Here's the uncomfortable truth about most ecommerce SEO audit checklists: they were built for content sites and retrofitted for stores. The result is a list of 90 generic checkboxes - 'check robots.txt,' 'fix 404s,' 'add meta descriptions' - that keeps SEOs busy without actually moving revenue.
When I first started running ecommerce audits, I used those same templates. And the results were predictably underwhelming. The technical scores improved. The rankings didn't. The reason? Ecommerce SEO has fundamentally different failure modes than content SEO.
A blog has maybe 200 pages. A mid-size ecommerce store can have 50,000 URLs generated overnight by filters, sorting options, and session parameters. A blog competes on informational intent. A store competes on commercial and transactional intent - where Google's quality signals are entirely different.
This guide is built from the ground up for stores. We developed a proprietary audit structure we call the 7-Layer Audit System - a sequenced, depth-first framework that surfaces the issues that actually suppress revenue.
We're going to cover Faceted navigation is the single biggest crawl budget killer, duplicate product content, commercial intent cannibalization, Silo-to-Shelf internal linking, and the often-ignored revenue architecture layer that most technical audits never touch.
If you run an ecommerce store - or you audit them - this is the checklist that will change how you think about the work.
Key Takeaways
- 1Every finding should cite evidence, state a pass or fail condition, assign severity, name an owner, prescribe a correction, and define validation
- 2The 7-stage review covers crawl access, intent, product content, internal links, performance, structured data, and commercial prioritization
- 3Faceted navigation should be tested by URL pattern, demand, indexation, crawl behavior, canonicalization, and internal linking before any blanket directive
- 4Category and product overlap should be confirmed with query and landing-page evidence before consolidation or retargeting
- 5Manufacturer copy requires measured similarity and product-level prioritization rather than an unsupported sitewide duplicate-content claim
- 6Internal-link findings should identify the source page, destination page, anchor, crawl path, commercial role, and corrected route
- 7Core Web Vitals findings need field and template evidence, a responsible technical owner, and post-release validation
- 8The final output should be a sequenced implementation sprint with blockers, dependencies, owners, validation dates, and commercial context
- 9Structured-data checks should validate rendered values against visible price, availability, rating, product, and breadcrumb information
- 10Competitor comparison should isolate a controllable difference and avoid treating correlation as proof of why another page ranks
1Stage 1: Can Search Engines Reach and Index the Right URLs?
Evidence required: export crawl results, index reports, sitemaps, canonical targets, robots directives, rendered links, server responses, and representative log data where available. Record the crawl mode, user agent, JavaScript setting, date, and URL samples so another reviewer can reproduce the finding. A 500-product store may be reviewed differently from a 10,000 SKU catalog, but size alone does not prove crawl waste.
Pass condition: priority categories and products are discoverable through internal links, return the intended status, declare consistent canonicals, and appear in the intended index state. Non-search pages and duplicate URL patterns have a documented control based on demand and function.
Fail condition: priority URLs are blocked, orphaned, redirected incorrectly, canonicalized elsewhere without justification, or crowded out by uncontrolled URL patterns.
Test inflation patterns separately. A category with 3 filters and 10 choices can create up to 1,000 combinations, but the audit should count actual generated and linked URLs rather than assume the theoretical maximum exists.
Review session and tracking parameters, internal-search URLs, sorted views, pagination, and filter combinations. Page 47 should be inspected against the real sequence, including cases where only 12 useful pages exist.
Severity: Critical for blocked or deindexed revenue pages, incorrect canonicals on priority templates, or widespread infinite URL generation. High for large index drift or crawl concentration on non-search pages.
Medium for isolated waste without material coverage loss. Owner: technical SEO with engineering, platform, merchandising, or infrastructure support.
Corrective action: choose the control that matches the evidence, such as removing crawlable links, consolidating templates, correcting canonical logic, changing index directives, fixing status codes, or revising robots rules.
Do not use robots.txt to solve an index problem without understanding the current indexed state. Validation: rerun the same crawl, inspect rendered directives, test representative URLs, compare index reports over the appropriate recrawl period, and confirm that priority pages remain accessible.
2Stage 2: Does Each Priority Page Match a Distinct Buyer Intent?
Evidence required: Google Search Console queries by landing page, paid search terms where available, internal-site-search data, page titles, headings, product or category scope, conversion behavior, and the current result types for target queries. Build a reproducible sample of the top 20-50 priority pages rather than reviewing pages only because they rank.
