AI visibility, AEO and LLM SEO tools:
an honest comparison.

Four names, one category

Vendors sell the same product under four labels. An AI visibility tool, an AEO tool (answer engine optimization), a GEO tool (generative engine optimization) and an LLM SEO tool all do one job: ask the engines a fixed set of questions on a schedule, record who gets named and which pages get cited, and chart it. The label tells you which crowd the vendor is selling to, not what the software does.

The names that matter are the columns in the table below. Does it cover the engines your buyers use. Does it separate the prompts that contain your name from the prompts that do not. Does it show the sources the engine pulled from. Does it publish a price. Judge on those four and ignore the acronym on the pricing page.

The number most dashboards hide

Ask a model "tell me about Acme" and it will tell you about Acme. That is recall, and almost every brand scores near 100% on it. Ask "what should a 50-person SaaS company use for X" and the model has to choose. That is discovery, and it is the only number that describes buyers who have never heard of you.

Most dashboards run both kinds of prompt and report one blended score. The blend averages a solved problem into an unsolved one. A brand at 100% recall and 0% discovery shows up as "50% visible," which sounds like a marketing problem when it is a distribution problem. The methodology behind this page, with the four numbers a report should show, is in Your AI visibility number is lying to you.

One first-party example. On 10 September 2026, GTM Labs ran its own scorer on three unbranded buyer categories it sells into. Organic score on each: 0 of 10 answers. Branded recall on each: 5 of 5. A blended score would have reported 50% and a healthy trend. The honest read is that the engines know the name and never reach for it, and the citation lists said where they reach instead: linkedin.com was cited in all three categories. That is the gap this page is built to expose, and the reason the comparison table has a column for it.

Tool by tool

Which one, for whom

• You want the split and a low bill: Otterly.AI from $29 a month has Branded and Non-Branded filters on the dashboard. Peec AI classifies every prompt but does not publish its figures.
• You already pay Ahrefs: Brand Radar at $50 a month for custom prompts is the cheapest way to add the metric to a stack you have, and the 456M-prompt index is useful context even if the split is not stated.
• You want the widest engine list at a published price: Athena, 11+ engines, $295 a month or $252 on annual.
• You are enterprise and procurement asks about SSO: Profound or Scrunch. Both put citations at the center and both price on request.
• You want the number before you buy anything: the open-source scorer, or the free AI Visibility Checker. Either will tell you in an afternoon whether you have a discovery problem at all.

How to test any tool in ten minutes

1. Write five questions a buyer in your category would ask an assistant. None of them may contain your name or a competitor's name. "What is the best way to track Meta conversions on Shopify in 2026" qualifies. "Is Acme good" does not.
2. Paste each into ChatGPT with search on, and into Perplexity. Ten answers total.
3. Count the answers that name you. Count the ones that cite a page of yours. Write down every domain the answers cite.
4. Open the tool's dashboard for the same week. If it says 60% and your ten answers said zero, the 60% is recall wearing a costume.
5. Look for your ten cited domains in the tool's source report. If they are there, the tool sees what you see. If the report only shows your own score, you bought a thermometer that reads your temperature and nobody else's.

The scorer automates exactly this: it generates the unbranded questions, runs them per engine, and prints the organic share, the branded recall, who was named instead, and the supply chain, on four separate lines. It is on GitHub, MIT licensed, and a run costs cents.