What is included in an SEO audit? A buyer checklist

An SEO audit should include enough evidence and implementation detail for a team to decide what to fix, who should fix it, and how to confirm the defect is closed. The familiar categories—technical SEO, content, internal links, structured data, performance, and authority—are only headings. The quality of the audit depends on what sits underneath them.

A crawler export is not a finished audit. Neither is a presentation full of best practices. The useful deliverable is a prioritized finding register tied to affected URLs, business relevance, owners, dependencies, and verification. If answer-engine optimization is included, the report should also preserve the sampled prompts and responses while being explicit that generated visibility is variable.

The audit should begin with scope and business context

Before a provider crawls the site, it should confirm the properties and decisions involved. At minimum, the scope should name:

• domains, subdomains, applications, and environments included;
• target markets and languages;
• priority products, services, audiences, and conversion routes;
• known migrations, redesigns, platform constraints, or recent incidents;
• available access to Search Console, analytics, log files, and the CMS;
• page or crawl limits and representative template families;
• stakeholders available for product, engineering, legal, or content questions;
• explicit exclusions and assumptions.

This context prevents a provider from assigning the same priority to every warning. A broken canonical on a pricing page and a duplicate title on an expired event page are both technical observations, but they do not have the same business exposure.

Scope should also explain sampling. A large patterned site may be crawled broadly and reviewed by template. Content quality may be sampled across commercial and editorial families. The audit should state what was inspected manually and what was inferred from a pattern.

For help evaluating fees against that scope, see Leaf’s guide to SEO audit cost.

Technical deliverables need reproducible URL evidence

The technical section should inspect how search systems and users encounter the site. Depending on scope, it normally covers:

• HTTP status behavior and redirect chains;
• robots.txt and crawler-specific rules;
• meta robots and relevant HTTP headers;
• canonical tags and conflicting canonical signals;
• XML sitemaps and discovery paths;
• internal links, orphan patterns, and crawl depth;
• duplicate and parameter URL patterns;
• mobile and JavaScript rendering on key templates;
• international and language targeting where relevant;
• structured-data inventory, syntax, and visible accuracy;
• performance evidence with its data source labeled;
• broken forms or navigation when they affect organic landing paths.

The report must keep controls distinct. Google explains that robots.txt controls crawling and does not reliably prevent a URL from appearing in results. A finding should not call a robots-blocked URL “deindexed” without additional evidence.

For every material issue, require an export or exact list of affected URLs, a representative example, the expected state, and reproduction steps. Screenshots are helpful for rendered defects, but screenshots alone are poor implementation inputs for a pattern affecting hundreds of pages.

A buyer needing more detail can use Leaf’s technical SEO audit service checklist alongside a proposal.

Content deliverables should map pages to buyer intent

A content review should do more than count words or flag missing keywords. It should identify what question each priority page owns, whether that page answers it credibly, and where multiple URLs compete for the same job.

Expect the review to cover:

• commercial page purpose and intended audience;
• overlap, cannibalization, and consolidation candidates;
• titles, descriptions, headings, and opening answer clarity;
• unsupported, inconsistent, stale, or overbroad claims;
• authorship, review ownership, dates, and source quality where relevant;
• missing comparison, implementation, pricing, risk, or proof information;
• internal links between discovery, evaluation, and conversion pages;
• thin archives, campaigns, tags, or programmatic patterns;
• editorial content that receives traffic but has no useful buyer path.

Reject recommendations such as “add more content” unless the provider identifies the audience, missing decision, suitable evidence, affected page, and expected next action. A word count may describe a page; it does not diagnose its usefulness.

Google’s people-first content guidance can inform questions about audience, expertise, and sourcing. It is official platform guidance, not proof that following a checklist will produce a specific ranking.

The deliverable should clearly distinguish factual defects from editorial judgment. An outdated pricing claim can be verified against an owned source. A recommendation to merge two pages is a strategic decision that should include rationale and risk.

Structured data and entity findings must match visible facts

A structured-data section should inventory detected types, identify parsing errors, compare properties with visible page content, and explain whether a type is relevant to a supported search feature or simply part of Schema.org vocabulary.

Google states that structured data should represent the page content and does not guarantee a rich result. A finished recommendation therefore specifies:

• the affected template and current code;
• the inaccurate, missing, or invalid property;
• the visible fact that supports the proposed value;
• the relevant official documentation;
• the validation method after release.

“Add schema” is not a deliverable. Nor should an audit recommend review ratings, people, products, services, or claims that the visible page cannot support.

Entity consistency extends beyond markup. Company names, product names, addresses, authors, credentials, and policy details should agree across maintained pages and profiles. The audit can identify discrepancies; responsible internal owners must approve sensitive facts.

