This SEO audit template turns technical, content, authority and AI search checks into an assignable delivery plan. It includes a downloadable Excel workbook with 79 preloaded checks, scoring fields, evidence requirements, owners, dependencies, due dates and verification steps.

The template is built for consultants, in-house teams and agencies that already know how to run a crawl but need a better way to decide what should happen next. Instead of treating every warning as equal, it separates evidence, impact, confidence, effort and dependency so the team can distinguish a serious diagnosis from a noisy tool alert.

Download the DIY AI SEO aud it template for Excel

What is included in the SEO audit template?

The workbook contains four sheets:

• Start Here: the scoring rules, priority formula and recommended workflow.
• Audit Register: 79 checks covering crawling, canonicals, internal linking, content, cannibalisation, structured data, page experience, backlinks, local SEO, AI crawler access and citations in AI answers.
• Executive Summary: live counts for open issues, priority bands, completed fixes and category-level progress.
• Evidence Examples: examples of weak evidence, stronger evidence, sensible impact statements and repeatable verification methods.

The register is deliberately editable. A 50-page brochure site should not carry the same audit rows as a marketplace with millions of parameter URLs. Delete irrelevant checks, add platform-specific risks and keep the working queue small enough that owners can use it.

---

---

Why most SEO audit checklists fail after delivery

A crawler can produce thousands of warnings in minutes. That does not mean the site has thousands of independent problems. One faulty canonical component may create 40,000 reported URL issues, while a single unlinked revenue page may have a greater commercial cost.

The recurring operational failure is a technically impressive spreadsheet that nobody can explain, assign or close. Long checklists often mix observations, diagnoses and recommendations in the same cell. The developer cannot tell which URLs prove the issue. The content lead cannot tell what outcome is expected. Three months later, a task is marked complete because code was deployed, even though nobody checked the live result.

This template separates those decisions. Raw evidence proves the condition. The diagnosis explains the cause. Impact describes what is affected. The action states what should change. Verification defines the test that closes the task.

The eight fields that turn an audit into a decision system

Field	Purpose	What good looks like
Evidence	Proves the problem exists	A crawl export, live header, log sample, screenshot, query report or representative URL set
Impact	Explains what the issue affects	A clear consequence for discovery, indexation, relevance, authority, user experience or conversion
Confidence	Shows how certain the diagnosis is	A lower score for sampled or indirect evidence, and a higher score for confirmed live behaviour
Effort	Estimates the work required	Includes implementation, quality assurance, deployment, approvals and migration work
Dependency	Records what must happen first	None, minor, major or blocked, with the prerequisite named in the notes
Owner	Creates accountability	One named person or role responsible for moving the task forward
Verification	Defines how the fix will be checked	A repeatable live test, preferably using the same evidence source that identified the issue
Completion	Separates deployed work from confirmed outcomes	A completed date plus a passed, failed or partially passed recheck result

A useful audit row should be understandable without opening the original crawler project. Compare these two findings:

Weak finding	Actionable finding
1,247 pages have canonical errors.	1,247 filter URLs in /shoes/ declare self-referencing canonicals while internal navigation and the XML sitemap point to the parent category. Evidence: crawl export and five live samples. Recommended action: remove filter URLs from the sitemap, point allowed variants to the agreed category canonical and update navigation links.

The second version gives a developer enough context to investigate the component rather than patching random URLs.

How the priority score works without hiding serious risks

The workbook uses this formula:

(Impact × Confidence) ÷ Effort × dependency multiplier

Dependencies receive a multiplier of 1.00 for none, 0.85 for minor, 0.65 for major and 0.40 for blocked. The resulting bands are Critical, High, Medium and Low.

Treat this as a delivery sequence, not an objective measure of SEO severity. A blocked issue can have very high impact while receiving a lower active priority because the team cannot start it yet. Keep impact visible, isolate high-impact blocked work for management escalation, then sort the remaining unblocked tasks by priority score.

Do not score issues during the first crawl. Score them after the cause and affected scope are understood. Otherwise, teams tend to give every large URL count an impact of five and every familiar fix an effort of one.

Run the audit in five passes rather than 79 disconnected checks

1. Define the indexation and business scope

List the page types that should attract search demand, the sections that support conversion and the areas that should remain private or excluded. Record migrations, redesigns, international folders, faceted navigation, JavaScript frameworks and known platform limits before opening a crawler.

This prevents an auditor from recommending indexation for utility pages that were intentionally excluded or treating every parameter as a defect.

