The short version

  • Lead with the action, compress the situation into context, end on the result. One line, in that order.
  • This is our reordering of STAR for a medium that gives you a line rather than ninety seconds. We are not presenting it as an established framework, because it is not one.
  • The traceable relatives are real: STAR dates to DDI in 1974, and CCAR is the model the US Office of Personnel Management requires for senior executive applications.
  • Parsers do not read your numbers. Textkernel's output spec extracts title, employer, dates and location as structured fields; bullet prose stays prose.
  • The famous six-second figure comes from a 2012 job board study of about 30 recruiters, never peer reviewed. Useful metaphor, not a measured constant.

Most resume bullets describe a job rather than a person. "Responsible for monthly reporting" tells a reader what the role involved and nothing about whether you were good at it. Every hiring guide says to fix this and most of them hand you a framework built for a different medium.

STAR works in interviews, where you have a minute or two to walk through a situation, a task, an action and a result. A resume bullet gets roughly one line. Front-loading two clauses of setup before you reach the verb wastes the part of the line most likely to be read.

The reordering, and where it comes from

So we reorder it. Lead with the action, compress the situation and task into a short context clause, and finish on the result. Action, context, result.

One thing to be straight about, because plenty of articles are not: this is a reordering we find useful, not an established methodology with an institutional pedigree. If you search for the "ACR framework" you will find it asserted confidently and sourced to nobody. Treat it as a writing habit that suits the format.

The frameworks it borrows from are traceable. DDI states it introduced STAR in 1974 alongside its Targeted Selection product. CCAR, which stands for challenge, context, action and result, is the structure the US Office of Personnel Management requires in its guide to senior executive service qualifications. That is a government hiring authority mandating a context clause, which is about as solid a precedent as this subject offers.

Core Principle
“STAR was built for ninety seconds of speech. A resume bullet gets one line. Order the line accordingly.”

What the three parts do

  • Action. Start with the verb and the thing you did. It is the highest-value position in the line and the part a skimming reader is most likely to reach.
  • Context. One clause on the constraint that made it hard, or the scale it happened at. Without it a result is unreadable, because nobody knows whether it was difficult.
  • Result. What changed, stated as concretely as your evidence honestly allows.

What parsers actually extract from your bullets

Here is where a lot of advice on this topic invents a mechanism. You will read that numerals parse better than spelled-out numbers, or that percentage signs help an algorithm recognise achievements. We went looking for a vendor source for any of that and there is none.

What the documentation shows is the opposite. Textkernel's parser output specification extracts employment history into structured fields: title, organisation name, start and end dates, location, job category, and a description. Bullet prose is not converted into structured data. There is no field for a metric, a percentage or an achievement.

Greenhouse is consistent with this. Its parsing populates profile fields like name, contact details, current title and current employer, and its documented causes of a failed parse are all structural: columned layouts, complex tables, unclear section headings. Nothing about how you write a number.

So write numbers for the human. Use whatever is clearest to read, because the parser is not scoring them and the recruiter is the one you are persuading. That also means the file has to extract cleanly in the first place, since a bullet that never came out of the document cannot persuade anyone.

Recommended Action

See how your bullets read after extraction

Optimize your application materials in under 2 minutes.

Try it now

Running your resume through extraction shows the bullets as a recruiter's system stores them, stripped of the formatting that made them look organised on your screen.

How much evidence is behind any of this

Less than the confident tone of most career advice suggests, and it is worth knowing which parts are solid.

The closest thing to real research is a 2000 study by Bright and Hutton in the International Journal of Selection and Assessment, which had 62 managers and HR consultants rate genuine resumes manipulated to contain varying amounts of competency information. Resumes carrying competency statements scored significantly better on suitability and interview decisions, and where they sat in the document made no difference. Note what that tested: competency statements present against absent, with 62 raters, twenty-six years ago. It did not test the same bullet with and without a number, which is the experiment everybody cites and nobody appears to have run.

Two numbers you will see attached to this topic should be ignored outright. "Resumes with measurable achievements get 40% more callbacks" and "75% of hiring managers want achievements not duties" appear only on resume-tool marketing pages, with no named study, no sample size and no methodology behind either.

About the six seconds

The claim that recruiters spend six seconds on a resume comes from an eye-tracking study published by TheLadders in 2012, a job board, as marketing research. Trade coverage puts the sample at about 30 recruiters. It was never peer reviewed. A 2018 follow-up revised the figure to 7.4 seconds without publishing a sample size at all, and the original study PDF is no longer publicly available.

The 2012 release also noted that recruiters self-reported spending around four minutes per resume, against six seconds measured. That gap tells you what the study actually captured: a fast triage pass, not the whole evaluation.

Which is still a useful thing to design for. Something quick happens first, and a bullet that front-loads the verb survives it better than one that opens with two clauses of setup. Just do not carry the number around as though it were a physical constant, and be suspicious of anyone who does.

Where to start tonight

Take the three bullets you are proudest of, whether you wrote them yourself or started from an AI resume builder draft, and rewrite each one to start with a verb. For each, ask what made it hard and whether a reader could tell. Add one clause of context, and only then reach for the result.

If you cannot find a number, do not invent one. Count something honest instead, like frequency, volume, headcount or time saved, or describe the change qualitatively. A precise qualitative claim beats a fabricated metric, and it survives the interview where a made-up number does not. Then make sure the words survive parsing and that the skills the posting names appear where the evidence is.

Frequently asked questions

No, and we would rather say so. It is a reordering we find useful for a medium that gives you a line rather than ninety seconds. The frameworks it borrows from are traceable: DDI introduced STAR in 1974, and CCAR is the structure the US Office of Personnel Management requires in its guide to senior executive service qualifications. Anyone citing an "ACR framework" to an institution is citing nobody.

No. Textkernel's parser output specification extracts employment history into structured fields — title, organisation, dates, location, category and a description. Bullet prose stays prose. There is no field for a metric, a percentage or an achievement, so the number is working on the recruiter, not the algorithm.

Write whichever reads more clearly, because no vendor documents a preference. The claims that numerals parse better or that percentage signs help an algorithm recognise achievements have no source behind them. Greenhouse's documented causes of a failed parse are all structural — columns, tables, unclear section headings — and say nothing about how you format a number.

Do not invent one. Count something honest instead — frequency, volume, headcount, time saved — or describe the change qualitatively. A precise qualitative claim beats a fabricated metric, and it survives the interview where a made-up number does not.

It is a useful metaphor, not a measured constant. The figure comes from an eye-tracking study published by TheLadders in 2012 — a job board, as marketing research, with a sample of about 30 recruiters and no peer review. A 2018 follow-up revised it to 7.4 seconds without publishing a sample size at all. The same 2012 release noted recruiters self-reported around four minutes per resume, which tells you it captured a triage pass rather than the whole evaluation.

Some, and less than the confident tone of most advice suggests. A 2000 study by Bright and Hutton in the International Journal of Selection and Assessment had 62 managers and HR consultants rate genuine resumes manipulated to contain varying amounts of competency information; resumes carrying competency statements scored significantly better on suitability and interview decisions. Note what that tested — statements present against absent, twenty-six years ago. It did not test the same bullet with and without a number. The figures you will see quoted as "40% more callbacks" and "75% of hiring managers" appear only on resume-tool marketing pages, with no named study, sample size or methodology.

Use STAR there. It was built for ninety seconds of speech and it works in that setting, where you have room to walk through a situation and a task before reaching the action. The reordering exists because a resume bullet gets one line, and front-loading two clauses of setup wastes the part of the line most likely to be read.