How to check a site before you buy an SEO audit

Before you buy an SEO audit, check what your site can already prove. You do not need a full tool stack to see whether Google can crawl key pages, whether important URLs are indexed, which pages actually earn impressions, and where you have no evidence yet. That matters because a good audit is a diagnostic exercise, not a promise machine.

If someone offers rankings, leads or sales as the direct output of an audit, treat that as marketing, not evidence. What an audit can do is show you what is technically blocking discovery, what content appears weak or duplicated, what users and search engines can already reach, and what should be tested next. That is the standard we use at The SEO Blog, because a crawler visit, an indexed page, an impression, a click and an enquiry are different things.

What an SEO audit is actually for

An SEO audit is a structured check of how a website can be discovered, crawled, indexed and understood, and whether its pages are useful enough to deserve attention. In practice, it should answer questions such as:

• Can Googlebot reach the pages that matter?
• Are important URLs indexable, or blocked by noindex, canonicals, redirects or status codes?
• Which pages already get impressions, and for what kinds of searches?
• Where is the site duplicating pages or competing with itself?
• Do key pages answer a real need clearly, or do they only mention keywords?
• Are there obvious performance or accessibility problems that make pages harder to use?

That is different from promising an outcome. Google’s own guidance is careful on this point. The Google SEO Starter Guide explains useful practices for discovery and content, but it does not promise rankings. For local businesses, Google also states that local ranking depends on relevance, distance and prominence, and that a better local position is not something you can buy from Google’s systems directly, see Google local ranking guidance.

So when you assess an audit, separate evidence from interpretation:

• Evidence is a blocked URL, a noindex tag, a canonical pointing elsewhere, a page with impressions but poor click through, or a duplicate template repeated across hundreds of pages.
• Interpretation is the likely effect of fixing it.
• Promise is “this will get you to page one” or “this will increase sales”.

A competent audit should be strong on the first two and restrained on the third.

Start with the evidence you already have

Before buying tools or hiring help, gather three things:

• access to Google Search Console
• access to your analytics package
• a simple list of important URLs

That URL list can be a spreadsheet with columns for page type, purpose, canonical URL, current status and whether you think it should be indexed. Include your homepage, service pages, product or category pages, location pages, articles, contact page and any pages used in campaigns.

Then compare your list with what Google Search Console already shows.

First, open the Page indexing report. Google’s documentation is clear that this report helps you investigate why a URL is or is not indexed, and that not every excluded URL should be indexed, see the Google Page indexing report documentation. That point matters. Many sites contain pages that should stay excluded, such as duplicate filtered URLs, internal search results, thank-you pages, staging copies or parameter variants.

Work through your URL list like this:

1. Mark pages that should be indexed.
2. Check whether they are actually indexed.
3. Note the exclusion reason where they are not.
4. Mark pages that should not be indexed.
5. Confirm that their exclusion is intentional.

Next, use the Performance report in Google Search Console. Export the pages with impressions and clicks for a meaningful recent period. You are looking for three broad groups:

• Pages with impressions and clicks, these have evidence of search visibility.
• Pages with impressions but very low clicks, these may have snippet, intent or position issues.
• Pages with no impressions at all, these may be new, weak, blocked, not indexed, or simply not aligned with any search demand.

Then check analytics. Do not use analytics to prove SEO success on its own, but do use it to understand what happens after a visit. If a page gets search clicks but users leave immediately because the page does not answer the query, that is useful context. If a page gets no search traffic but performs well in email or paid campaigns, you may still want it, just not as an indexed SEO landing page.

At this stage, you are not solving everything. You are building a baseline. If you want a related checklist, we have already set out a practical site audit you can run before buying tools, and our Search Console evidence and limitations guide explains what those reports can and cannot prove.

Which technical checks matter first?

Technical auditing can become a long list very quickly. Start with the issues that can stop discovery or indexing altogether.

