Website SEO audit: what to check yourself and what to commission

After a redesign, the clicks chart in Search Console starts to look like a staircase going down. The contractor says, “Google probably updated its algorithm.” That is possible. But you can check something simpler first: do the old page addresses still open, and where do they lead now? If shoppers and search engines find errors instead of products, it is too early to accept the algorithm explanation.

A website SEO audit investigates why important pages may be inaccessible, unclear, or a poor match for search queries. The useful output is not a thick PDF. It is a list of verified problems with URLs, evidence, priorities, and a way to check each fix. An owner can spot some of the most important issues in 40 minutes. The rest calls for access, tools, and someone who can distinguish scanner noise from an actual loss.

We will run the checks on a hypothetical online home goods store. After a redesign, it changed category and product URLs, but the team did not redirect the old ones. This is a teaching example, not a GrandMa client case or a confirmed explanation for a real site’s decline.

Write down the redesign date and choose two periods of equal length before and after it. In Search Console, compare clicks and impressions separately, then filter by pages and queries. A fall in clicks with similar impressions can have a different cause from a fall in both. If only one category changed, do not diagnose the entire store. With very little data, a handful of visits can make a graph look dramatic.

For our hypothetical store, select five important URLs: the homepage, two categories, and two products that received search visits before the redesign. Find their old URLs in the earlier Search Console report, an old sitemap, or a saved page list. Write the corresponding new URLs next to them. Now you have something to compare, rather than pressing “crawl the entire site” and waiting for a verdict.

This is not a complete technical SEO audit. It is a route to finding out whether you need an urgent fix, deeper diagnosis, or both. Search Console access helps considerably; without it, you will see less.

The report names above follow Search Console’s English interface; localized menus and configurations may differ. Google explains what Search Console does. URL Inspection shows Google’s known version of a page and offers a separate live URL test. These are different states: the site may already have changed while Google has yet to crawl its new version.

The site: operator is a quick orientation tool, but Google warns that its results are not a complete list of indexed URLs. Do not treat the number at the top of the results as an indexation ledger. Use URL Inspection for a specific page.

If you lack Search Console access, record that as the first organizational issue. Ask the property owner to grant access to the relevant domain or property rather than sending a screenshot of the graph once a month. Meanwhile, inspect live pages, old URLs, and the mobile route to purchase. Separate what you observed directly from what you cannot yet know without Search Console.

In our scenario, an old product URL returns 404 after the redesign. The new URL opens, but the old one does not redirect to it. Two old category links in the menu also lead nowhere. This is a verified technical problem within the example that could affect people and search crawling. It does not prove how much of the fall in clicks this one fault caused: demand, seasonality, and changes to the pages may also have played a part.

Start the fix with a URL mapping: old category → corresponding new category; old product → the same product at its new address. For permanently moved pages, Google recommends permanent server-side redirects, such as 301 or 308. Do not send every old product to the homepage. A visitor asked for a particular product and needs an understandable path to it. If a product has been discontinued with no suitable replacement, a 404 can be the honest response; it is not automatically an “SEO error.”

Once the redirects are in place, open every old/new pair and check the final HTTP status, product, title, canonical, internal links, and current sitemap. Prefer one direct redirect to a chain of several. Then watch the specific URLs in Search Console over the following weeks. Google says recrawling during a site move takes time; I would not promise a date for the old click count to return.

The example also shows why a commissioned audit should not begin with a list of colored scores. One missing path from old URLs may matter more than dozens of warnings about decorative images if those URLs brought shoppers in.

Write the finding so a developer does not have to repeat your investigation. For this store, a short table could look like this:

There is no “expected traffic increase” column. Three checks cannot honestly produce that number. There are concrete actions and acceptance conditions. If the store has thousands of broken addresses, apply the same logic to the full list rather than checking only three examples.

In Page indexing, “not indexed” is not a diagnosis. Google explains that a page may be absent from the index for legitimate reasons: a duplicate, deliberate noindex, or a discontinued product without a replacement. If a URL was never indexed, do not call its absence “deindexing.” First ask whether that page should appear in search and why Google has not included it.

robots.txt controls what a crawler may visit but is not a reliable way to hide a page from search. noindex serves another purpose: it tells a crawler not to include a page in the index, provided the crawler can read that instruction. If the rules conflict, simply “removing all blocks” could make matters worse. Record the finding and ask a specialist to explain the plan.

A canonical indicates a preferred main address among duplicate or very similar pages. Google treats it as a signal, not an unconditional command. A new product with a canonical pointing to an old URL that returns 404 deserves attention before the length of its description. Likewise, a sitemap helps Google discover URLs but does not guarantee indexing or rankings.

PageSpeed Insights may show both real-user data and a lab test. Core Web Vitals describe loading, responsiveness, and visual stability. One yellow score neither proves why traffic fell nor predicts a Google position. But if a product page moves under your finger while you tap “Buy,” you have a problem for the customer regardless of the rankings graph.

Forty minutes is enough to notice symptoms and collect examples. It cannot replace checking the whole store. If thousands of URLs changed, you need a list of old and new URLs, a site crawl, analysis of category and product templates, and checks of redirects, internal links, canonicals, sitemaps, and Search Console data. Sometimes server logs are needed. The audit should also ask whether pages answer real queries and whether they compete with one another.

A ten-page site with one obvious 404 needs a different scope. Do not buy an elaborate diagnosis just because the report is thicker. We explain the difference between a quick inspection and paid diagnosis separately. If you want to start with one page’s problem, GrandMa offers a site check. Our SEO service covers research, priorities, and continued work when one inspection is not enough.

A useful audit considers the pages that should bring enquiries to the business. Fewer clicks to an old blog post and the disappearance of a core service page may deserve different priorities. For a local business, its company profile and Google Maps presence can be another part of search visibility that a technical website crawler cannot measure.

When accepting a professional audit, I would ask for three layers. First, a short summary: what happened, what is proven, and what remains a hypothesis. Second, an issue register with exact URLs, reproduction steps, impact, priority, and owner. Third, a post-fix check plan: which URLs to open, which Search Console data to review, and when to assess again. If the contractor also implements changes, treat the two deliverables separately: a report identifying a problem does not prove that a live site has been fixed.

Ask for dependencies. There is little point in rewriting the title on an old URL that should redirect to a new one. Decide which address survives first, then edit fields and internal links. That sequence saves the editor work and keeps new content off a page that will soon be removed.

1. Which URLs and page types will you inspect? Ask about categories, products, services, blog posts, and exceptions. “The whole site” is too vague for acceptance.
2. Which evidence will you use? Search Console, real HTTP responses, a site crawl, an old-URL map, and business questions each show a different piece of the picture.
3. How will you separate a cause from something that happened at the same time? Several things may have changed during a redesign. Ask for verified facts and the limits of the conclusion.
4. How will you set priorities? Ask for the impact on important pages, risk, effort, and dependencies between changes.
5. Who will make the fixes? Auditing and implementation can be separate jobs. Name the owner and agree on the task format for developers and editors.
6. What will I receive? Expect exact URLs, evidence, a first-priority plan, a way to retest, and a short explanation for the owner.

If the result is an 80-page PDF with 300 identical “critical” warnings and no first five actions, the team still does not know what to do on Monday. A good audit lets you open a URL, repeat the check, and know who should receive the fix. After implementation, verify the live pages separately from a promise in the document.