2. Crawl, render and compare sources

Run at least one crawl from internal links and compare it with XML sitemaps, analytics landing pages and any available indexation exports. For JavaScript sites, compare raw and rendered HTML. On large sites, use server logs or edge logs to confirm how bots actually spend requests rather than assuming the crawl simulation tells the whole story.

Collect evidence by URL pattern. Ten representative samples with one shared cause are normally more useful than pasting thousands of URLs into the action register.

3. Review content by task and template

Content quality should be audited against the user task, not a universal word count. Check whether the page matches the likely intent, provides original evidence, answers the decision and differs meaningfully from neighbouring pages.

Review templates separately from individual editorial pages. A weak product-description component may affect every product. A poorly mapped comparison article is a page-level problem. Mixing both in one finding makes ownership unclear.

4. Audit authority, local signals and AI visibility

Join backlink data to live HTTP status so valuable links pointing to retired or redirected URLs are visible. For local businesses, compare website details, location pages, profiles, priority citations and local structured data field by field.

AI visibility needs its own evidence, but it should not become a separate collection of speculative tricks. Record crawler access, public text availability, citation sources, entity consistency and repeatable prompt results. The goal is to find access failures, missing evidence and inconsistent facts.

5. Build the action queue and verification plan

Merge duplicate symptoms under their root cause, score the remaining issues and assign one owner per row. Add due dates only after dependencies are understood. A task with no owner is an observation. A task with no verification method is a recommendation that cannot be closed safely.

SEO audit checklist by category

Category	What to inspect	Evidence to retain
Crawling and indexation	Robots rules, noindex controls, status codes, sitemaps, orphan pages, parameter spaces, JavaScript rendering and index bloat	Live directives, crawl joins, rendered HTML, logs and URL samples
Canonicals and redirects	Canonical targets, signal conflicts, redirect chains, URL variants, internal links to redirects and retired URLs	Canonical matrix, redirect paths, sitemap membership and inlink sources
Internal linking	Crawl depth, contextual links, broken targets, anchors, orphan pages, pagination, internal nofollow and script-added links	Inlink exports, source URLs, rendered anchors and architecture maps
Content quality and intent	Intent match, unique value, stale claims, page purpose, visible text, scaled templates and low-value archives	Query data, SERP review, source checks, content samples and review dates
Cannibalisation	Multiple URLs competing for the same task, duplicate variants, archive overlap and unstable ranking ownership	Query-to-page mapping, ranking URL history and content similarity
Structured data	Eligible types, accurate properties, visible-content alignment, conflicting graphs, breadcrumbs and stable entity identifiers	Rendered JSON-LD, validator results and representative page checks
Page experience	Field performance, mobile journeys, overlays, layout shifts, script weight, accessibility and useful error pages	Field data, traces, request waterfalls, device tests and interaction recordings
Backlinks and authority	Linked-to status, authority distribution, suspicious links, anchor context, unlinked mentions and realistic competitor gaps	Backlink joins, source samples, acquisition history and destination decisions
Local SEO	Business details, duplicate listings, location-page quality, local schema, reviews and priority citations	Profile records, citation samples, location-page evidence and review workflow
AI crawler access	Named user-agent policy, CDN or WAF blocks, accessible HTML, snippet controls, private content protection and change ownership	Policy matrix, edge logs, live responses, rendered text and access tests
Brand citations in AI answers	Prompt coverage, cited sources, incorrect facts, competitor substitution, third-party corroboration and source-page quality	Dated answer captures, citation URLs, prompt context and issue classifications

How to audit structured data without chasing warning counts

Structured data tools can produce a long list of optional-field warnings. Prioritise by eligibility and factual accuracy. Incorrect prices, availability, business details or entity relationships create a more serious problem than an absent optional field that does not affect the intended result.

For each marked-up template, compare the rendered schema with visible page content. Check that duplicate plugins or theme components are not publishing conflicting graphs. Ecommerce teams can use our guide to AI-generated ecommerce schema markup to review implementation choices, but the audit row should still point to live evidence from the site being assessed.

How to audit AI crawler access and brand citations

Start with an access policy, not a list copied from another site’s robots.txt. Search crawling, model training and browser-agent access can have different owners and risk tolerances. Record the named user-agent, intended access, technical control, policy owner and last review date.

Robots.txt is a crawler instruction, not protection for private data. Sensitive or licensed material should rely on authentication and server-side access control. Test the unauthenticated response and check whether caches or public assets expose the same information elsewhere.