Pass condition: every priority category, collection, and product page has a documented primary intent, a distinct role, and content that matches the buyer decision represented in the search results. Fail condition: one page attracts the wrong intent, several pages compete for the same query cluster without distinct roles, or the selected page type conflicts with the dominant result type.
Record the matrix in order:
- current query cluster,
- intended commercial query cluster,
- intent classification, and
- corrective decision.
The decision may be to reorient the page, separate informational support, consolidate overlap, change internal links, or keep both pages with clearer roles. Product pages should not be expanded into comparison pages unless the product can genuinely answer that decision.
Severity: Critical when cannibalization or intent mismatch affects a major category or revenue page. High when high-impression pages attract materially wrong traffic. Medium when the issue is limited and does not displace a priority page. Owner: SEO strategist with merchandising, content, and analytics support.
Corrective action: change the target page, title, headings, copy, product set, internal anchors, or supporting content only after the intended role is approved. Validation: compare the same query clusters, page selection, click behavior, and conversion quality after recrawl and sufficient search exposure. Do not attribute a ranking change to the edit without considering other material changes.
3Stage 3: Are Product Pages Accurate, Distinct, and Complete?
Evidence required: a full product URL export, organic traffic, revenue, stock state, product type, manufacturer-copy similarity, rendered content, structured data, variants, customer questions, returns or support themes, and product-data ownership. Do not infer a sitewide problem from a few pages.
Use three evidence tiers while preserving the source sequence. Tier 1 covers high-revenue or high-traffic products that justify full manual review. Tier 2 covers medium or emerging products that need targeted differentiation and completeness. Tier 3 covers low-demand or low-value records that may need scalable templates, consolidation, or index controls. The tier is a prioritization device, not a claim about page quality.
Pass condition: the page accurately identifies the product and variant, contains useful original information for the intended buyer, answers material pre-purchase questions, and matches the visible offer and structured data.
Fail condition: the page copies supplier text without store-specific value, conflicts with product data, omits decisive information, or creates near-duplicate variant pages without a documented search role.
Execute in order:
- export product URLs with traffic and revenue,
- assign Tiers 1-3,
- run similarity and completeness checks for Tier 1 and Tier 2, and
- review title, description, visible facts, offer, and markup.
A similarity score is evidence for review, not proof of a ranking cause.
Severity: Critical for inaccurate price, stock, safety, compatibility, or identity data. High for widespread duplication or incomplete priority pages. Medium for low-priority thin records with controlled indexation. Owner: merchandising and product content, supported by SEO and engineering.
Corrective action: rewrite from verified product evidence, merge variants where they represent one offer, preserve separate pages only when demand and product differences justify them, and align feeds and structured data.
Validation: compare rendered content, product data, similarity, index state, search queries, and conversion or support behavior after publication.
4Stage 4: Do Internal Links Support Discovery, Relevance, and Buying Paths?
Evidence required: a crawlable internal-link graph, rendered navigation, source and destination URLs, anchor text, link type, click depth, orphan reports, related-product logic, external-link targets, and commercial page priorities. Do not assume a fixed hierarchy is correct for every catalog.
Use the source levels as an audit sequence. Level 1 reviews the homepage and its durable routes. Level 2 reviews category hubs. Level 3 reviews subcategories or collections. Level 4 reviews product destinations. Level 5 reviews supporting guides and editorial resources. For each level, record which users and crawlers should move to the next page and why.
Pass condition: priority products and categories are reachable through descriptive, relevant links; supporting content routes authority and users toward the appropriate commercial destination; utility pages do not dominate link allocation; and automated widgets select related items using explainable logic.
Fail condition: orphaned products, broken paths, excessive utility linking, generic anchors, or unrelated automated recommendations.
Audit in order:
- map actual architecture to the five levels,
- export the link graph,
- identify orphaned priority products,
- identify over-linked or bypassed paths, and
- review anchor and image-link meaning.
Severity is Critical for orphaned or unreachable revenue pages, High for template-wide irrelevant linking, and Medium for isolated anchor or depth issues. Owner: SEO architecture with UX, merchandising, content, and engineering.
Corrective action: change navigation, breadcrumbs, contextual links, collection membership, related-product rules, or utility templates while preserving user usefulness. Validation: rerun the rendered crawl, compare inbound links and click depth, inspect anchor samples, and test the buying journey. Link-count changes alone do not prove ranking impact.
