Method
How the comparison is made
Each tool is read on its own public pages, scored on five parts of the loop from search to revenue, and every cell keeps the link and the day it was checked. This page sets out the rules, the limits and how to ask for a fix.
What Is Compared
The parts follow a case from search data to a verified result. A tool can be strong in one part and absent from another; the table shows each part apart.
Revenue
Does the tool report revenue from search and AI assistants, and does it say how each figure was measured?
Decisions
Does it turn its data into ranked work, each item with its evidence and its value?
Governed Execution
Can it change a site, and under what approvals, records of what changed and ways back?
Verification
Does it measure a change against what would have happened without it?
Agency Fit
Can an agency run many clients in it, and what is it metered by?
Scores and Evidence
The score says how much of a part the tool’s own pages describe. The evidence says what the score rests on. Neither is a rating of the tool.
Score
- Full
- The part is described in full on the tool’s own pages.
- Partial
- Some of the part is described, or it is described without a mechanism.
- None found
- Nothing was found on the pages read. Not proof that it is absent.
- None or partial
- The pages read were unclear.
“Not stated” means the matrix has no sourced finding for that part, so the cell is left empty rather than guessed.
Evidence
- DDocumented
- Docs, a help centre, a changelog, or a feature page that describes how it works.
- MMarketing claim
- A marketing claim without a described mechanism.
- 3PThird party
- A third party or a search snippet only.
The Rules
Chosen by what they sell
The tools are AI visibility trackers, SEO suites, enterprise SEO platforms, automation and testing tools, revenue attribution and agency reporting tools whose public pages describe at least one part of the loop. Y Combinator’s SEO category was scanned for small companies; Product Hunt, G2 and Crunchbase were not browsed directly.
Read on the vendor’s own pages
Docs, help centres, changelogs, feature pages and pricing pages, each read on a stated day. A third party is used only where the vendor’s page could not be read, and is marked as such.
Dated, and re-checked
Every cell was checked on 25 October 2026, and the five tools with head-to-head pages were read again on their live pages on 6 November 2026. Each fact is re-read at least every 120 days, and every 30 days for Profound and Peec AI, with a monthly sweep for new products. A row past its window shows “Re-check due” and its page is withdrawn from search until it is read again.
Prices as the page showed them
A price is the one the vendor’s pricing page showed on the day it was read, with its billing where the page said. Where a vendor publishes none, the table says so and no estimate from elsewhere is shown.
Paraphrased, never quoted
Many pages were read through a summarising fetch, so the tables paraphrase them and mark the prices read that way. Follow the link to read the vendor’s own words.
Our own column, marked
QueryRev’s column is our description of our own product, each line linked to the page of this site that describes it. It is not scored by this method, and every table marks it as ours.
Limits
- “None found” means not found on the pages read, never shown to be absent.
- Enterprise vendors keep pricing and behaviour behind demos, so their cells rest on less.
- Private betas, enterprise demos and unannounced products are not visible to a public survey.
- Some pages could not be opened on the day; where a cell depends on a snippet or a third party, it says so.
- Products change quickly. A dated cell is true of the day it was read, which is why each one shows that day.
Corrections
Write to research@queryrev.com with the cell, the page of your site that shows the correct fact, and its date. We read that page again, correct the cell with the new check date and record the change. The same rules apply to every tool, ours included.
Keep exploring