Quick answer
Test buyer selection questions, then inspect shortlist inclusion, factual accuracy and recommendation reasons. Having a website or receiving a brand mention does not establish that your SaaS is considered for a buyer's requirements. Keep category discovery, comparisons and named fact checks separate, and retain the original answers.
See Edith for Winin's public description. Your product owners should confirm the software's suitability, deployment and integration facts.
Details
Start with the buyer's problem
An illustrative discovery question is “Which software options suit a sales team working across several markets?” Follow with the buyer's industry, deployment and integration constraints. Use named questions separately to check whether AI describes existing features correctly. Putting your brand into every question tests recognition rather than natural shortlist inclusion.
Record the exact question, use case, language, target market, model or product interface, search setting and sampling time. Mark conditions you cannot control as unknown. The retesting method explains how to keep comparisons interpretable.
Connect observations to evidence
| Observation | What to inspect | Suggested action |
|---|---|---|
| Missing from category answers | Public description of the problem and intended users | Clarify supported use cases with approved facts |
| Mentioned, but a competitor gets the reasons | The cited comparison dimensions and your corresponding evidence | Explain features and limits against the same dimensions |
| Wrong deployment or integration claims | Current pages, old documentation and channel information | Confirm the valid version before correcting content |
| Answers change with test conditions | Market, question, interface and search differences | Keep results separate rather than merging them |
These are investigation steps, not proven causes. A missing brand does not establish why it was excluded. Inspect visible citations where available; when none are shown, do not claim to know which page supplied the answer.
Ask for a report that supports action
A useful procurement sample should connect each finding to the exact prompt, interface or model, time, raw answer, visible citations and recommendation reasons. Separate mentions, shortlist inclusion and accurate recommendations using the stage definitions.
Confirm product facts before preparing page tasks for approval, then retest the same questions after verified publication. Agree the actual deliverables before buying. See Homer and Friday for Winin's public descriptions.
Common questions
Q: Why is our SaaS missing from AI vendor shortlists?
A: Start with the buyer's requirements and retained answers. Check whether your public facts explain those requirements accurately, while treating absence as an observation rather than proof of a content problem. One answer cannot represent every buyer.
Q: AI mentions our software but recommends a competitor. What should we check?
A: Inspect the reasons attached to the competitor, such as industry fit, deployment or integrations. Check whether your corresponding facts have clear, current public sources; do not infer what optimization work the competitor performed.
Q: What evidence should a SaaS team ask for in a GEO monitoring report?
A: Ask for traceable answers, the question set and conditions, separate shortlist and accuracy findings, and proposed fact checks and retests. Have each candidate demonstrate this using your agreed software selection scenarios.