Performance deliverables must label field and lab data

Performance belongs in an SEO audit when it affects important templates and user tasks. The report should identify which data comes from real-user field sources and which comes from controlled lab tests. The two are complementary, not interchangeable.

The official Core Web Vitals documentation defines the current metrics and explains their user-centered purpose. A responsible audit cites the source and collection context, then connects diagnostics to specific templates or components. It does not claim that one laboratory score caused a ranking or revenue change.

Useful performance findings include the page family, device context, metric source, affected component or request, likely technical owner, proposed investigation, and retest environment. If field data is unavailable, the audit should say so rather than present lab output as observed user experience.

Measurement deliverables should verify the commercial path

An audit cannot credibly prioritize for pipeline if the primary conversion events are broken or ambiguous. The measurement review should identify the intended actions for major landing pages and test whether those events are captured reliably.

Depending on access and scope, this may include:

• form submission, demo, trial, purchase, call, or booking events;
• duplicate event firing and test or internal traffic;
• source and landing-page preservation;
• cross-domain or third-party form behavior;
• connection between marketing events and qualified CRM stages;
• annotations for migrations and material site releases.

The audit should not promise an attribution model it did not inspect. It should state which systems were reviewed, which events were tested, and which downstream outcomes remain outside scope.

Separate implementation completion from external outcomes. A repaired analytics event can pass immediately. Search impressions, qualified leads, or answer citations require observation and have variables beyond the change itself.

An AEO add-on should preserve prompts, sources, and limitations

A combined SEO and AEO audit should not simply add an “AI score.” It should define the answer products tested, prompt panel, market and language, date, account context where visible, repetitions, and evidence captured.

Useful fields include:

• exact neutral discovery and branded accuracy prompts;
• full or reviewable response text;
• cited URLs and domains;
• linked versus unlinked mentions;
• recommendation appearances kept separate from mentions;
• incorrect company or product claims;
• competitor sources that answer the buyer question;
• proposed owned-page correction or experiment;
• retest conditions.

Google confirms that existing SEO fundamentals apply to its AI features. That supports including technical and content foundations in an AEO review. It does not establish that a page change will earn inclusion. The audit should label such recommendations as experiments and avoid guarantees.

Crawler access is also product-specific. OpenAI publishes official bot and crawler controls. A report should state which user agent and purpose it checked rather than treating one robots rule as a universal “AI visibility” switch.

Every finding needs priority, ownership, and closure

The finding register is the operational center of the audit. Require these fields for every material item:

A practical priority model considers business exposure, severity, confidence, effort, and dependency. It should not hide these judgments inside a proprietary score. Access and measurement defects often need resolution before content experiments can be interpreted, but the specific order depends on the site.

The final package should include more than a presentation

A complete handoff commonly includes:

1. an executive summary tied to the business decision;
2. a scope and methodology statement;
3. a prioritized finding register;
4. affected URL and crawl evidence exports;
5. content or page-ownership maps where relevant;
6. implementation notes and examples;
7. a now, next, and later roadmap;
8. a readout, walkthrough, or defined question period;
9. retesting instructions or an included retest;
10. exclusions, unresolved questions, and access limitations.

Not every audit must include code changes, final copy, backlinks, content production, migration support, or ongoing monitoring. Those exclusions are acceptable when explicit. Ambiguity is the problem: buyers should know whether the provider diagnoses, implements, validates, or does all three.

Use this buyer acceptance checklist

Before accepting an audit, confirm:

• □ The reviewed domains, templates, markets, and data sources match the agreed scope.
• □ Manual samples and automated coverage are clearly distinguished.
• □ Material technical findings include affected URLs or patterns.
• □ Content recommendations identify the buyer question and missing evidence.
• □ Structured-data recommendations match visible facts and official documentation.
• □ Field and lab performance observations are labeled correctly.
• □ Primary conversion events were tested if measurement was in scope.
• □ AEO prompts, products, dates, responses, and citations are preserved if included.
• □ Mentions, citations, recommendations, and conversions are not collapsed into one score.
• □ Every priority item has an owner, dependency, and closure method.
• □ Confirmed defects and external-outcome experiments are separated.
• □ Deliverables, exclusions, implementation, and retesting responsibilities are explicit.

Leaf’s free homepage assessment is intentionally narrower than this list. Leaf’s $875 SEO and AEO audit provides a broader prioritized report for established B2B sites that fit its scope, but implementation is separate and no rankings or citations are guaranteed.

A good audit does not impress buyers with the number of checks. It reduces ambiguity. Your team should leave knowing which evidence is reliable, which decisions still require judgment, what work comes first, and what observation will show that each specific change is complete.