How to audit an online shop without missing the basics

An online shop audit should answer a practical question, not produce a long list of warnings. We want to know whether search engines can reach the pages that matter, whether the right URLs are being indexed, whether category and product pages provide enough evidence of purpose and usefulness, and whether technical or usability problems are getting in the way.

A good ecommerce SEO audit also separates evidence types. A crawler finding, a Google Search Console report, a Core Web Vitals metric, and a drop in sales are not the same thing. If we mix them together, we end up buying fixes for the wrong problem. If you want a broader framework for checking evidence before you spend money, our guide to how to audit a site without buying the wrong fix is a useful companion.

What an online shop audit should prove

The purpose of an ecommerce SEO audit is not to prove that every page is perfect. It is to prove, or disprove, a smaller set of things that affect discovery.

For an online shop, the main questions are usually these:

• Can search engines reach category and product pages through normal crawl paths?
• Are the URLs we want in search actually indexable?
• Are low value or duplicate URLs being kept out of the index deliberately?
• Do key pages make their purpose clear to users and search engines?
• Are speed, layout, and accessibility issues making those pages harder to use?

That sounds simple, but the evidence comes from different places and should stay separate.

First, there is crawl evidence. This includes robots rules, internal links, XML sitemaps, status codes, canonicals, and what a crawler such as Googlebot can fetch. This tells us about access and discovery, not quality on its own.

Second, there is indexing evidence. In practice, this often comes from Google Search Console and especially the Page indexing report. This tells us what Google says happened to a URL or group of URLs. It does not automatically tell us whether the result is wrong. Google’s own documentation is explicit that excluded pages are not always a problem, and some should stay excluded (Google Page indexing report).

Third, there is page evidence. This is what a category or product page actually contains. Titles, headings, descriptive copy, links, images, and page structure all matter because they help users and search engines understand what the page is for (Google SEO Starter Guide).

Fourth, there is performance and usability evidence. Core Web Vitals, mobile rendering, contrast, and stable layouts can show where a page is awkward or frustrating to use, but they do not prove a ranking outcome on their own (web.dev Core Web Vitals, W3C accessibility design tips).

If we keep those evidence types separate, the audit stays useful. If not, a shop can end up rewriting copy to solve a crawl problem, or chasing speed scores when the real issue is that the main category pages are blocked.

Can search engines crawl the pages that matter?

Start with the basics. Can a search engine get from your home page or top navigation to the categories and products you care about?

Check robots.txt first. Look for rules that block key folders such as /category/, /shop/, or product paths. Blocking internal search results or basket and checkout URLs is often sensible. Blocking main commercial pages is not. Also remember that different crawlers may identify themselves differently. If you are reviewing AI-related access, separate Googlebot for Search from Google’s AI features guidance, which says existing Search fundamentals apply and no special AI text file or schema is required (Google AI features). Likewise, OAI-SearchBot and GPTBot are separate controls with separate purposes, and OpenAI documents that search access and model training controls are not the same choice (OpenAI crawler documentation).

Then test internal linking. A category page buried behind JavaScript interactions, faceted combinations, or site search may technically exist but still be hard to discover. In most shops, we expect important categories to be linked from primary navigation, subcategory listings, breadcrumbs, and related sections. Product pages should be reachable from their category pages without a long chain of clicks or reliance on on-site search.

Status codes come next. Important category and product pages should usually return 200 OK. Common problems include:

• 404 for discontinued products that still receive internal links
• 301 chains caused by URL changes or trailing slash inconsistencies
• 302 temporary redirects left in place permanently
• soft 404 behaviour where thin or out of stock pages load but offer little content

Canonicals need a separate check. A product page may be crawlable but canonicalised to a parent category, a variant, or a parameter-free version that strips useful distinctions. That can be correct in some setups, but it should be a deliberate choice. For categories with sorting or tracking parameters, a canonical to the clean category URL often makes sense. For products, a self-referencing canonical is common where the page is intended to stand on its own.

Review crawl paths, not just isolated URLs. Pick a few important categories and follow the route a crawler would take. Home page, category, subcategory, product. Are there plain HTML links? Are breadcrumbs present? Do pagination links exist where stock lists span multiple pages? Google’s starter guidance still emphasises that links help discovery, and descriptive anchor text helps users understand where they are going (Google SEO Starter Guide).

