What a Technical SEO Audit Checks

September 18, 2026

No comments

Photo of author

Brent W Peterson

Updated

A technical SEO audit is the check of everything that decides whether a page can be found, read, and ranked before a single word of copy gets judged. Search engines still run that check the old way, through crawling, indexing, and rendering, and the answer engines now run a version of it too, so an audit in 2026 has to ask whether Googlebot and GPTBot can both get in the door. Content Cucumber runs its technical SEO audit in a fixed order, from access to security, and hands back a ranked fix list with each issue tagged critical, high, medium, or low. The order below is that order, with the question each check answers.

Start With Whether Anything Can Reach the Page

The first check is access. A site fails it in minutes. Robots.txt gets read line by line, the XML sitemap gets fetched and compared against the pages that matter, and each URL in the sample gets requested to see what the server sends back, a 200, a redirect, a 404, or a 500 under load. The 2026 addition is the AI crawlers. GPTBot, ClaudeBot, PerplexityBot and Google-Extended each check robots.txt for their own name, and a site that blocked them in a 2023 panic has no chance of being cited by the engines they feed. I think it is crazy for a merchant to block a bot unless the merchant does not want to be discovered by the LLMs at all, and that is the failure covered in your website is fine, the AI still cannot see you. The primer on what site crawlers are covers the mechanics for a reader who has not opened a robots file before. On the same July to September audits, 3 of the 5 sites checked had no sitemap declared in robots.txt, and 4 of the 7 checked for an llms.txt file had none, so the access check fails on the basics more often than on anything exotic.

Count the Pages Google Kept Against the Pages You Published

Indexation is a count, the pages Google kept against the pages published, and the two numbers seldom agree. Search Console’s Pages report lists what Google indexed and what it left out, with a reason for each exclusion, and the audit compares that list with the CMS’s own count of published pages. A store with 2,400 products and 900 indexed product pages has a problem the copy cannot fix. The usual causes are a noindex tag that shipped with a theme, a staging setting that survived launch, pages Google discovered and did not crawl because the site gave it no reason to, and parameter URLs that swallowed the crawl budget meant for the catalog.

Find the Duplicates Before Google Picks One for You

Ecommerce sites generate duplicates by design, a product reachable at three category paths, a filter combination with its own URL, an http and an https version, a trailing-slash twin. Google resolves each set to one canonical page, and if the site does not say which, Google decides. The audit reads the rel=”canonical” on each sampled page, follows the redirects, and checks that the sitemap agrees. Google’s own canonicalization guidance ranks the signals it listens to, a redirect and a rel=”canonical” annotation each count as “a strong signal,” while sitemap inclusion is “a weak signal,” so a site relying on the sitemap alone to settle its duplicates has not settled them.

Measure the Vitals on Real Visits

Before I open a robots file on a site I am auditing, I load the homepage on my phone and watch how long it takes, because that one load tells me more about how the site is run than the rest of the checklist. Google describes Core Web Vitals as “a set of metrics that measure real-world user experience for loading performance, interactivity, and visual stability of the page,” and its search documentation says it highly recommends site owners achieve good scores “for success with Search.” The three numbers are Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift. The audit reads them from field data, the Chrome User Experience Report that reflects real visitors on real connections, before it reads a lab tool, because a lab run on a fast laptop passes pages that fail for a shopper on a phone in a parking lot. Each failing page gets a cause attached, an unsized hero image, a third-party script blocking the main thread, a font swap that shifts the layout, so the fix list names the change and not only the metric. Across the client sites Content Cucumber audited between July and September of 2026, mobile Largest Contentful Paint failed on 11 of the 11 sites where it was measured, and desktop passed on most of them. A lab run on a laptop hides that gap. My read on why it keeps happening is that even a team that built the site mobile first stops looking at mobile over time, so the mobile site drifts while the desktop version everyone works on stays fast.

