What we do and don't do
We compare published specs and documentation. For every product we read the vendor's pricing page, help centre and product documentation, record what they say in one dataset, and link each price to its source with the date we checked it. Every review on this site was built that way.
We don't claim hands-on testing we didn't do. None of our current reviews comes from running the tools side by side, and each one says so. If we test something in future, the page will say what we tested, when, and how.
We don't take money from vendors. No affiliate commissions, sponsored placements, paid reviews or free upgrades in exchange for coverage. There are no ads on this site.
We do own two of the businesses that appear here: cite-met, a hosting and AI analytics platform, and Space & Story, a web studio. See how we handle our own products below.
The rubric
Every third-party AI visibility tool is scored 1 to 5 on six criteria, weighted for a small business buying its first tool. The weighted average, to one decimal place, is the score shown on the review and in the reviews table. Each review explains every score in a sentence.
| Criterion | Weight | What we ask |
|---|---|---|
| Cost for one small business | 25% | What does a single business pay for a setup that answers its questions, and what does each tracked prompt cost? |
| Engine coverage at entry tier | 20% | Which assistants does the cheapest useful plan track: ChatGPT, Google AI Overviews and AI Mode, Perplexity, Copilot, Gemini, Claude? |
| Data method and freshness | 15% | Does the vendor document how it collects answers (consumer interface or API) and how often it refreshes them? |
| Insight and next steps | 15% | Beyond counting mentions, does it show which sources are cited and what to change: audits, recommendations, content help? |
| View of your own site | 10% | Can it show which AI agents read your pages (crawler or agent analytics), not just what assistants say? |
| Self-serve and trial | 15% | Is the price published, is there a free trial or plan, and can a small team start without a sales call? |
A different reader would weight these differently. An enterprise team would care less about cost and self-serve and more about engine coverage and insight, and Profound would rank higher. We publish the per-criterion scores so you can re-weight them for yourself.
What the scores mean: 5 is the best we found in the category on that criterion; 1 means the product doesn't offer it or doesn't publish enough to judge. "Doesn't publish" scores low on purpose: if a vendor won't say how it collects data or what it costs, a small business can't evaluate it either.
Where our data comes from
- Prices and plans: the vendor's own pricing page, or its help centre where the pricing page is incomplete. Monthly billing in US dollars unless stated. Every price table links to its source and says "checked" with a date.
- Features and methods: the vendor's documentation and help centre. Where a vendor's own pages contradict each other, we say so on the page.
- Vendor comparison pages (such as Profound's and Peec's pages about each other) are used only for context and labelled as the vendor's claims. Where they conflict with a pricing page, the pricing page wins.
- Studies: we cite the original study (Ahrefs, Surfer, Vercel and others), with its sample size and date, and note when the company publishing it sells a related product.
- Our own measurements: the AI Search Observatory is the only measured data on this site. It's aggregated from small-business sites hosted on cite-met under fixed privacy rules: every figure covers at least five sites and no site is identifiable. Details are on the About page.
All product data lives in one dataset, so a price changed in one place changes on every review, comparison and table at once.
Our own products
cite-met appears in our reviews and comparisons because it's a real option for some readers. When it does:
- The page carries a disclosure box at the top and says the relationship again in the body text.
- We don't score it. A score from its owner would be a self-serving rating, so its review shows the same facts as every other product, with no number and no rating markup for search engines.
- It gets real weaknesses: it's new, it's a hosting platform rather than a standalone tracker, and it doesn't do prompt-level rank tracking like Profound or Peec AI.
- We never rank it first by default, and every page it appears on names who should pick something else.
- We don't put it in head-to-heads with prompt trackers, because the comparison would flatter it.
Space & Story is mentioned only where a reader might need a studio, and never in rankings.
How often we update
- Prices and plans: rechecked at least quarterly with each Observatory edition, and sooner when we learn of a change.
- Crawler registry: rechecked against each operator's documentation every quarter.
- The "Updated" date on a page changes only when its content changes. Every change is logged on the Updates page and in our RSS feed.
Corrections
If something here is wrong, or you work for a vendor and our description is out of date, tell us through Space & Story and include a link to the source. We fix verified errors, note the change on the Updates page, and move the page's "Updated" date. We don't change a score because a vendor asks; we change it when the facts behind it change.
Who does this
Three Impact is written by Mahmoud Halat, founder of Space & Story.