We measured how many resume bullet points actually contain a number (n=542)

Every resume guide says: “Quantify your bullet points.”

Nobody says how many people actually do it.

We build ResumeAtlas, which compares a resume against a specific job description, so we had the raw data to check.

The result:

• Median: 36 bullet points per resume
• Median quantified share: 5.4%
• 25.3% had zero quantified bullets
• 65% quantified fewer than 10% of their bullets

How we measured it

ResumeAtlas stores an evidence_dashboard.snapshot JSON blob for each analysis.

The relevant fields are:

• bullets[] — extracted work-experience bullets
• hasMetric — whether a bullet contains a measurable number

We analyzed 542 snapshots from June 8 to September 7, 2026 every run in that window that contained a snapshot.

There was no sampling.

For each run:

quantifiedShare = bulletsWithMetric / totalBullets

What counts as a number?

Our hasMetric check uses two stages:

1. Regex detects percentages, currency, counts, time changes, multipliers, rates, and scale units.
2. A model removes false positives such as Python 3.11, 2024, ZIP codes, and phone numbers.

The check is deliberately conservative, so 5.4% should be treated as a floor, not a ceiling.

The distribution

Quantified bullets	Resumes
0–10%	353 (65%)
10–20%	77 (14%)
20–30%	42 (8%)
30–40%	60 (11%)
40%+	10 (2%)

The interesting part isn't just the 5.4% median.

65% of resumes are concentrated in the 0–10% range.

Most candidates aren't slightly under-quantifying their experience. They're barely quantifying it at all.

Why?

“Quantify your accomplishments” is often interpreted as:

“I need a percentage improvement for every bullet.”

That's too narrow.

Numbers can describe:

• people managed
• clients supported
• tickets handled
• contracts reviewed
• budget managed
• frequency
• project scale
• before/after changes

For example:

“Reviewed ~400 contracts annually.”

or

“Reduced month-end close from 5 days to 2.”

No dashboard required.

Caveats

This isn't a representative sample of all job seekers. These are ResumeAtlas analysis runs and currently skew toward software, data, and engineering roles.

Our PDF extraction spot-check showed approximately 94% bullet recall.

And having a number doesn't make a bullet good. “Improved synergy by 30%” can still be meaningless if the candidate can't explain the number.

We measured metric presence, not bullet quality.

The takeaway

The advice is correct.

The adoption is the surprising part.

You don't need a number on every bullet. You need credible numbers where they actually exist.

And apparently, most resumes have very few.

Full breakdown: https://resumeatlas.io/quantify-resume-accomplishments

Questions about the methodology? Happy to answer them in the comments.