Revenue BenchThe same email finder costs 5.4 times as much through Clay as through Treg.
5.4× the Clay price over the Treg price
Findymail gave the same answers through both stores. 1,000 right answers cost $113.66 through Clay and $20.91 through Treg.
Which email verifier, email finder or job-change tool is worth paying for?
We tested 83 vendors on records where we already knew the right answer, and we publish what we found.
Latest findings
Seven email verifiers are in the top tier.
7 of 25 verifiers tested
BounceBan through Treg scores 90 out of 100. It catches 91% of bounces and keeps 99% of good emails. The other six are too close to call.
Cheap decision models match general LLMs on three of four data-hygiene jobs.
3 of 4 jobs where the best decision model keeps up
On duplicate records, the best decision model is right on 79% of cases at $0.0123 per 1,000, and the best LLM on 78% at $1.1463. On own work email address the best LLM is right on 77% and the best decision model on 59%.
AI can't clean your CRM on its own yet.
17% on main employer for the best model
Allow at most 1 mistake in 100 and the best model clears 17% on main employer, 14% on duplicate records, 7% on same person and 4% on own work email address by itself. A person still has to check the rest.
Reports
Email verification, work email finding, job change detection and people search, tested on records where we already knew the right answer.
- Released
- 10 Nov 2026
- Jobs
- 4
- Tested
- 83 vendorsthrough 7 stores
- Spend
- $311.69
Jump to accuracy, accuracy and cost, store markup, speed and method.
Duplicate records, own work email address, same person and main employer, answered by 25 decision models, 8 LLMs and 1 web agent.
- Run
- 6 Nov 2026
- Jobs
- 4
- Tested
- 34 models25 decision models, 8 LLMs, 1 web agent
- Cases
- 702each asked 3 times
- Spend
- $8.53judge calls
Jump to right answers, guard share, cost, speed and method.
How we score
Every score comes with a likely range. Vendors we can't separate from the leader share the top tier, so pick on price, speed or where you already buy. Accuracy, cost and speed stay on separate charts, not one rank.
Run it on your own records
The kit tests vendors on your own records with a known outcome and prints a cost estimate before you spend anything. It ships with the revenuebench repository, which is not public yet.