Every SEO Loop audit boils down to one number between 0 and 100 — but that number is a summary, not the point. The point is the 35+ checks underneath it, grouped into four categories: SEO, performance, security, and best practices. This post walks through exactly what happens when you run one, how the score gets calculated, and what to actually do once you have a report in front of you.

What happens when you click "Run Audit"

When you submit a URL, SEO Loop makes a real, live HTTP request to that page — not a cached copy, not a stale index entry. It records the response time, reads every response header, and parses the returned HTML with a proper DOM parser rather than regex-guessing. From that single fetch, it runs every check in parallel against the same snapshot of the page, so the results are internally consistent — you're never comparing a check that ran against yesterday's version of the page with one that ran today.

If you choose Full Site Crawl instead of Single Page, that same process repeats across up to 20 pages, discovered by following internal links from your starting URL, with a small delay between requests so we're not hammering your server. The crawl also respects your site's robots.txt — if you've disallowed a path for crawlers, we won't audit it, the same courtesy any well-behaved bot should extend.

The four categories, and what actually lives in each

SEO covers the things search engines use to understand and rank a page: title tag presence and length, meta description, a single H1, canonical tags, whether the page is actually indexable (no stray noindex), heading structure, internal linking density, content length, image alt attributes, and whether you have a sitemap and robots.txt at all.

Performance covers response time, whether your server compresses responses with gzip or brotli, render-blocking scripts missing defer/async, image lazy-loading, HTML payload size, and — if you've added a free PageSpeed Insights API key in Site Audit settings — a real Lighthouse mobile performance score, not just our own heuristic.

Security covers HTTPS itself, HSTS headers, mixed content (HTTP resources loading on an HTTPS page), your SSL certificate's validity, and a set of response headers that matter more than people think: Content-Security-Policy, X-Content-Type-Options, Referrer-Policy, and clickjacking protection via X-Frame-Options or a CSP frame-ancestors directive.

Best practices is the catch-all for things that don't fit neatly elsewhere but still matter: a mobile viewport meta tag, character encoding declaration, favicon, HTML5 doctype, Open Graph tags for social sharing, Twitter Card tags, and structured data (JSON-LD).

How the score is actually calculated

Each check carries a weight — high, medium, or low importance — reflecting how much it actually affects rankings, security, or user experience. A missing title tag (high) hurts your score far more than a missing favicon (low). The final score is: sum of weights for checks you passed, divided by the sum of all possible weights, times 100. This is why two sites can both "fail three checks" and land at wildly different scores — three high-importance failures will tank a score much faster than three low-importance ones.

A score of 60 with three high-importance fixes is more useful to know than a score of 95 with none — because it tells you exactly what to do next, in the right order.

Reading a report: what to fix first

Start with anything marked high-importance. Those are weighted heaviest for a reason — they're the checks most directly tied to whether your page can be found and trusted at all. A missing canonical tag or a page that's accidentally set to noindex will do more damage than a dozen minor issues combined. Work through high, then medium, then low. Don't chase a perfect 100 — chase "no high-importance issues left," which is a realistic and meaningful target for almost any real site.

For a full-site crawl, the report groups issues differently: instead of a flat checklist, you get an issue-by-issue view showing exactly which pages are affected. A missing meta description on one page is a five-minute fix. The same issue on every product page on an e-commerce site is a template bug — and the crawl report is what makes that distinction visible in the first place, which a single-page audit never could.

What the score doesn't tell you

It's worth being honest about the limits here. A 100/100 audit score doesn't mean you'll rank #1 — it means the technical foundation is solid. Rankings are also a function of content quality, backlink profile, competition, and search intent match, none of which a technical audit can measure. Think of the score as a health check, not a ranking guarantee: it tells you whether anything is actively working against you, not whether you'll win.

Using the AI report on top of your results

If you've added your own API key for ChatGPT, Claude, Gemini, or DeepSeek in Profile Settings, every audit report gets a "Generate report" option. This doesn't just re-read the checklist back to you — it looks at your specific pattern of passes and failures and writes a prioritized, plain-language action plan: what to fix first and why it actually matters for your specific site, quick wins you can knock out in minutes, and longer-term suggestions that go beyond what any fixed checklist could cover. It's particularly useful if you're going to hand the report to someone non-technical, like a client or a stakeholder who doesn't want to parse 35 checkboxes.

Common surprises on a first audit

A few things trip up almost everyone the first time they run an audit. Missing HSTS headers is extremely common even on sites that have had HTTPS for years — it's a header nobody thinks to add manually, and most frameworks don't set it by default. Missing canonical tags show up constantly on sites built without a strict CMS workflow. And render-blocking scripts are nearly universal on older WordPress sites loaded with plugins, each adding its own unoptimized script tag to the page head.

A note on re-running audits over time

If you label your audits with a project name, SEO Loop groups repeat scans of the same project on your dashboard and shows you a score trend over time — not just a single snapshot. This is genuinely more useful than a one-off audit, because it turns "is this fixed?" into an answerable question instead of a guess. Re-audit after any deploy that touches your templates, headers, or hosting configuration; that's when regressions actually happen.

Run your first audit now at seoloop.in/site-audit — it's free, takes under a minute for a single page, and doesn't require you to hand over a credit card to find out what's wrong.

Reading category scores in isolation

It's worth looking at your four category breakdowns separately, not just the overall number, because a single blended score can hide a real problem. A site can post an 85 overall while its security category alone sits at 40 — dragged up by a clean sweep on SEO and best-practices checks. If you're specifically worried about one dimension (say, you just migrated hosting and want to confirm nothing broke on the security side), check that category's sub-score directly rather than trusting the headline number to reflect it.

Single audits versus ongoing monitoring

A one-off audit is a snapshot: it tells you the state of a page or site at the exact moment you ran it. That's genuinely useful before a launch, after a migration, or when investigating a sudden traffic drop. But technical SEO health isn't static — a plugin update, a CDN configuration change, or a new page template can silently introduce a regression weeks after everything looked fine. This is why labeling your audits with a consistent project name matters: it turns a series of one-off snapshots into an actual trend line on your dashboard, and a sudden score drop between two audits is a much stronger signal than either audit alone.

Frequently asked questions

Does a low score mean Google has penalized my site? No — a technical audit score and a Google manual action are unrelated. A low score means there are fixable technical issues; it says nothing about whether you've been penalized, which would show up in Google Search Console, not here.

Why did my score change without me editing anything? A few checks depend on live, external state: your SSL certificate's remaining validity, whether a third-party script your page loads has changed, or whether your hosting provider altered a response header. It's also worth checking whether a CDN or caching layer served a different version of the page than what you expect.

Can I audit a page that requires login? No — the audit fetches the URL the same way a normal visitor or search crawler would, without any authentication, so login-gated content can't be evaluated this way.

Is there a limit to how many audits I can run? No hard limit for reasonable personal or small-business use. A full-site crawl checks up to 20 pages per run and includes a small delay between requests out of courtesy to your server.

One last practical note: bookmark the report page once it's generated, or better, log in before you run the audit so it saves to your dashboard automatically. A logged-out single-page audit still runs and shows you the full result, but without an account it won't be saved anywhere for you to compare against next time — and comparison over time is where a lot of the real value in auditing regularly actually lives.