Methodology
How we test AI cover letter tools
Every score on LetterBench comes from hands-on testing, not vendor claims. Here is exactly what we do with every tool, and how each score is built.
The three test passes
Apply with a real job
We feed each tool the same set of real job descriptions and generate letters. We compare structure, specificity, and how naturally the writing reads — the things a hiring manager actually notices.
Check the ATS basics
Formatting, keyword handling, file export, and whether the output survives the kind of scan an applicant tracking system would run. A great letter that breaks in ATS is not a great letter.
Weigh price against value
We test the free tier and the paid plan. We look at real pricing, what changes when you pay, and whether the upgrade is worth it.
How a score is built
Each tool gets a score out of 10 in four dimensions. The weights are public — you can see exactly what goes into every number.
| Dimension | Weight | What we grade |
|---|---|---|
| Output quality | 40% | Writing that sounds human, is specific to the job, and has a clear structure. No template clichés. |
| ATS compatibility | 25% | Safe formatting, sensible keyword handling, and clean export to PDF, DOCX, and plain text. |
| Value for money | 20% | Free tier limits, real subscription price, and what actually changes on the paid plan. |
| Ease of use | 15% | How fast it takes to go from paste-job to finished letter, and how easy it is to edit the result. |
Our honesty rules
- No paid scores. Tools cannot buy a higher rating, a review slot, or placement in our rankings.
- No scores without testing. If we have not run a tool, it is marked "testing" or "review coming soon" — never given a number.
- No reviewing our own products. LetterBench never reviews tools made by its own team. That would be a conflict of interest.
- Changes are logged. When a tool changes and we re-test, the review is updated with a date so you can trust what you are reading is current.