1. Crawl access

Check whether important areas are blocked in robots.txt. A blocked page may still be known to Google, but blocking can prevent proper crawling and diagnosis. Also check whether navigation relies on elements that are hard to crawl, when plain HTML links would do the job better. The Google SEO Starter Guide supports clear internal linking and crawlable site structure.

2. Indexability

For pages that should appear in search, check for:

• noindex directives
• x-robots-tag headers
• password protection
• soft 404 behaviour
• pages only available after form interaction or scripts fail

A page can exist for users and still be a poor candidate for indexing. That is why “excluded” is not automatically “broken”, which Google states in the Page indexing report documentation.

3. Canonicals

Review canonical tags on important pages. A self-referencing canonical is common on the preferred version. Problems arise when a page that should stand on its own points to another page by mistake, or when templates set canonicals inconsistently across variants.

Common failures include:

• paginated pages canonicalised to page one when they contain distinct items
• local landing pages canonicalised to a generic national page
• tracking parameter URLs left indexable instead of canonicalised cleanly
• duplicate CMS routes that each claim to be canonical

4. Redirects

Export redirect chains and loops if you can. At minimum, test key pages manually. You want old URLs to resolve cleanly to the best current equivalent, ideally in one hop. Watch for:

• internal links still pointing to redirected URLs
• http to https plus non-www to www chains
• removed pages redirected to the homepage regardless of topic
• mixed use of 302 and 301 where the intent is permanent

5. Status codes

Important pages should return 200. Not 3xx, not intermittent 5xx, not a branded error page with a 200 status. A surprising amount of confusion comes from pages that look live in the browser but return the wrong code.

6. XML sitemaps

A sitemap does not force indexing, but it is a useful declaration of preferred URLs. Check that your sitemap includes canonical, indexable URLs and excludes obvious non-indexable ones. If the sitemap is full of redirected, duplicate or noindex pages, it becomes less useful as a signal.

7. Obvious duplication

Look for repeated title tags, repeated H1s, near-identical location pages, duplicate category text, and multiple URLs serving the same purpose. Duplication is not always a penalty issue. Often it is simply a sign that the site does not make clear which page should rank for which need.

If you are checking a small business site and need a simpler decision framework, our piece on which SEO audit suits a small business site best may help.

How to review content without guessing

Content review goes wrong when people reduce it to word counts and keyword frequency. A better approach is to ask what job the page is doing, for whom, and what evidence shows it is doing it.

Start with page purpose. For each important page, write one plain sentence:

• what need is this page meant to meet?
• what type of visitor is it for?
• what action should be possible after reading it?

Then review the page against that purpose.

Does it answer a real need?

A service page should explain what is offered, where, for whom, and any constraints or next steps. A product category should help users compare options. A guide should teach something specific. If the page exists mainly because a keyword tool suggested a phrase, that usually shows.

Are headings clear?

Headings should help a reader scan the page and understand the structure. If every heading is vague marketing copy, the page may be hard to interpret quickly. The Google SEO Starter Guide supports content that is useful and organised.

Are links descriptive?

“Click here” and repeated “read more” links tell users little. Descriptive internal links help both people and search engines understand what sits behind the link. They also make audits easier, because you can see whether the site is reinforcing the right page relationships.

Is the page duplicating another page?

Sometimes two pages target the same need with slightly different wording. That can split internal links, confuse canonicals and leave neither page strong. Consolidation is often better than expansion.

Does it match likely search intent?

You do not need to guess every query. Look at the queries and landing pages you already have in Google Search Console. If users arrive on a page for informational searches but the page opens with only a hard sell, it may not fit. If the page is meant to convert high-intent visitors but spends most of its length on generic education, that can be a mismatch too.

Is it stuffed with keywords?

Awkward repetition is usually a symptom of writing for tools instead of readers. Use the terms that belong naturally on the page, but do not force them. Google’s guidance has been consistent for years, useful content first, not mechanical repetition, see the Google SEO Starter Guide.

