Evidence
SaaS GEO case studies
GEO is new enough that most published "case studies" are unverifiable. We would rather show one result we can stand behind than a wall of logos. Below is the case we have measured end to end, followed by how to judge anyone else's numbers — including ours.
finopsllm.com — 2,300+ Copilot citations in two months
Disclosure: finopsllm.com is our own property, which is exactly why we can publish the full method rather than a headline number. It went from no measurable AI citations to over 2,300 Microsoft Copilot citations in roughly two months.
What changed, in order of impact:
- ✓ Definition-first pages. Every article opens by answering its own title in one extractable sentence.
- ✓ Structured data on every page. Article and FAQ schema, so claims survive being parsed out of context.
- ✓ Crawler access. An explicit robots policy and llms.txt, so AI crawlers were permitted and pointed at the right pages.
- ✓ Publishing depth. Sustained topical coverage rather than a handful of landing pages.
The full write-up, including the measurement window and caveats, is on the case study page.
How to read any GEO case study
Citation counts are easy to inflate and hard to audit. Before you believe a number — ours included — check:
- ✓ Is the baseline stated? "2,300 citations" means nothing without the starting figure and the window.
- ✓ Which assistant, and measured how? Copilot, ChatGPT and Perplexity report differently and are not interchangeable.
- ✓ Does the count include backfill? Some dashboards retro-fill historical data, which can make a flat period look like growth.
- ✓ Is the property disclosed? A vendor's own site is legitimate evidence, but you should be told that is what you are looking at.
- ✓ Did citations convert? Being cited and being visited are different outcomes; ask which one is being reported.
More case studies will be published here as results become measurable. If you want the playbook behind this one, start with GEO for SaaS or the methodology.
Frequently asked questions
Are there real GEO case studies for SaaS?
Few that can be verified. GEO is new enough that most published case studies omit the baseline, the measurement window, or which assistant was measured — which makes the headline number uncheckable. The one we publish is our own property, finopsllm.com, disclosed as such so the full method can be shown rather than just the result.
What did the finopsllm.com case actually change?
Four things, in order of impact: every article opens by answering its own title in one extractable sentence; Article and FAQ schema on every page so claims survive being parsed out of context; an explicit robots policy plus llms.txt so AI crawlers were permitted and pointed at the right pages; and sustained topical publishing rather than a handful of landing pages.
Why should I trust a vendor's case study about its own site?
You should not trust it more than any other — you should be able to check it. That means the baseline is stated, the window is stated, the assistant is named, and the ownership is disclosed. Ours is disclosed for exactly that reason. Apply the same five tests to anyone else's numbers, including a competitor's.
How long did the results take?
Roughly two months from no measurable AI citations to over 2,300 Microsoft Copilot citations. That timeline reflects a browse-and-index surface picking up newly structured pages; it is not a promise, and it says nothing about how quickly answers drawn from model training data would change.
What is a realistic starting point for a SaaS company?
Measure before you change anything. A baseline of how often each assistant names you today, on the prompts your buyers actually ask, is the only thing that makes any later number meaningful — and it takes less time than writing the first page.