A SaaS site is usually two sites sharing a domain. The marketing pages — home, features, pricing, comparisons — are few, hand-written and carry most of the impressions that lead to a trial. The documentation, changelog and help centre are many, generated, and carry long-tail queries from people who are already customers. An audit that sorts by severity fills the top of the list with the second kind, because there are more of them.
The searches that matter are category searches ('invoicing software for freelancers'), comparison searches ('X vs Y') and, increasingly, the same questions asked to ChatGPT and Perplexity, where the answer is a list of names and the only question is whether yours is on it. A SaaS company that ranks third for its category and is unnamed by every assistant has a measurable problem, and it is a different one from the ranking.
The other thing SaaS sites do is deploy constantly. Every release is a chance for a template to lose its canonical or pick up a staging noindex, which is why catching regressions after a deploy matters more here than anywhere.
The findings that recur on this kind of site
- Docs pages with the product name as the titleThe help centre template that titles every article 'Help — Product'. It is one duplicate-title cause across hundreds of pages, and it is fixed once in the template.
- Noindex that shipped from stagingA noindex-page finding appearing on a marketing page the week after a release is almost always an environment flag, and it costs the page its impressions until somebody notices.
- Comparison pages that canonicalise to each other'X vs Y' and 'Y vs X' pointing at one URL, or at neither. Canonical conflicts on the pages built to catch comparison searches.
- Feature pages nothing links toLaunched from a campaign, never added to the nav. Orphan pages among the pages that were supposed to sell.
- Thin changelog entries in the indexHundreds of three-line release notes, each a thin-content finding, competing with the pages that matter for the crawler's attention.
What to change first, and what to compare it against
- Fix the marketing template first, then the docs templateRank by impressions, not by count. Twenty marketing pages with real traffic outrank two thousand docs pages with little, however many findings the docs generate.
- Judge a title fix by click-through rate, a linking fix by impressionsChoose the metric before the release. Click-through rate for anything the searcher sees; impressions or position for anything the crawler sees.
- Compare against the pages you did not touchFeature pages you changed against feature pages you did not, matched by traffic. A control set frozen when the release lands, read four weeks later. Without it, a launch week and a fix are indistinguishable.
- Track the assistants with the same disciplineA fixed set of category questions, asked weekly. Whether ChatGPT names you is a rate over weeks, and who it names instead is a list you would not otherwise have.
Questions from people running one
Should the docs be on a subdomain or a path?
A path keeps everything in one Search Console property and lets the docs pass authority to the marketing pages through internal links. A subdomain separates them cleanly and makes the docs template's findings easier to ignore. Neither is wrong; what matters is that the marketing pages are ranked and measured on their own, whichever you choose.
We ship daily. Does that make measurement impossible?
It makes a control set essential. If everything changes every day, the only way to attribute a result to one change is to compare the pages it touched against comparable pages it did not, over the same window. That is what the ledger does for every landed fix, and it is the reason a daily-deploy site can still produce a verdict.
Everything above is the method, usable with any crawler and a Search Console export. Rankelle runs it on a schedule: the crawl, the grouping into causes, the watch for the fix landing, the frozen control set and the verdict four weeks on — including the verdicts that say the work did nothing. One site is free, permanently, because the first verdict takes six weeks to arrive.
Try it on your own site, free
The whole audit of every website you add, no card and no expiry. Pay only when you want the proof.