The Age-Signal Resume Line, Scored Per Element to Keep, Compress, or Drop

This guide is for anyone with 20 or more years of work history who is deciding, line by line, what stays on the resume. It replaces the blanket "cut to 15 years and delete your grad date" rule with five scored dimensions. Each graduation year, early role, legacy tool, tenure phrase, and contact detail gets a Keep, Compress, or Drop verdict that still holds up against the application form, the background check, the ATS, and your LinkedIn profile.

Why does a blanket age-proofing rule fail?

A blanket rule fails because the resume is only one of five places your history shows up. A cut that hides your age on the resume can open a gap, a mismatch, or a knockout everywhere else. The age penalty is real, but it attaches to specific signals, not to experience in general. So each signal has to be judged on its own.

The penalty itself is well measured. Neumark, Burn and Button sent fictitious applicants aged 29 to 31, 49 to 51, and 64 to 66 across more than 40,000 job applications. Across all four occupations, callback rates were about 18% lower for the middle-aged group and about 35% lower for the oldest group. The resume signalled age through the school graduation year, nothing else. The same work attributes around 90 percent of the discrimination found in earlier audit studies to the interview-selection stage. That is the stage your resume controls.

The penalty is not uniform, so one rule cannot fit every applicant:

The gap runs from 30% for men in retail to 47% for women in admin. If your group and sector sit at the high end, give the age-signal dimension more weight. The studies do not explain why the gap differs, so treat this as a calibration, not a theory. Readers applying in Europe should also note one finding: a meta-analysis by Lippens, Vermeiren and Baert found that hiring discrimination against older applicants is more prominent in Europe than in the United States.

The usual "how far back should a resume go after 50" answers are reasonable defaults, but they are only defaults:

• 10 to 12 years: AARP's Carly Roszkowski.
• 15 years: Amanda Augustine of TopResume, via AARP.
• 15 to 20 years: acceptable on executive resumes, per one practitioner guide.
• Cumulative: academic CVs list everything.
• Medicine and other credential-led fields: I found no source. Treat licensure length as a relevance question, covered below.

A window tells you where to start cutting. It does not tell you whether a specific cut creates a problem somewhere else. The rest of this guide is about that second question.

Where else does your history get checked?

Four checkpoints see your history besides the hiring manager: the application form, the ATS parse and knockout questions, recruiter search on LinkedIn, and the background check after an offer. Each one sees different elements at a different moment. A verdict only holds if it survives all four.

The application form and state bans

The form matters because it can ask directly for what your resume leaves out. Several states now bar that at the application stage:

A BFOQ (bona fide occupational qualification) is a narrow legal exception where a trait like age is genuinely required for the job. Details worth knowing:

• Colorado (the Job Application Fairness Act, C.R.S. 8-2-131) applies only "at the time of an initial employment application." Employers can still ask for transcripts if they first tell you that you may redact those details. The state labor department (CDLE) enforces it. There is no private right of action, which means you cannot sue under it yourself. A third violation can bring a fine of up to $2,500, and each application counts as a separate violation.
• Oregon (HB 3187, signed May 22, 2025) runs until the first interview is complete, or until a conditional offer if there is no interview. It explicitly covers information collected through applicant tracking systems. You may still choose to disclose the dates yourself.
• Connecticut bans these questions unless the job requires a minimum age.
• California, Minnesota, and Pennsylvania are reported to have similar laws. I have not verified their coverage. Wisconsin's law does not specifically ban age questions.

Laws change. To re-check, read the state labor department's guidance for the state where the job is located. Then open the employer's actual form, which is the only thing that tells you what this employer will ask.

The ATS years-of-experience count

An ATS (applicant tracking system) is the software that parses your resume into fields. Practitioner sources say most ATS engines calculate experience by adding up the length of each work entry. Some count overlapping periods once and some count them twice. An entry whose dates fail to parse drops out of the count. That means an undated "Earlier Experience" block most likely adds zero years. This is my inference from how parsing works, not something vendors document.

Minimum-years knockouts are usually form questions you answer yourself, and failing one removes your application entirely. Systems also judge how recent a skill is from the dates around it. A skill that appears only in a job from seven years ago reads as seven years stale. Workday and Greenhouse do not publicly document their exact methods, so test your own file rather than trusting any description, including this one.

LinkedIn Recruiter

Recruiter's search sidebar includes a Year of Graduation filter. Third-party guides describe Recruiter's years of experience as the span since your earliest Experience start date, not a sum of your roles. I found no official LinkedIn documentation confirming this. LinkedIn does let you show years without months. I have not confirmed against LinkedIn Help whether education dates can be left blank, so check that in your own profile editor.

The background check

Employment verification confirms dates and titles. Education verification confirms the degree and its conferral date through the registrar or the National Student Clearinghouse, which holds records for more than 3,600 colleges and universities. Screening companies flag any mismatch. The report does not separate lying from rounding. Employers do, and they care about the direction and size of the gap. A "graduated 2018" against a January 2019 conferral is paperwork if you resolve it quickly. It is not publicly established whether leaving a role off entirely creates a reportable discrepancy. Verification checks the entries you submit.

The five scoring dimensions

Score every element 0, 1, or 2 on five dimensions:

• Age signal (A) pushes toward removal.
• Relevance (R), parse and knockout effect (P), and cross-material consistency (C) push toward keeping.
• Verification exposure (V) decides whether Compress is a safe option at all.