One caution here. A crawler tool can tell us that a URL exists and is linked, but not whether Googlebot spends crawl budget on it in the way we hope. For most small and mid-sized UK shops, the immediate question is simpler. Are we making it easy or hard for search engines to find our commercial pages? Usually, the basics answer that.

Which URLs should be indexed and which should not?

This is where many audits go wrong. They treat every excluded URL in Google Search Console as an error. That is not what the report means.

Open the Page indexing report in Google Search Console and review URL groups by type. For an online shop, split them into at least:

• category pages
• product pages
• filtered or faceted URLs
• internal search results
• duplicate or parameterised URLs
• utility pages such as basket, account, and checkout

Now ask a business question for each type. Do we actually want these pages in the index?

Category pages usually should be indexed if they represent real product groupings with search demand or navigational value. Product pages usually should too, unless the products are obsolete and have no replacement logic.

Filtered URLs are more nuanced. A filter for a meaningful product type or size range may deserve indexing if it has stable demand and unique value. A URL created only by sort order, stock state, or a combination of five filters usually does not. Internal search results generally should not be indexed. Duplicate URLs with session IDs, tracking parameters, or alternate sort orders usually should not be indexed either.

Use the report’s exclusion reasons carefully. “Excluded by ‘noindex’ tag” can be correct. “Alternate page with proper canonical tag” can be correct. “Crawled - currently not indexed” is not a verdict on quality by itself, but it is worth investigating for key products or categories. “Discovered - currently not indexed” can suggest scale, weak internal linking, or low perceived value. Google’s own documentation says to investigate the reason and not assume every exclusion is a problem (Google Page indexing report).

For products, compare indexed counts with your live catalogue, but do not expect a perfect one-to-one match. Variant handling, discontinued items, and duplicate URLs all affect the picture. What matters is whether commercially important products and categories are represented.

A useful method is to export a sample of excluded URLs and label them manually. Ten to twenty examples per exclusion reason can tell you far more than staring at totals. You may find, for example, that excluded parameter URLs are healthy housekeeping, while a whole branch of category pages is excluded because of an accidental noindex.

If you need help interpreting Search Console evidence without overclaiming from it, we have a more detailed piece on Search Console evidence and its limits.

Do category and product pages give enough evidence to rank?

Once crawlability and indexing are in order, we can look at whether the page itself gives enough evidence of purpose and usefulness.

Start with titles and headings. A category page title should identify the product group clearly, without relying on boilerplate alone. A product page title should identify the product and, where useful, the distinguishing attributes a buyer needs. This is established guidance in principle, not a formula. Google recommends descriptive titles and useful content, but it does not provide a guaranteed template for rankings (Google SEO Starter Guide).

Then check descriptive copy. Many shops either have too little, or they repeat manufacturer text across dozens of pages. A category page often needs a short introduction that explains what the range is, who it is for, or how to choose between options. A product page usually needs more than a stock line and a specification table. The aim is not to hit a word count. It is to make the page’s purpose and distinctions clear.

Page structure matters too. Ask whether the page has a clear main purpose:

• category pages that help people browse and compare
• product pages that help people decide and act
• supporting information such as delivery, returns, dimensions, compatibility, or materials where relevant

That is partly interpretation, but it follows established usability and accessibility principles. Clear headings, understandable labels, and predictable navigation help users make sense of a page (W3C accessibility design tips).

Internal anchor text is often overlooked. If every link to a category says “shop now” or “view all”, we lose useful context. More descriptive links from navigation, related categories, buying guides, and breadcrumbs help both users and discovery. Google’s starter guidance supports descriptive link text for this reason (Google SEO Starter Guide).

Images deserve a practical review. Are product images large enough to inspect? Do they show the item clearly? Do alt attributes describe the image usefully where needed, especially for product detail imagery? W3C guidance supports text alternatives and clear design, though alt text should describe the image’s function and content rather than repeat keywords (W3C accessibility design tips).