What performance and accessibility checks can tell you

Performance and accessibility checks are worth doing, but they need careful interpretation.

Core Web Vitals describe aspects of loading, responsiveness and visual stability. They are useful because they show whether pages feel clumsy or unreliable to use. Google and web.dev explain what these metrics represent, but they do not present a score as a guarantee of rankings or sales, see web.dev on Core Web Vitals.

In an audit, these checks can tell you:

• whether layout shifts are disrupting reading or clicking
• whether pages feel slow to become interactive
• whether large media or scripts are obvious bottlenecks
• whether templates differ sharply between page types

They cannot prove:

• that improving a metric will increase revenue
• that a “green” score means the page is good
• that a poor score is the main reason a page underperforms in search

Accessibility checks are similar. Basic checks can reveal barriers that affect real users and often improve clarity for everyone. The W3C’s accessibility design tips cover practical issues such as readable contrast, understandable navigation and clear controls.

A basic review should look at:

• colour contrast in text and controls
• keyboard access for menus, forms and pop-ups
• form labels and error messages
• link text that makes sense out of context
• heading order and page structure
• alt text where images carry meaning

These checks can tell you that the page may be harder to use than it needs to be. They cannot, on their own, tell you why a page is not converting or where it should rank. They are part of diagnosis, not a complete explanation.

How to judge an audit report before acting on it

A useful audit report is not just a list of issues. It should help you decide what to do first, what evidence supports the recommendation, and how you will know whether the change helped.

Look for these qualities.

Findings are prioritised

A report that gives equal weight to a missing alt attribute and a sitewide noindex problem is not helping. The best reports separate:

• critical blockers
• high-value improvements
• useful but non-urgent tidy-up work
• observations that need testing before rollout

Findings are evidenced

Each issue should show where it appears, how often, and why it matters. Good evidence includes example URLs, screenshots, source code references, crawl samples and report exports. Vague statements such as “your site authority is low” are not enough on their own.

Recommendations are testable

You should be able to implement a change and then check what happened. For example:

• after fixing a canonical error, the preferred page should become the indexed version
• after improving internal links, a deeper page may start receiving crawl and impression evidence
• after consolidating duplicates, one page may gather clearer query signals

If there is no way to observe the result, the recommendation may be too abstract.

The report separates search access from AI crawler controls

This is an area where confusion is common. Google Search access is not controlled by Google-Extended. Google says its existing search fundamentals apply to AI features in Search, and no special AI text file or special AI schema is required, see Google AI features guidance.

Likewise, OpenAI’s documentation separates search access from training controls. OAI-SearchBot and GPTBot are not the same thing. OAI-SearchBot is for search-related discovery, while GPTBot relates to training access controls in OpenAI’s documentation, see the OpenAI crawler documentation.

So if an audit says:

• “block Google-Extended to improve Google Search rankings”
• “allow GPTBot to get cited in search”
• “add llms.txt for Google AI results”

treat those claims carefully. They do not follow from the cited platform guidance. A crawler allowance, a citation, an impression and a business lead are separate events.

The report does not confuse diagnosis with outcome

An audit can show that pages are blocked, duplicated, slow, unclear or misaligned. It cannot honestly guarantee that fixing those issues will produce a specific ranking or revenue outcome. Search performance also depends on competition, demand, location, site history, brand signals and the quality of alternatives.

That does not make auditing pointless. It makes it useful in the right way. You buy an audit to reduce uncertainty, expose avoidable faults, and set better tests.

If you want to go further after reading one, our guide to investigating a change in search performance is a practical next step, and our editorial responsibility and evidence page explains how we separate source guidance from interpretation.

Before you buy an SEO audit, ask for the method, not just the deliverable. Ask what data will be checked, how findings will be prioritised, what counts as evidence, and which claims remain uncertain. If the answers are clear, specific and testable, you are much more likely to get a report you can use.