Check That the Structured Data Says What the Page Says

Schema markup gets validated twice, once for syntax and once against the visible page. Valid JSON-LD that describes a product price the page does not show, or an FAQ block the reader cannot see, passes the validator and still fails rich results, and Google has been removing that mismatch from its results for a while now, which Content Cucumber documented in why valid structured data still fails Google rich results. The audit covers Organization, Article, Product, FAQ, and BreadcrumbList as the baseline set, then the types the industry calls for, and the audit flags the pages where the markup is missing outright, since a product page with no Product schema is invisible to the shopping and answer surfaces that read it.

Follow the Internal Links to the Pages That Earn

Internal links decide which pages a crawler reaches and how much weight they carry when it does. The audit maps the click depth from the home page to each product and post, finds the orphans that nothing links to, and follows the redirect chains where a link points at a URL that points at another URL before it lands. A category page four clicks deep with the highest revenue in the catalog is a finding, and so is a blog post that ranks and links to nothing that sells. The product page audit runs the same triage from the conversion side.

Read the Security Headers

HTTPS is the floor, and the audit checks the certificate, the redirect from http, and the mixed-content warnings that break the padlock on a page that loaded one image over the old protocol. Then it reads the headers, HSTS, Content-Security-Policy, X-Frame-Options, and the rest, because they affect trust signals as well as safety, and a header that blocks a legitimate embed can break a checkout flow while it protects it.

What to Do With the List

The output is a report with each issue categorized by severity, critical for the fixes that go in the same day, high for the coming week, medium for the coming month, and low for the backlog, and most audits finish inside a week. For WordPress and Shopify sites Content Cucumber implements the fixes itself, and for Magento, BigCommerce and custom platforms the report carries the instructions a development team needs. After the fixes deploy the same checks run again, and monitoring stays on Core Web Vitals and crawl health so the next regression gets caught before the rankings do. When a merchant tells me the agency that built the site already ran a technical audit, I tell them a technical audit is a monthly job, weekly if they can manage it, and the one from launch day was for the site as it stood on launch day. The foundation post on technical SEO covers what to keep in place once the list is clear, and the Search Discovery audit adds the layer the technical audit stops at, whether the answer engines name you once they can read you.

Content Cucumber’s technical SEO audit starts with a crawl of the whole site and ends with a ranked fix list, and the first audit is free.

Let us help you rank higher with human-written SEO content

Our SEO Content Services →

Frequently Asked Questions

What does a technical SEO audit check?

A technical SEO audit checks whether a site can be found, read, and ranked before the copy is judged. In the order an auditor runs them, the checks are crawl access (robots.txt, sitemaps, server responses, and AI-crawler access), indexation against the published page count, canonicals and duplicate URLs, Core Web Vitals on real visits, structured data validated against the visible page, internal links and click depth, and HTTPS with the security headers.

Does a technical SEO audit cover AI crawlers like GPTBot?

Yes. GPTBot, ClaudeBot, PerplexityBot, and Google-Extended each read robots.txt for their own name, so the access check now covers whether the answer engines can reach a page as well as whether Googlebot can. A site that blocked those crawlers cannot be cited by the engines they feed.

How does an audit tell whether pages are indexed?

The Search Console Pages report lists what Google indexed and what it excluded, with a reason for each exclusion. The audit compares that count with the CMS's own count of published pages, and the gap points at noindex tags, staging settings that survived launch, pages discovered but not crawled, and parameter URLs that consumed the crawl budget.

Why does the audit read Core Web Vitals from field data instead of a lab tool?

Field data from the Chrome User Experience Report reflects real visitors on real devices and connections, while a lab run on a fast laptop passes pages that fail for a shopper on a phone. The audit reads Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift from field data first and attaches a cause to each failing page.

What does the technical SEO audit deliver at the end?

A report with each issue categorized by severity, critical, high, medium, or low, most audits finished inside a week. For WordPress and Shopify sites Content Cucumber implements the fixes; for Magento, BigCommerce, and custom platforms the report carries the instructions a development team needs, and the same checks run again after the fixes deploy.

Why Broad Search Queries Stopped Sending You Clicks

What an Email Copywriting Service Delivers

Leave a Comment