One structural change, measured twice: 63, then 71
We took a weak document, made exactly one change, and ran the free check again. Both sums are photographed.
— same words, different section, eight points.
Re-measured on 26 September 2026. This page was first measured on 13 September, at 40 before and 80 after. The checker then did not read lines under a Projects heading, so the weak version returned content 0; that was a fault in the checker and it has been fixed, and on 25 September the rules moved to grading in proportion. The numbers and pictures below are from 26 September, and the one change is now worth eight points rather than forty.
Advice about résumés is usually unfalsifiable, so here is a small falsifiable experiment. One document, one change, two measurements, both photographed.
The document before
An invented backend engineer's work, written up under a PROJECTS heading: two project names with three achievement lines between them, plus a skills list and an email address. The ATS checker read 48 words and scored 100 × 30% + 48 × 30% + 46 × 40% = 63.
Why it scored that way
The projects read as experience evidence but not as employment, and neither carried a date of its own, so completeness lost both of those and more: no education, no phone number, no summary. Content graded the three lines as they are, two of them with a figure, three lines where the full depth score asks for twelve.
The one change
The PROJECTS heading became EXPERIENCE, with a role title, an employer and a date range under it, and the same three achievement lines were gathered beneath that. Not one line was rewritten. Nothing was added except the role line itself.
The document after
100 × 30% + 75 × 30% + 46 × 40% = 71. Completeness went from 48 to 75, and content stayed at 46.
Where the eight points came from
All of it from completeness. The lines now sit under a role with an employer and dates, so they count as employment, worth all 30 experience points where projects earn 18, and the entry carries its own date, which earns the dates points. Content did not move, because today's rules grade the same three lines the same way wherever they are filed. On 13 September the same change was worth forty points, because the rules of that day could not see project lines at all; that is the pattern described in evidence ceilings explained.
What is still wrong at 71
There is no education section, no phone number and no summary, and the checks under the score name each. On content, three achievement lines and 52 words are graded as the thin page they are. The fully written version of the same career measures 88, as shown in the same work, written twice, and the habit of comparing runs is described in what a fix was worth.
Why one change at a time is worth the extra minute
Because a pair of measurements tells you something about the next document as well as this one. Changing five things and re-running gives you a better résumé and no knowledge: you cannot tell which of the five mattered, so the next time you are in the same position you will do all five again and still not know. Changing one thing costs a second run of the check, which takes a minute, and buys a fact you keep.
What this page does not claim
That every document has points like these in it, or that a score changes an outcome. The candidate is invented and nobody was hired. What the pair shows is a mechanism worth trusting: the scorer measures what it can see, real work filed in the wrong section is read as less than it is, and moving it is a ten-minute edit rather than a career change.
The habit this supports is the useful part. Run the check, take the single check marked Review that is worth the most, do only that, and run it again. Previous checks keeps both runs on this device with their scores, so you can see what the change bought rather than guessing. Doing five things at once and re-running produces one number and no information, which is the difference between measuring and hoping.
One last thing this pair demonstrates about advice in general. A claim of the form fix this and your score improves is testable in about three minutes here, because the rules are published and the reports are kept. That is a rare property in this category, and it is worth using on anything you read, including the pages on this site. If a piece of advice cannot be shown to move a named row, it is a preference rather than a finding, and it should be read that way.
Run it, change one thing, run it again
Previous checks keeps earlier runs with their scores, so the pair can be compared and the difference is what the change was worth.
Open the ATS checkerWhat a fix was worthQuestions
- What exactly was changed
- A PROJECTS heading became an EXPERIENCE heading with a role title, an employer and a date range, and the three achievement lines were gathered under it. No line was rewritten and nothing was added except the role line.
- What were the two scores
- Measured on 26 September: 100 × 30% + 48 × 30% + 46 × 40% = 63 before, and 100 × 30% + 75 × 30% + 46 × 40% = 71 after. On 13 September, under the rules of that day, the same pair measured 40 and 80.
- Why did only completeness move
- Because today's checker already reads achievement lines under a Projects heading, so the content dimension graded the same three lines the same way both times. What changed was their standing: projects earn 18 of the 30 experience points and employment earns all 30, and a dated role earns the dates points that undated projects did not.
- Is the document finished at 71
- No. There is still no education section, no phone number and no summary, and on content three achievement lines and 52 words are graded as the thin page they are. The checks marked Review say what each would earn.
- Does my document have points like these in it
- That depends where it starts. The mechanism is general: find what the scorer could not see, and put it where it can.
Conxfolio is a free set of four career tools: a résumé builder with 37 rendered layouts, an ATS résumé checker that prints its own arithmetic, a cover letter builder that traces every proof paragraph back to the line it came from, and a portfolio builder with 22 authored designs. There is no account to create, nothing is held back for a paid plan, and no language model is used anywhere in the product, so the readers, the score and the letter are deterministic code you can check.