For Google AI search features, the practical audit remains rooted in crawlability, indexability, internal links, accessible text and useful content. No special AI schema is required. Review Google’s official guidance for generative AI search before turning unverified AI optimisation advice into an audit task.

Brand citation testing should use a repeatable prompt set. Record the date, market, model or product context, answer, cited sources and classification:

• Absent: the brand is relevant but not mentioned or cited.
• Incorrect: the answer states a wrong or outdated fact.
• Substituted: a competitor is cited for evidence your site could credibly provide.
• Unsupported: the brand is mentioned, but the answer relies on weak or unrelated sources.

A screenshot alone is not a useful task. Link the finding to the source-page gap, inconsistent fact, access problem or missing third-party corroboration that the team can address.

Keep the template tool-agnostic

The workbook does not require a specific crawler, rank tracker or AI visibility platform. Evidence quality matters more than the logo on the export. A strong workflow may combine a crawler, analytics, server logs, backlink data, performance tools and manual page review.

Do not buy another platform solely because it reports more warnings. Check whether it can export URL-level evidence, preserve historical comparisons, segment templates and let the team reproduce a finding. Our comparison of AI SEO tools practitioners repeatedly recommend can help with tool selection, but the template should remain usable after a subscription changes.

Common mistakes that make an SEO audit harder to implement

Scoring symptoms instead of causes

If 20 reports all stem from one CMS component, create one root-cause task and attach the affected reports as evidence. Separate rows encourage duplicate work and inflate the apparent backlog.

Using URL counts as impact

A defect affecting 100,000 low-value filtered URLs may be less important than one noindexed category responsible for a large share of revenue. URL count describes scope. It does not prove business impact.

Giving every task to the SEO owner

SEO may diagnose the issue, but ownership should sit with the person who can move it. Typical owners include engineering, product, content, design, legal, infrastructure and local operations. Keep the SEO lead responsible for verification where appropriate.

Marking work complete at deployment

A release can succeed while the intended search signal remains wrong. Cache layers, template overrides, plugin output and redirect rules can all produce a different live result. Use a separate completion date and recheck the result.

Presenting the full register to every stakeholder

The working register is for diagnosis and delivery. Leadership usually needs the decisions: the highest-impact risks, the work approved now, blocked dependencies, expected outcomes and what will be rechecked. Use the Executive Summary sheet, then open the detailed evidence only when a decision needs defending.

How to adapt the audit template by site type

Site type	Checks to emphasise	Checks commonly overused
Small service site	Indexation, page intent, internal links, local details, conversions and stale claims	Crawl-budget analysis and complex log segmentation
Publisher	Topic ownership, cannibalisation, archive policy, author/entity consistency, internal discovery and update workflow	Uniform word-count rules across different article types
Ecommerce	Facets, canonicals, product availability, category depth, internal links, structured data and retired products	Fixing every filtered URL individually rather than changing the generating component
Marketplace or large platform	URL governance, logs, crawl traps, template rendering, sitemap segmentation, migration controls and ownership	Manual page-by-page review before template sampling
Multi-location business	Location uniqueness, business details, duplicate listings, local citations, profiles, reviews and local schema	Creating near-identical pages for every nearby area
SaaS	Feature and comparison intent, documentation access, JavaScript rendering, international pages, brand facts and third-party citations	Treating app pages and public acquisition pages as one indexation group

Copyable audit register columns

Teams using another project system can copy these headers into Google Sheets, Airtable, Notion or a ticketing platform:

ID    Category    Audit check Scope   Evidence required   Evidence link   Finding Affected URLs   Status  Impact  Confidence  Effort  Dependency  Priority score  Priority band   Owner   Recommended action  Verification method Due date    Completed date  Recheck result  Notes

Keep the evidence link, diagnosis and verification method as separate fields. Combining them into one notes column makes filtering and handover harder.

Turn the audit into a controlled improvement cycle

An audit is useful only when it changes decisions. Start with the highest-impact blocked items so leadership can remove dependencies. Then move the highest-scoring unblocked tasks into the active delivery queue. Keep low-confidence findings in investigation rather than presenting them as confirmed defects.

After implementation, repeat the defined verification test and record the result. Failed checks return to the queue with new evidence. Passed checks can be closed, but template-level fixes should be sampled again after later releases because regressions often reappear through CMS, plugin or component changes.

The strongest SEO audit is not the one with the most rows. It is the one where each important problem has proof, a defensible priority, one accountable owner and a test that confirms the live outcome.