How to choose an AI SEO tool: 10 questions for every demo

GEO tool demos all look alike: a dashboard with charts going up. What separates them is what they do not show. Here are 10 questions worth asking in any demo, the three must-haves first, and how Lumen AI answers each one so you can use it as a yardstick.

The 3 must-have questions

1. 01
   Which AI engines does it track?: Ask for the exact list and what each plan includes. ChatGPT and Gemini are the floor; Perplexity and Google AI Overviews add to it. In Lumen, ChatGPT and Gemini come with the first paid plan and all 6 engines are in Agencia Full Coverage.
2. 02
   Are the results from the real app or the API?: A model’s API and the app people use can recommend different brands. Ask how the answer is obtained. Lumen reads the logged-out web experience of chatgpt.com and gemini.google.com and uses the API only as a fallback. Below we explain why it matters.
3. 03
   Which of my URLs get cited?: Knowing you are mentioned is not enough: you need to see which sources each engine cites and whether they talk about you. It is the basis of any authority sources plan.

Why does real answer vs API matter?

A monitor that only calls the API measures how the model answers in a development setting. People use the app, with its web search layer and its sources. In a test of ours with 48 runs, the top-5 brand overlap with the real web was 0.55 when reading the web experience and 0.37 with the API. It is not an industry benchmark, it is what we measured, so ask any vendor for an equivalent comparison. Details in real answers.

What else should you ask about prompts, frequency and competitors?

• Your own prompts: write your own, accept suggestions and test them before spending quota. Lumen has ranking, sentiment and precision prompts, and a free test mode (prompt tracking).
• Repeat runs: models vary from run to run, so a single snapshot is not enough. Ask how often it re-runs and whether you can choose. In Lumen, daily, weekly or monthly cadence is set with the team when you sign up.
• Competitor view: share of voice, average position and a per-prompt matrix, not just your own number (competitors).

How do you know AI is right about your brand?

A mention can be wrong. Look for a tool that checks whether AI knows what you offer (precision prompts) and lets you review each claim made about your brand. In Lumen each idea can be marked true, false or in progress. It is part of AI visibility that measures reputation, not just presence.

What happens after the diagnosis?

A tool that only shows numbers leaves the work to you. Ask whether it tells you what to fix, generates the content and checks afterward whether it worked. Lumen proposes GEO Actions with ready-to-publish content, notifies indexers and checks the result 14 days later.

Does it fit your stack, and how does pricing scale?

Ask for PDF and CSV for your reports, an MCP server or an API for your team, and Search Console and GA4 connections (integrations). And check how it charges: per prompt, brand or seat. A prompt pool shared across clients, with no per-seat fee, scales best for an agency.