Resume · Artificial intelligence
The Metric Mirage: Why Overusing Numbers Is Killing Your Resume
For years the advice was to quantify everything. Then ChatGPT learned the rule, and now every resume reads like a sales deck. Recruiters can spot the fakes — and the fakes are sinking real applications too.
In thirty seconds
- "Quantify your achievements" was great advice — until ChatGPT learned the rule, and now every AI-generated resume is stuffed with suspiciously round percentages.
- Recruiters now read "improved efficiency by 37%" as a tell, not a strength. Round numbers, vague metrics, and identical formats across every bullet are the new red flags.
- Real metrics still work — but only when they're specific, contextual, and defensible in interview. Three tests: is it real, is it specific, can you talk about it for two minutes if asked.
- Roles that don't produce numbers (most of them, honestly) shouldn't be forced to. There are stronger ways to show impact than fabricating a percentage.
- The goal isn't to impress an algorithm. It's to read like a person describing real work to another person.
For years, the standard advice on resume writing went: quantify everything. Vague duties? Replace them with numbers. "Managed a team" became "managed a team of 8". "Improved processes" became "improved processes by 23%". Hiring managers responded well, the advice spread, and for a long time it was simply true that resumes with metrics outperformed resumes without them.
Then ChatGPT learned the rule.
Now every AI-generated resume — and there are a lot of them — is stuffed with confident, suspiciously round percentages. "Increased efficiency by 37%." "Boosted sales by 80%." "Cut costs by $200K annually." On the surface, that should still be a good thing. In practice, it's actively undermining the documents it appears in.
Recruiters can spot the fakes
The pattern is now obvious to anyone who reads resumes for a living. AI-generated metrics have a tell: they're rounded, they're convenient, they appear on every bullet, and they often don't quite match the role they're attached to. A junior administration officer doesn't usually save $200K a year. A graduate marketing coordinator rarely "increases conversion by 47%." When the numbers don't fit the seniority, the role, or the industry, recruiters notice.
What's worse: once a recruiter spots one fabricated metric, they start questioning the whole document. The credibility damage isn't local to the dodgy bullet. It infects every other claim on the page.
The metric used to be the thing that made the bullet credible. Now it's often the thing that makes the recruiter doubt the rest.
This is the metric mirage. Numbers were the gold standard for so long that AI tools were trained to favour them — and now that everyone's using AI, the same numbers that used to differentiate a candidate make them blend into the AI-generated pile.
What a fake metric looks like
The pattern is recognisable once you know what to look for. Three signatures of AI-typical metric inflation:
"Increased efficiency by 37%." Round, vague, no method, no baseline, no time frame. Efficiency of what? Measured how? Compared to what?
"Reduced quarterly compliance reporting time from 5 days to 1.5 by automating data extraction in Power BI, freeing 14 staff hours per quarter for higher-value analysis."
"Boosted team performance by 25%." Performance measured how? Over what window? On whose say-so? "Boosted" alone is the giveaway — it's a verb that only appears on resumes.
"Restructured weekly stand-ups and introduced a shared sprint board, lifting on-time ticket close rate from 68% to 82% over two quarters."
"Saved $200K annually through process optimisation." A junior or mid-level role rarely owns $200K of P&L. The size doesn't fit the seniority, which is the fastest way to flag a metric as invented.
"Identified and remediated three duplicate vendor agreements during AP audit, saving the team approximately $18K per year and improving invoice match rate."
The good versions don't sound smaller — they sound real. They have specific numbers, specific contexts, specific methods, and they fit the role. A reader can imagine the work happening behind them, which is the point of a metric in the first place.
The three tests for any metric on your resume
Before you commit a number to your resume, run it through these three checks. If it doesn't pass all three, take it out.
Test 01
Is it real?
Did this number actually happen? Can you point to where it came from — a report, a system, a manager's email, a personal log of your work? Or did it appear out of nowhere because the bullet "needed a number"?
The bar isn't certified accuracy. It's that you genuinely believe the figure and could explain how you arrived at it. "We didn't have formal metrics, but I tracked my own throughput in a spreadsheet and it averaged X" is fine. "I have no idea but 30% sounds about right" is not.
Test 02
Is it specific?
"Improved efficiency by 30%" fails this test. "Efficiency" of what? Compared to when? Measured how? Vague metrics signal that the number was bolted on after the fact, not derived from real work.
Specific metrics name the thing being measured, the baseline, and the time window. "Reduced average customer call resolution time from 8 minutes to 5 minutes over six months" passes. "Improved customer service by 38%" doesn't.
Test 03
Can you talk about it for two minutes in interview?
This is the killer test, and it's the one most candidates fail. If a recruiter asks "tell me about that 37% efficiency improvement" — can you talk about it for two minutes? What you measured, what you changed, what the result was, what you'd do differently? If you can't, the number doesn't belong on the resume.
The interview is where fabricated metrics collapse. The candidate can't expand on the figure, can't explain the method, can't describe the team or the timeline — and the recruiter quietly downgrades the application. Better to never claim a number than to claim one you can't defend.
What to use when the role didn't produce numbers
Here's the thing nobody quite says out loud: most roles don't produce clean numbers. Government policy work, clinical care, teaching, customer service, legal advisory, internal HR, project coordination, research — vast swathes of professional work create real value that doesn't compress neatly into a percentage. That's not a failure of the work. It's a feature of the work.
For roles like these, forcing fake metrics is worse than telling the story properly. The strong alternatives:
- Scope and complexity. "Caseload of 80 active matters across three jurisdictions, including five Federal Court appearances." Numbers, but not invented ones — they're scope markers.
- Stakeholder count and seniority. "Briefed Deputy Secretary fortnightly on national policy developments affecting 14 portfolio agencies." Tells the reader the level of work without claiming an outcome metric.
- Concrete examples. "Drafted Cabinet submission on aged care funding reform; submission progressed without amendment." Specific, defensible, and shows the quality of work without inventing a percentage.
- Recognition signals. "Selected from 22 graduates for the rotational leadership program." A fact, verifiable, gives the reader information about how you were perceived in the role.
- Frequency or consistency. "Maintained 99% medication chart accuracy across 36 months on rotating ward shifts." When the actual job is consistent execution under pressure, that consistency is the metric.
None of these require fabrication. All of them communicate value to a recruiter, and crucially, they survive the interview test — you can talk about them because they happened.
A practical rule
Metrics are seasoning, not the dish. A good resume has 4–6 strong, defensible numbers across the document — not a percentage in every single bullet.
If your resume currently has a number in every bullet, it's almost certainly working against you. Pick the four to six numbers you can absolutely defend, place them where they have the most impact, and write the rest of the bullets with strong action verbs and specific scope language. The recruiter will trust the numbers more because there are fewer of them, and the document as a whole will read like a person wrote it.
That last bit is the actual game now. The resumes that win in 2026 don't out-metric the AI spam. They read, unmistakably, like a real person describing real work — numbers where the numbers belong, scope where the numbers don't, and credibility throughout.
Real numbers, written by a real person
We pull the metrics out of you. We don't invent them
An hour-long interview where a senior writer asks the questions that surface the figures you've actually got — caseloads, throughput, project values, headcount, savings, time-to-resolution — and writes them into the document where they belong. No AI inflation. No invented percentages. No "improved by 37%". Real, specific, defensible. 4.8 on Google. Trading since 2016.