Watch for duplication risks. Reused manufacturer descriptions, near-identical variant pages, and repeated category intros can make it harder to show why one URL should be indexed over another. That does not mean every repeated phrase is a problem. Shops naturally reuse brand names, specifications, and policy text. The issue is whether key pages offer enough unique context to justify their existence.

This is also the point where people often overstate what an audit can prove. We can say a page gives weak or strong evidence of purpose. We cannot say that changing three headings will secure a ranking outcome.

Are speed, usability and accessibility blocking discovery or use?

Performance checks matter, but they need careful interpretation.

Core Web Vitals measure aspects of loading, responsiveness, and visual stability. They are useful because they describe real page behaviour, not because they predict a commercial result. A poor score can indicate that images are oversized, scripts are delaying interaction, or page elements shift as they load. A good score does not prove that a page will rank, and a bad score does not tell us why sales changed (web.dev Core Web Vitals).

For an online shop, review templates by page type. Home, category, product, basket. A product page with multiple review widgets, image galleries, and recommendation blocks often behaves very differently from a plain content page.

Mobile usability is still the practical baseline. Check whether navigation is readable, taps are easy, filters can be opened and closed without trapping the user, and key actions remain visible. These are not only conversion concerns. If pages are awkward to use, they are less likely to satisfy the people who land on them.

Accessibility checks overlap with usability. W3C guidance highlights readable contrast, clear controls, understandable navigation, and layouts that do not confuse users (W3C accessibility design tips). For shops, common failures include:

• pale text on white backgrounds in filters or product options
• menus that only work on hover
• accordion sections with unclear labels
• low contrast price or stock messages
• layout shifts that move the add-to-basket button while the page loads

These checks can show whether a page is harder to use than it should be. They cannot prove legal compliance, and they cannot prove rankings. They are still worth doing because they reveal friction that affects both discovery and use.

If you are also reviewing visibility in AI search interfaces, keep the evidence separate there too. Google’s AI features documentation says standard Search guidance still applies, and there is no special AI text file or schema required for those features (Google AI features). A citation in an AI answer, a search impression, and a visit from Googlebot are different events.

How to turn audit findings into a sensible action list

A useful audit ends with decisions, not just observations.

We recommend sorting findings by page type first. Group together category issues, product issues, indexation issues, and sitewide technical issues. That makes it easier to estimate impact. A broken canonical pattern across all products is usually more urgent than one weak product description.

Then prioritise by three factors.

First, business risk. Does the issue affect high value categories, core products, or a large proportion of the catalogue?

Second, certainty of evidence. An accidental noindex on revenue-driving categories is high certainty. A suspicion that copy is “not rich enough” is lower certainty and may need testing or editorial judgement.

Third, effort and dependency. Some fixes are quick, such as correcting a robots rule or restoring internal links. Others depend on platform changes, feed logic, or template redevelopment.

A simple action list might look like this:

1. Fix access blockers on key category and product templates.
2. Correct indexation signals for URLs that should be indexed.
3. Consolidate duplicates and parameter handling.
4. Improve category and product page clarity where evidence is weak.
5. Address template-level speed and usability issues.
6. Review long tail opportunities such as filter pages only after the basics are stable.

This order matters. There is little point polishing copy on pages that search engines cannot crawl or index properly.

Keep your recommendations tied to evidence. If a finding comes from Google Search Console, say so. If it comes from a crawl, say so. If it is an editorial interpretation, label it as judgement rather than fact. That discipline helps when several teams are involved, especially in businesses where marketing, development, and merchandising all touch the shop.

If you want a lighter process before commissioning a formal review, our checklist on checking a site before you buy an SEO audit and our practical site audit before buying tools both help you sort obvious faults from tool noise.

One final point for UK readers. Many ecommerce platforms and plugins are built for global or EU-wide defaults. Where there is a difference between UK practice and wider EU assumptions, treat legal and compliance questions separately from SEO. An audit can identify crawl, indexation, content, and usability issues. It should not pretend to settle tax, consumer law, or accessibility compliance questions without specialist review.

That is the basic shape of an ecommerce SEO audit that does not miss the essentials. Prove access first. Decide what should be indexed. Check whether key pages provide enough evidence of purpose. Then deal with speed and usability in proportion to what the evidence actually shows.