5Stage 5: Do Product and Category Templates Meet Performance Requirements?
Evidence required: Core Web Vitals field data, lab traces, template groups, device breakdown, connection assumptions, image requests, JavaScript tasks, third-party scripts, and representative product and category URLs. Record whether the issue is field-confirmed, lab-only, or an observation requiring more data.
Review five areas in sequence.
1. LCP on product pages. Identify the actual LCP element, request priority, image dimensions, format, preload behavior, server timing, and cache path. Preserve the 3MB example as a defect scenario, not a universal threshold.
2. CLS on listing templates. Confirm whether image dimensions, banners, consent tools, fonts, reviews, or dynamic merchandising cause movement in real sessions.
3. INP on add-to-cart and variant interactions. Capture the event, long tasks, script owner, and user impact. Do not assume JavaScript volume alone proves failure.
4. Mobile and desktop comparison. A 20-40 point lab gap can direct investigation, but field data and the affected user journey determine severity.
5. Third-party scripts. List purpose, owner, load timing, blocking cost, data dependency, and removal risk.
Pass condition: priority templates meet approved field targets or have documented exceptions, interactions remain usable, and asset delivery matches the rendered need. Fail condition: field failures affect a material template, or severe lab defects explain a reproducible user problem.
Severity is Critical for broken purchase interactions, High for template-wide field failures, and Medium for isolated lab regressions. Owner: performance engineering with platform, UX, analytics, and vendor owners.
Corrective action: optimize the responsible image, code, rendering, or script dependency. Validation: deploy through controlled release, retest lab traces, and confirm field movement after enough real-user data.
6Stage 6: Is Structured Data Accurate, Eligible, and Consistent?
Evidence required: rendered structured data, visible page content, product database values, merchant feed values, eligibility guidance, Search Console enhancement reports, and representative tests for each template. The audit must validate accuracy and eligibility, not merely presence.
Review five source areas in sequence.
1. Product markup. Confirm that name, description, image, brand, SKU, offer, price, currency, condition, and availability match the selected visible product and variant.
2. AggregateRating. Confirm that ratings are visible, current, eligible, and drawn from reviews represented on the page. Ask eligible customers consistently for honest feedback without incentives, discouraging negative feedback, or selecting only satisfied customers.
3. Breadcrumb markup. Confirm that the rendered path matches the intended hierarchy and does not point through blocked, duplicate, or irrelevant categories.
4. FAQ content on category pages. Useful questions may be published for readers, but do not claim FAQPage markup can earn a Google FAQ rich result. Google no longer shows that feature. Do not add or change schema under this contract.
5. Organization and site-search information. Confirm that the business entity, name, URL, logo, and supported search behavior match the visible site.
Pass condition: rendered values are current, visible, eligible, and consistent with the page and product source. Fail condition: stale price, unavailable inventory, unsupported rating, mismatched breadcrumb, invisible content, or invalid entity data.
Severity is Critical for misleading price, availability, or identity information; High for template-wide invalid data; Medium for non-critical missing recommended properties. Owner: engineering and product-data owners, with SEO validation.
Corrective action: repair the data source or rendering logic. Validation: test rendered output, sample live products, and monitor enhancement reports without promising a rich result.
7Stage 7: Which Findings Deserve the First Implementation Sprint?
Evidence required: organic landing pages, revenue or qualified-conversion data, stock and margin context where available, ranking positions, query demand, issue severity, dependencies, competitor samples, implementation effort, and confidence in the finding. The purpose is prioritization, not a guaranteed revenue forecast.
Use four source components in order.
1. Revenue-weighted URL priority. Export the top 100 organic landing pages and identify Priority 1 pages using documented commercial and operational criteria. Do not assume revenue per session alone is sufficient.
2. Ranking-gap review. For the top 20 commercial pages, record current position ranges such as 8-15 and desired opportunity ranges such as 1-5 only as observed planning bands. Confirm result type, intent, page quality, and realistic competitiveness before estimating value.
3. Competitor comparison. Review 3-5 relevant competitors and isolate differences in the previous six stages. A difference is an actionable hypothesis, not proof of causation.
4. Conversion and SEO alignment. Compare landing-page traffic with conversion quality, stock, returns, device, and user intent. Strong traffic with weak conversion may require merchandising, UX, policy, or intent correction rather than more ranking work.
