Industries / SaaS
SEO for a SaaS company means making sure buyers, Google and AI tools can find the answers a purchase depends on: what it costs and on what basis, whether it connects to the tools they use, how to set it up, and how it compares. We checked 60 SaaS products by hand. 45 showed a real price, but only 17 also had readable help docs, comparison pages and integration information.
GetBorvis / Organic visibilityWhat it costs and on what basis. Whether it works with the tools they already use. How hard it is to set up. How it compares with the product they are leaving. And who else uses it. Most of those answers live on pages marketing does not own: the pricing page, the help centre, the integrations directory.
Google says its AI features use the same index as normal search, and that no special file or markup is needed to appear in them. So the work is not a new kind of SEO. It is making sure those five answers exist, can be read, and say the same thing everywhere.
Fact From Google's guidance on AI features and your website.
| On the website | Checked by hand (60 products) | Our checker (50 sites) |
|---|---|---|
| A public pricing or plans page | 54 of 60 | 48 of 50 |
| A real price shown on it | 45 of 60 | 41 of 50 |
| Help or docs pages readable without signing in | 38 of 60 (15 blocked our browser) | 46 of 50 loaded |
| Comparison or "alternative to" pages | 41 of 60 | 48 of 50 |
| Integration pages or a directory | 46 of 60 | 45 of 50 |
| Software structured data | 25 of 60 | 20 of 50 |
| robots.txt names any AI crawler | 9 of 56 sites | 10 of 49 |
| An llms.txt file | 38 of 56 sites | 32 of 50 |
Ours Two independent checks on 16 October 2026. By hand: 60 SaaS products in seven categories (CRM, accounting and invoicing, HR, scheduling, help desk, email marketing, project management) on 56 marketing sites, with every check recorded as found, not found, uncertain or blocked. A blocked page is not counted as missing. Our checker: 56 sites found through searches for pricing in seven categories, 50 answered. On the 27 sites in both samples, the two methods agreed on 30 of 30 AI crawler checks and 29 of 30 llms.txt checks. Both samples lean towards established, English-language products. Products are never named.
45 of 60 products showed a real price. In a separate look at five pricing pages chosen for their different billing models, every one showed a price, the unit it applies to, the billing period and a route to talk to sales. Public prices and sales conversations sit side by side. Spell out what the number means: per seat or per account, monthly or yearly, minimums, included usage and overage.
For 15 of 60 products, bot protection stopped our research browser from reading the help centre. 13 of those had readable integration pages, so the buying information and the setup information sat behind different rules. 19 of our 50 sites kept their docs on a separate address. Check what your docs host allows: a block meant for scrapers can also turn away search crawlers.
41 of 60 products had readable comparison or "alternative to" pages, and the median site in our checker's sample listed about 35 in its sitemap. Having them is no longer an edge. Being believed is: compare what buyers weigh, date the comparison, and correct it when the other product changes.
38 of 56 sites published an llms.txt file, the highest of any industry we have checked. Only 9 named any AI crawler in robots.txt. Ahrefs found that 97% of about 38,000 sites with a valid llms.txt got no requests for it in May 2026. Decide your crawler rules on purpose, for example search access versus training, before adding optional files.
Study Ahrefs: llms.txt study, 137,210 domains in its web analytics, May 2026, published 15 June 2026. Its customers skew technical, which it says makes its adoption figure an upper bound.
We check that each price says what it is for: the amount, the currency, per seat or per account, monthly or yearly, any minimum, what usage is included and what happens above it. A buyer and an AI answer can only repeat what the page actually states.
Buyers read setup guides before they buy. We check that your help centre can be read without signing in, that search engines are allowed to reach it, and that bot protection on the docs address is not turning away the crawlers you want.
Almost every SaaS site has them. We check that yours compare what buyers actually weigh, state when they were last checked, and stay fair enough to be believed. A one-sided comparison convinces nobody who has seen the other product.
A logo in a directory answers "does it connect?". It does not answer "can my team set it up?". We check that your most important integrations each have a page explaining what syncs, what plan it needs and where the setup guide is.
For a worked example, our audit of a flashcard app found a broken upgrade link and flashcard markup one property short, and all four priority fixes were live within two weeks. Why markup alone does not earn AI citations is in schema markup and AI citations, and which AI crawler does which job is in how AI decides which businesses to name.
We plan and specify the search work: what each page needs to answer, where it should live and how to measure it. Your team makes the changes. Prices, plan limits, integration behaviour and security claims are confirmed by the people who own them before anything is published. We do not write competitor claims we cannot source.
Send your website and the buying question you most want answered well. We will reply with where you stand against the products we checked, and the first thing worth fixing.
…it was worth gold!!Founder of Notum Learn, a SaaS company we audited