SEO and GEO for SaaS with many programmatic pages: how to measure if they work
How to tell whether programmatic SEO and GEO pages for a SaaS are working: index rate, traffic per template, AI citations and when to prune.
The short answer
You cannot tell whether programmatic pages work by looking at one page. You look at the template. For each template, track four things: how many pages are indexed, how many earn impressions and clicks, how many lead to signups, and whether AI engines cite them. If a template fails on most of those for several months, fix it or remove it.
What I mean by programmatic
A programmatic page is one that a template generates from a data set: an integration page for each tool, a comparison for each competitor, a use case for each industry. SaaS companies like them because one template can produce hundreds of pages that match specific searches.
They go wrong when the data is thin. A page with a title, an empty section and a call to action is not useful to anybody. Google’s guide to generative AI features says that more pages do not make a site better, and that creating separate pages for each variation of a search mainly to manipulate rankings or AI answers violates its spam policy on scaled content abuse. Source: Google Search Central, guide to generative AI features in Search.
Measure by template
Group your URLs by template, usually by URL pattern, and keep a small table for each one.
| Question | Where to look |
|---|---|
| How many pages exist? | Your CMS or sitemap |
| How many are indexed? | Search Console, page indexing report |
| How many get impressions and clicks? | Search Console performance, filtered by URL pattern |
| How many bring signups or leads? | Your analytics, by landing page |
| Do crawlers fetch them? | Server logs |
| Do AI engines cite them? | Your prompt set (see below) |
A practical trick: give each template its own sitemap. Search Console then shows submitted versus indexed per sitemap, so you can see which template Google is skipping.
Read the index status honestly
Two statuses deserve attention in the page indexing report: pages that were discovered but not crawled, and pages that were crawled but not indexed. A large share of either, concentrated in one template, usually means the pages look thin or duplicated to the search engine. It is feedback on quality, so resist the urge to submit them again.
Rules of thumb I use
These are my own judgment, not published thresholds:
- Give a new template 8 to 12 weeks before judging it, and look at the share of pages that get at least one impression, not the average.
- If most pages in a template are not indexed after that period, improve the content before generating more.
- If a template gets impressions but no clicks, the title and description are probably weak.
- If pages get clicks but no signups, the page may attract the wrong intent.
Quality gates before you publish
- Real data on every page. If a page has nothing specific to say, do not generate it, or mark it
noindexuntil it does. - A distinct answer. Two pages should not differ only by a name.
- A direct answer in the first lines, so it can be quoted.
- Structured data that matches the visible content, and a visible author or company.
- Internal links from related pages, and a clear path to the next step.
- Dates and facts you can maintain. Pricing and features change, so decide who updates them.
The GEO side
AI engines tend to quote pages that state a specific fact clearly. For a SaaS that means integrations, supported features, plans and limits, written plainly and kept current. Build a prompt set that includes questions your template pages should answer, run it regularly and record which URLs get cited. The method is in how to check if ChatGPT and Perplexity cite your product.
Pruning
When a template is not working, you have three options: improve it, merge it into a stronger page with a permanent redirect, or noindex it. Do one at a time and write down what you changed and the date, or you will not know what helped.
A simple review cadence
Look at each template monthly for the first quarter, then quarterly. If you want help setting up the measurement, or an outside review of a template before you scale it, the GEO audit covers it, and the multilingual guide covers the language side when your pages exist in more than one language.
Frequently asked questions
How do I know if programmatic SEO is working?
Measure by template, not by page. For each template track how many pages are indexed, how many get impressions and clicks, how many produce signups, and whether AI engines cite them. If most pages of a template are not indexed or never get impressions after a few months, fix or prune that template.
Is programmatic SEO against Google's policies?
Not by itself. Generating pages from real data that helps people is fine. Google's spam policy targets scaled content made mainly to manipulate rankings, such as near-identical pages for every query variation.
How many programmatic pages should I publish at once?
Start with one template and a small batch. Check that pages get indexed and that the content is useful before you generate more.
Do programmatic pages get cited by ChatGPT or Perplexity?
They can, when a page gives a specific, accurate answer that is easy to quote, such as a clear integration, pricing or comparison fact. Test with a fixed prompt set and record which URLs get cited.
When should I remove programmatic pages?
When a template stays unindexed, gets no impressions, duplicates another page or has thin or empty content. Merge, redirect or noindex them, and keep a log of what you changed.