Pass condition: every sprint item has evidence, a defined failure, severity, commercial context, owner, dependency, correction, validation method, and success measure. Fail condition: the plan is a flat technical list or uses unsupported revenue estimates. A page at position 7 moving toward 2-3 is an example of an opportunity pattern, not a promised multiplication.
Severity and priority are separate. A Critical issue can be expedited even on a lower-value page when it creates legal, security, data, or customer harm; otherwise commercial impact and dependency help sequence the sprint.
Owner: SEO lead with engineering, merchandising, content, analytics, and product stakeholders. Validation: review each completed fix against its original evidence and record the result at the agreed review checkpoint.
8What Most Guides Get Wrong
A generic audit records whether a tag or feature exists. A useful ecommerce audit verifies whether the page should exist, whether search systems can access and interpret it, whether it serves a distinct buyer intent, whether the product facts are accurate, and whether the page supports a commercially meaningful path.
Checklists also fail when they mix symptoms and causes, assign no owner, or close findings without validation. A technically valid page can still fail because it targets the wrong intent or duplicates another page's role.
A technically imperfect low-value page may deserve lower priority than a small defect on a major revenue page. The audit must therefore preserve evidence, severity, commercial context, corrective ownership, and a repeatable validation method.
Because catalogs, filters, feeds, templates, and policies change, the audit should leave behind queries, crawl settings, samples, dashboards, and decision rules that can be rerun.
9Why Evidence and Closure Matter More Than Audit Volume
A long audit can still fail if nobody can reproduce the evidence, understand the pass condition, own the correction, or verify the result. Ecommerce stores amplify that problem because templates, products, feeds, variants, filters, and policies change continuously.
The useful unit of work is not a discovered defect; it is a closed finding with evidence before and after the repair. The 7-stage structure is therefore a way to organize responsibility. It asks what the site is exposing, which buyer decision each page serves, whether product information is accurate, how users and crawlers move through the catalog, whether templates perform, whether machine-readable facts agree with visible content, and which correction deserves priority.
The audit becomes institutional knowledge only when the crawl settings, samples, decisions, owners, changes, and validation results are preserved.
10Your 30-Day Evidence and Validation Plan
Evidence setup, days 1-3
Collect the full crawl, Google Search Console Coverage, Performance, Core Web Vitals, and Enhancements evidence. Export the top 100 organic landing pages with traffic and commercial data, and document crawl settings.
Outcome: A reproducible URL inventory, evidence baseline, owner list, and revenue-weighted review sample.
Crawl controls, days 4-6
Complete Stage 1 by testing URL patterns, robots.txt, canonicals, sitemaps, status codes, rendered links, and representative logs. Record pass or fail, severity, owner, correction, and validation for each finding.
Outcome: A crawl and index control register with sample URLs, responsible owners, and validation commands.
Intent review, days 7-10
Complete Stage 2 by mapping intent for the top 30 category and product pages, documenting query clusters, result types, competing URLs, approved page roles, and corrective decisions.
Outcome: An intent and cannibalization register with evidence, ownership, and retest criteria.
Product review, days 11-14
Complete Stage 3 by assigning product review Tiers, checking similarity for Tier 1 and 2 products, validating product facts, and recording completeness or variant defects.
Outcome: A product-content remediation plan with evidence requirements, page owners, and rendered validation steps.
Links and data, days 15-18
Complete Stages 4 and 6 by exporting the internal-link graph, checking orphan and utility-page patterns, and validating rendered structured data without adding or changing frozen schema.
Outcome: An internal-navigation repair plan and structured-data accuracy register with implementation owners.
Performance tests, days 19-22
Complete Stage 5 by testing mobile product and category templates, identifying LCP, CLS, INP, image, and script causes, and assigning severity based on field and lab evidence.
Outcome: A template performance backlog with reproducible traces, technical owners, and post-release validation.
Priority model, days 23-27
Complete Stage 7 by reviewing the top 5 commercial keyword opportunities, relevant competitors, landing-page conversion quality, dependencies, effort, confidence, and commercial context.
Outcome: A sequenced sprint combining all 7 stages with supported priority, ownership, corrective action, and validation.
Audit closure, days 28-30
Compile the audit by High, Medium, and Low revenue-impact tiers while preserving separate technical severity. Assign owners, timelines, success measures, and a 90-day review checkpoint.
Outcome: A complete implementation register and a 90-day validation schedule that closes findings with evidence.