Outcomes & methodology
What we measure,
and what we won't claim.
Most resume tools publish impressive numbers without saying where they came from. This page documents how our scoring actually works, how we approach ATS compatibility, the rules that keep generated content truthful, and the specific claims we refuse to make until we can evidence them.
Last updated: August 2026
01 — Scoring
How your match score is built
The overall figure is a composite of three components. Two are computed; one is model judgment. We distinguish between them because the difference matters.
Hard skills coverage
Counted, not estimatedWe extract the skills and tools named in the job description, then check which appear in your resume. The matched and missing lists shown in your results are the actual sets — you can see every keyword we found and every one we didn't.
ATS compliance
Rule-based checksA fixed set of structural checks runs against the generated document: single-column layout, standard section headings, no tables or graphics, no content stranded in headers or footers, consistent and parseable date formats.
Experience alignment
Model judgmentAn AI assessment of how closely your actual responsibilities and scope match what the role asks for. This is a judgment, not a measurement, and we label it as such in your results.
What the score is not: a prediction of whether you will be interviewed or hired. It measures how well a document matches a job description. That is a useful signal, and it is the only thing it is.
02 — ATS
How we approach ATS compatibility
Applicant tracking systems parse resumes into structured fields, and they differ considerably in how they do it. Rather than claim to replicate any one system, we build to the practices they share and that vendors document publicly: a single-column layout, conventional section headings, machine-readable dates, and text that lives in the document body rather than in headers, footers, tables, or images.
This is deliberately a narrower promise than “we simulate Workday.” Employers configure their own screening rules, weightings, and knockout questions. No external tool can see that configuration, so any product claiming to reproduce it is overstating what it knows.
What our checks cover
- Single-column, parseable layout
- Standard section headings
- No tables, graphics, or text boxes
- Nothing critical in headers or footers
- Consistent, machine-readable dates
- Keyword coverage against the JD
03 — Integrity
How we keep generated content truthful
AI writing tools invent things. The design question is what constraints you place on them, and what you leave to the user to verify.
- —Never fabricate experience, companies, degrees, or certifications.
- —Never invent metrics. If a number isn't in your source material, it doesn't appear in the output.
- —If a degree field is empty in your resume, it stays empty — no plausible-sounding credential is filled in.
- —Rewrites reframe and sharpen what you actually did; they do not add new responsibilities.
- —Every generated document is editable, and every claim is yours to review before you send it.
Two features exist specifically to support this. Advise Only mode returns recommendations without rewriting your document, so you make every edit yourself. The Career Evidence Vault lets you store the raw material — achievements, review excerpts, quantified outcomes — that generated statements are built from, so a claim on your resume can be traced back to something you recorded.
We do not tell you the AI cannot make mistakes. We tell you the system is built to avoid them and that you remain the final check. Review everything before it reaches an employer.
04 — Claims
What we claim, and what we don't
We do claim
Your resume is restructured against documented ATS parsing practices.
We implement the formatting and structural rules that major ATS vendors publish or that are consistently observable across them.
Keyword coverage against a specific job description is measured.
Matched and missing keywords are computed from the JD you paste, and shown to you in full.
The system is designed not to fabricate experience.
Explicit integrity constraints in every AI call, plus an Advise Only mode that suggests changes without rewriting.
We do not claim
We do not promise interviews, offers, or a specific score increase.
Hiring outcomes depend on the market, the role, the competition, and the employer — variables no resume tool controls.
We do not claim to simulate a specific employer's ATS.
Every employer configures their ATS differently. We optimize for documented practices that platforms share; nobody outside the company can reproduce a given employer's exact configuration.
We do not publish aggregate placement or salary-increase statistics.
We do not yet have a large enough representative sample of verified outcomes. Publishing a number before then would be marketing, not evidence.
We do not publish peer benchmark percentiles.
Your dashboard compares your funnel to itself over time. Peer comparison appears only once our outcome data is representative enough to support it.
05 — Evidence
What we can evidence today
Named user testimonials
The quotes on our homepage are from real users who submitted them and agreed to their use. They are individual experiences, not typical results, and we present them that way rather than converting them into an implied average.
Your own tracked funnel
When you record what happened to each application — applied, recruiter screen, interview, final round, offer, accepted, or rejected — your dashboard reports your real conversion rates and tells you where your search is actually breaking down. That diagnosis uses only your data.
What is not here yet
We publish no aggregate interview rate, placement rate, or average salary increase, because we do not yet have a representative verified sample. When we do, we will publish the figure alongside its sample size, date range, and how the outcome was confirmed — so the number can be judged rather than just believed.
06 — Institutions
Reporting for employers and universities
Institutional buyers are measured on outcomes, so pilots are structured to produce evidence rather than impressions. A pilot reports activation and usage from platform data, and participant progress from self-reported application status — labelled as self-reported, because that is what it is.
Judge it on your own resume.
The free tier scores your resume, shows the matched and missing keywords, and runs the ATS structural checks. No payment required to see whether the diagnosis is worth anything.
Check my resume free →