We ran one resume through eight ATS checkers. It scored 70, 72, 80 and 85.
Same PDF, same afternoon. Here is what each tool measured, what it got wrong, and what none of them caught on a resume built to trip every parser.
The setup: two resumes, eight tools
We built two PDFs. The first is a clean, single-column resume for a Java developer with three and a half years at two well-known companies: standard headings, dates on every role, a metric in six of seven bullets. The second is the resume most Indian freshers actually have: two columns, a photo, contact details in a coloured header band, a 10th/12th marks table, skill bars, father’s name, date of birth and a signed declaration.
Then we uploaded both to every tool on the first page of Google India for “ats resume checker”, on a desktop and again at phone width, and wrote down exactly what each one showed before it asked for anything.1
| Tool | Clean resume | Time to score | What stood in the way |
|---|---|---|---|
| 1MillionResume | 70 | ~6 s | One run a week; “critical” evidence paywalled |
| MyPerfectResume | 72 | ~17 s | Findings are counts with no text |
| Enhancv | 80 | ~20 s | 14 of 27 checks locked; one free report |
| cxolanes | 85 | ~6 s | Ads over the button; findings change between runs |
| Naukri | “High” | ~4 s | No number, no findings |
| GoodSpace, Resume Worded, Jobscan | — | — | OTP or account wall before anything |
The invented findings
Enhancv told us the file contained the misspellings “projekts” and “crickets”. Neither word is in it. The fix was a Pro upgrade. 1MillionResume flagged “Personal pronoun detected, Critical” and hid the evidence behind an upgrade button; the file contains no pronouns. cxolanes, on its second run of the same file, advised adding a date of birth.
A score you cannot check is a sales number. Every finding on this site shows the text or the region of your file it came from. If we are wrong, there is a button to say so.
The hazards nobody named
On the second resume, the one designed to break parsers, no tool mentioned the two columns, the photo, the contact band, the marks table or the skill bars. MyPerfectResume re-rendered “Father’s Name” and “Marital Status” back to us without comment. The one thing every checker claims to test is the one thing none of them tested.
Why the numbers differ, honestly
Each tool weights something different: Enhancv leans on design and content, Jobscan on keyword overlap with a job description, 1MillionResume on a word-count band, cxolanes on whatever a language model felt like that minute. None of them publish the weights. Two of them are not deterministic, so the same file can move between runs.
What a score should measure instead
Parse fidelity (will the structured fields be right), findability (will a recruiter’s search surface you) and readability (will a human get it in seven seconds). Weighted, published, the same every time, with the extracted text on show. That is what this site does, and the rubric is on the How we score page.
-
Full log with screenshots: our hands-on test notes, 12 September 2026. Parser behaviour claims: Greenhouse “Unsuccessful resume parse” help article; Textkernel, “Improving extraction from column resumes”. ↩
Questions this article answers
Which ATS checker is the most accurate?+
None of the eight we tested showed its working, so accuracy cannot be judged from the number. Judge a checker by whether it shows the text it extracted and names the rule behind each finding.
Why does my resume get different ATS scores on different sites?+
They measure different things with different weights, and most do not publish either. One clean resume scored 70, 72, 80 and 85 on four page-one tools in one afternoon.
Is a two-column resume bad for ATS?+
Risky, not fatal. The market-leading parser renders about 90% of column CVs correctly, which still leaves one in ten reading wrongly. None of the eight tools flagged our two-column test file.