), and approval workflows auto-approve content above an AI Checker score of 4.0. AndromedAI also generates Google Merchant Center conversational attributes for check 39. More than 500 catalogs have been optimized on the platform.
organic traffic, sales +46.9%, 483 hours saved
Semprefarmaciaclicks from AI chats
Bomboogieadd-to-cart in two months
Altaforma Milanobounce rate
Ausiliumhours saved monthly in catalog management
Global MarkFAQ
A structured review of an online store that checks whether search engines and AI agents can crawl and index its product and category pages, and whether they understand them. It covers technical SEO, faceted navigation, on-page content, structured data, Merchant Center feeds, page speed, plus the quality of the product data itself.
Start with a baseline from Search Console and analytics. Crawl with JavaScript rendering, then compare the crawl with the Page indexing report. Review product and category templates on mobile, sample 100 to 200 products for data quality, then rank failures by affected URLs and revenue, then effort. A first audit of a mid-size store takes one to three weeks.
Run the full 40-point checklist every quarter, and again after any replatforming, redesign or large catalog import. Between audits, check the Page indexing report and Merchant Center diagnostics every week. On a store, one template bug or feed error can affect thousands of products before anyone notices a drop in traffic.
You need Google Search Console and Google Merchant Center for indexing and feed data, plus a crawler such as Screaming Frog or Sitebulb with JavaScript rendering. Add PageSpeed Insights for Core Web Vitals and the Rich Results Test for markup. A catalog export from your PIM, platform or feed covers the product data checks.
The two most common findings are uncontrolled faceted navigation and incomplete product data. Filters create duplicate URLs that waste crawling and dilute rankings. Empty attributes, missing GTINs and manufacturer copy reused by other retailers limit rankings and Shopping reach across the whole catalog, because the same gap repeats on every product.
A modern one should. Google says AI Overviews and AI Mode need no special markup, so the technical checks decide whether a page is eligible. Whether a product actually gets recommended depends on its data: complete attributes and identifiers, coverage of shopper keywords and intents, and rich shopping metadata in the feed.
Glossary
- Faceted navigation
- Category filters such as color, size or price that create a new URL for each combination.
- Canonical URL
- The preferred version of a page, declared with rel canonical. Search engines treat it as a hint.
- Page indexing report
- The Search Console report that shows which URLs Google has indexed and why the others are excluded.
- ProductGroup
- The schema.org type that groups product variants, using variesBy and hasVariant.
- Core Web Vitals
- Google's page experience metrics: Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift.
- hreflang
- An annotation that tells search engines which language and regional versions of a page exist.
- GTIN
- Global Trade Item Number, the barcode identifier of a product, such as an EAN or UPC.
- AI Readiness Score
- AndromedAI's score of how ready a product page is for AI shopping agents, built on Product Data Completeness, Keyword Coverage, Customer Intent Match and Shopping Metadata.
Keep reading
The strategy behind every check in this audit
GuideFaceted Navigation SEO: Filters, Crawl Budget and Indexable FacetsGo deeper on check 11, the most common technical failure
GuideProduct Page Checklist for AI ReadinessScore single product pages against checks 14 to 20 and 35 to 40
GuideGoogle Merchant Center Requirements ChecklistClear feed and account issues before you score check 29
Sources (19)
- Adobe: Holiday shopping season drove a record $257.8 billion online
- Adobe for Business: AI traffic surges across industries, retail sees biggest gains
- McKinsey: New front door to the internet, winning in the age of AI search
- Baymard Institute: The current state of ecommerce product page UX
- Google Search Central: AI features and your website
- Google Search Central: Managing crawling of faceted navigation URLs
- Google Search Central: Pagination and incremental page loading
- Google Search Central: Designing a URL structure for ecommerce sites
- Google Search Central: Build and submit a sitemap
- Google Search Central: Merchant listing structured data
- Google Search Central: Product variant structured data
- Google Search Central: Localized versions of your page
- Search Console Help: Page indexing report
- Google Merchant Center Help: Product data specification
- Google Merchant Center Help: Tips to optimize your product data
- Google Merchant Center Help: Title and structured title
- web.dev: Web Vitals
- OpenAI: Overview of OpenAI crawlers
- OpenAI: ChatGPT product feed specification
Featured in
