Insights

What the data actually shows

Resume advice is mostly assertion. We would rather publish measurements — including when the measurement is inconvenient, and including when we don't have enough of it yet to say anything.

Does our score predict a callback?

Every ATS tool gives you a number. As far as we can tell, none of them publish whether that number relates to what happens next. We ask users what came of their application and compare it against the score we gave them.

Outcome data is not available right now. Check back shortly.

How these numbers are produced

After a scan we store the score, grade, role category, which sections were detected, and which keywords were missing. No resume text, no contact details, no identifier that links a scan to a person.

If a user later tells us what happened to that application, we store the outcome against the scan id and nothing else. There is deliberately no way to work back from an outcome to an individual.

Nothing is published below 100 data points, and no individual band is reported below 30. Where a figure is missing you will see a dash and the reason, rather than an estimate.

Full detail of the scoring itself is on our methodology page.

What this can't tell you

Outcomes are self-reported and voluntary. That is a biased sample, and no amount of volume fixes the bias — it only narrows the error bars around a skewed estimate.
Correlation between score and callback would not prove the score caused it. People who scan their resume are already people who tailor their applications.
We see one scan, not the rest of an application: the cover letter, the referral, the timing, whether the role was already filled.
If the data ends up showing our score does not predict callbacks, we will publish that. Being the only people holding the evidence is the point, not being flattered by it.

Contribute a data point

Scan a resume, apply, then come back and tell us what happened. That is the whole dataset.

Scan my resume

How these numbers are produced

After a scan we store the score, grade, role category, which sections were detected, and which keywords were missing. No resume text, no contact details, no identifier that links a scan to a person.

If a user later tells us what happened to that application, we store the outcome against the scan id and nothing else. There is deliberately no way to work back from an outcome to an individual.

Nothing is published below 100 data points, and no individual band is reported below 30. Where a figure is missing you will see a dash and the reason, rather than an estimate.

Full detail of the scoring itself is on our methodology page.

What this can't tell you

Outcomes are self-reported and voluntary. That is a biased sample, and no amount of volume fixes the bias — it only narrows the error bars around a skewed estimate.

Correlation between score and callback would not prove the score caused it. People who scan their resume are already people who tailor their applications.

We see one scan, not the rest of an application: the cover letter, the referral, the timing, whether the role was already filled.

If the data ends up showing our score does not predict callbacks, we will publish that. Being the only people holding the evidence is the point, not being flattered by it.