In short: this is a walk-through of the RankThat site audit for people who will use it every month: agency SEOs, in-house marketers and business owners who look after their own site. It covers what happens when you start an audit, how the crawler behaves, what each tab of the report shows, how the health score is worked out, and where the audit stops and your judgment takes over. If you want a quick look first, the free Website SEO Checker runs a one-page check with no account.
Most SEO site audit tools stop at a list of problems. The list is useful, but somebody still has to work out which problems cost money, write the fixes and put them live. RankThat’s audit is built around that second half of the job, so the report is organized as a plan, and the most common fixes can be drafted and applied without leaving the app.
Starting an audit: mini or full
Open a project, go to SEO › Site Audit and click New audit. You get two choices.
| Mini audit | Full audit | |
|---|---|---|
| Pages crawled | Up to 50 (25 by default) | Up to 500 (200 by default) |
| Typical time | 1 to 3 minutes | Longer, depending on site size |
| Good for | A quick health check, a sales call, a single section of a site | The monthly audit of a client site |
| Plans | Every plan | Professional and Enterprise |
You can set the number of pages to crawl, and add a “Prepared for” name that appears on the PDF cover. Every audit counts toward your plan’s monthly audit allowance (see pricing).
You don’t have to remember to run it. Every project with Search Console connected or keywords tracked gets an audit automatically on the 1st of each month (a full one if your plan includes it), unless the site was already audited in the previous 20 days. That audit feeds the monthly client report, so the numbers in the report match the state of the site.
How the crawler behaves
RankThat crawls as RankThatBot and says so in its user agent, with a link to a page where site owners can check who it is (an honest user agent is how site owners tell good bots from bad ones). It reads robots.txt and follows the rules written for RankThatBot (or the general rules if there are none for it). It requests at most four pages at a time, never submits forms, never logs in, and never requests add-to-cart, wp-login or wp-admin links. Those rules matter more than they sound: a crawler that hammers a small WooCommerce store or fills carts with phantom products creates problems for the client.
When a firewall blocks it
Plenty of sites sit behind Cloudflare, Sucuri or a security plugin that blocks unknown bots. RankThat never tries to get past bot protection: no proxies, no fake browser identities, no challenge solving. Instead:
- WordPress sites with the free Bridge plugin send their pages from inside WordPress. The audit shows “Firewall detected” and waits for the site to send its pages, usually within five minutes. Speed and UX scores come from Google Lighthouse.
- Other sites are audited from public archives (Common Crawl, then the Wayback Machine). Every finding is dated so you know how old the evidence is.
- You can also allowlist RankThatBot so the next audit crawls normally. The robots.txt checker shows whether your rules let it in.
What the report shows
The report has seven tabs. Here is what each is for and when you would open it.
- Action plan. Start here. A banner tells you how many issues AI can fix for you, then “Progress since the last audit”, the top priorities, a 30/60/90-day plan, quick wins and “What’s working” (useful in client meetings, because it is not all bad news).
- Issues. Every problem found, filterable by severity (Critical, Warning, Notice) and by category. Open an issue to see “Why it matters”, “How to fix” and the affected pages.
- UX & Conversion. Checks that affect whether visitors call, book or buy, such as a hidden phone number or no clear call to action on mobile.
- Site health. The score and how each category contributes.
- Speed. Lighthouse results for mobile and desktop. Each device is tested three times and the middle result is kept, so one slow run doesn’t swing the score.
- Authority. Link-based strength, shown separately.
- Pages. Every crawled page with its own issues.
The checks cover crawlability, on-page basics, content, links, speed and Core Web Vitals, security, structured data and AI readiness (whether AI crawlers are allowed and whether the site has an llms.txt file).
How the health score is calculated
The headline score is a weighted average of the category scores. Crawlability carries the most weight, followed by on-page and content, then links and security, with structured data and AI readiness weighted least. Speed and authority are shown as their own scores rather than mixed in, and UX always stands alone. Grades follow the school pattern: A from 90, B from 80, C from 70, D from 60.
Two details agencies tend to appreciate. First, a client project’s score is never artificially capped, so improvement shows up honestly. Second, each run is compared with the last: the score change, the change per category, and lists of fixed issues, new issues and issues that are improving.
A worked example: Coastline Plumbing
Coastline Plumbing in Austin, TX has a 60-page WordPress site. The agency runs a full audit at the start of the engagement.
The Action plan opens with three priorities:
- Eleven service pages share near-identical titles (“Plumbing Services | Coastline Plumbing”). Google can’t tell the water heater page from the drain cleaning page by title alone, and searchers can’t either.
- Four old URLs return 404 but still receive internal links from blog posts written two years ago.
- No phone number is tappable on mobile. On a plumbing site, that is lost calls, not a technicality.
The banner says AI can fix most of these. The agency clicks Fix all with AI, which produces one review screen for every fixable issue, up to 50 pages per run. Each page gets one combined new title and description, the 404s get redirects to the closest live page, and the internal links that pointed at redirects are updated. The SEO lead edits two titles that don’t sound like the client, unticks one redirect she wants to discuss first, and clicks Apply fixes to WordPress.
About 20 seconds after each change, RankThat re-checks the live page and marks it “Fixed · checked live”. The tappable phone number and call bar are left out of Fix all on purpose, because they need choices (which number, which pages), so she sets those up separately through the WordPress Bridge.
At next month’s audit, “Progress since the last audit” shows the 404s and duplicate titles as fixed. The monthly report lists them under work completed without anyone typing them in.
What the audit does not do
Being clear about limits saves arguments with clients later.
- It doesn’t sign in. Pages behind a login are flagged as “Pages behind a sign-in” so you know they weren’t checked. Fixes for them are drafted from what the Bridge plugin sent.
- Automatic fixes need WordPress. The audit and every recommendation work on any site. On other platforms, Fix with AI drafts the change and you copy it into your CMS.
- It won’t make editorial calls. It can tell you two pages compete for the same search. It can’t tell you which one the business would rather sell from.
- It isn’t a replacement for Search Console. Whether Google has actually indexed a page comes from Google, which is why the audit has a separate tab for it.
The Google index tab
The third tab on the Site Audit page, Google index, uses Search Console’s URL Inspection data to show which important pages are in Google and why the others aren’t. Pages are rechecked at most weekly and automatically each month, and you can press Check pages now. It lists “Why pages aren’t indexed” with advice for each reason, “Pages not in Google”, and “Google picked a different main page” (a canonical mismatch). If a page drops out of the index, you get an alert.
How agencies use it each month
| When | What to do | Where |
|---|---|---|
| 1st of the month | The automatic audit runs. Nothing to do. | Site Audit |
| First week | Read the Action plan and “Progress since the last audit”. Run Fix all for titles, redirects and schema. | Site Audit |
| First week | Check new critical issues. An alert fires when they appear. | Alerts |
| Mid-month | Review Google index for pages that dropped out. | Site Audit › Google index |
| Report time | Confirm fixes appear under work completed. | Monthly Report |
The audit also connects to the rest of RankThat. Every fix is recorded and measured in Opportunities under “Results of changes” from day 14, so you can see whether the new title earned more clicks. And if you use Claude, the Claude connector can read an audit and draft fixes for you to approve.
Tip: run a mini audit on a prospect’s site before a sales call. The co-branded PDF and the “What’s working” list make a better first conversation than a scary error count.
For a manual method you can follow without any tool, read the guide on how to do an SEO audit. Agencies running many sites can see how audits fit into a portfolio on the agencies page.