V is the one dimension you score against the edit you plan to make, not the element itself.

A: Age signal strength

• 0: no age information. A current role, a current tool.
• 1: indirect. An early role with its dates, a legacy tool, a "20+ years" phrase in the summary, a landline, an AOL, Hotmail, or Yahoo address, a full street address.
• 2: direct. A graduation year, or a date range that starts decades back.

What it proves: how much this line, read alone, tells a screener your age. The graduation year is the only signal the callback experiments actually varied. The other tells come from AARP's resume guidance and practitioner lists. They are plausible, but no experiment measured them. When it lies: you score a tell as strong when the employer's form, or your LinkedIn profile, will reveal the same fact anyway. Removing it from one document then hides nothing.

R: Relevance to the target posting

• 0: the posting does not ask for it.
• 1: it supports a stated need, such as a tool the posting lists or a sector it prefers.
• 2: the posting requires it. A credential, a minimum years figure, or length of licensure.

What it proves: whether cutting the line costs you fit. When it lies: the tool is in the posting, but it only appears under a role from 15 years ago, so the system counts it as stale even though you scored it relevant.

V: Verification exposure

• 0: not a verified fact. Phrases, tools, contact details.
• 1: a verified fact (employment dates, titles, degree) that you would keep exactly as recorded, or drop whole.
• 2: a verified fact where the only available compression changes a date or title. Merging roles, rounding to years only, retitling.

What it proves: whether your edit will show up as a mismatch on a screening report. When it lies: you assume year-only dates are "close enough." The screen compares against payroll records and flags the difference regardless.

P: Parse and knockout effect

• 0: removing it changes neither your parsed years nor any skill's recency.
• 1: removing it makes a skill look less recent, or narrows your margin over a minimum-years threshold.
• 2: removing it drops your parsed years below a knockout, or below the figure you will claim on the form.

What it proves: whether the machine count still matches what you say you have. When it lies: you count years in your head, including undated entries the parser ignores.

C: Cross-material consistency

• 0: the element appears nowhere else.
• 1: it appears in materials you control, such as LinkedIn, a cover letter, or a portfolio, and you will have to change those too.
• 2: a form you cannot change asks for it directly. For example, a graduation date question in a state with no ban.

What it proves: whether the resume will disagree with something else the employer sees. When it lies: you score LinkedIn as consistent because you updated one entry, while the earliest start date still sits in the 1990s.

ELEMENT:
Appears in: resume / LinkedIn / cover letter / portfolio / form
Target posting:
Job location (state or country):

A  Age signal       0 none | 1 indirect | 2 direct (grad year, decades-old range)        Score:
R  Relevance        0 not asked | 1 supports a need | 2 required (credential, min years) Score:
V  Verification     0 not verified | 1 verified, kept exact or dropped | 2 edit alters it Score:
P  Parse/knockout   0 no effect | 1 recency or margin hit | 2 falls below a knockout     Score:
C  Consistency      0 nowhere else | 1 in materials I control | 2 a fixed form asks for it Score:

Rule fired (first match):
Verdict: KEEP / COMPRESS / DROP
Mirror on LinkedIn: yes / no, because:
Form answer this must match:

How do the scores become Keep, Compress, or Drop?

Apply five rules in order. The first one that matches gives the verdict. Requirements and fixed forms beat age signal. Age signal beats nothing else. Compress is only allowed when it leaves every verified fact intact.

The three verdicts:

• Keep: the element stays, with its true dates and title.
• Compress: the element stays but takes up less space. Fewer bullets, moved out of the summary, a tool shown in a current role. Verified facts do not change.
• Drop: the element comes off every material you control, at the same time.

The rules, first match wins:

1. R = 2 or P = 2: Keep. The posting requires it, or removing it breaks the years count. You may shorten the wording, never the dates.
2. C = 2: Keep. A form you cannot change will ask for it, so leaving it off the resume reads as hiding something.
3. A = 0: Keep. It carries no age information. Trim it only for space.
4. R = 1 or P = 1: Compress, if V is 0 or 1. If the only compression available alters a date or title (V = 2), Keep instead.
5. Everything else: Drop. Remove it from the resume, LinkedIn, and the cover letter together.

Worked examples, scored under the stated assumptions:

For the legacy tool, Compress means one of two things. If you use the tool today, show it under a current role. If you do not, leave it out of the skills list. Do not stretch the old role's dates to make the tool look current.

What does Refolk's index show about "25+ years" phrases?

In Refolk's index of professional profiles, "25+ years experience" shows up in US headlines at about half the volume of "10+ years experience". It clusters among founders and executives, who are selling their own experience rather than getting through an employer's screen. For an employed job seeker, it is a minority signal.

In a 25-profile sample of the US "25+" group, all 11 slots in the top-10 title list were founder, C-suite, board, or advisor titles. Read these numbers with their limits in mind:

• Matching is by keyword, so phrasing variants are missed.
• Each title sample is only 25 profiles.
• The US "10+" sample was skewed, with 8 of its 25 profiles at a single company.

What follows for scoring:

• Employed job seekers in corporate roles: a "25+ years" or "20+ years" phrase in the summary scores A = 1, R = 0, V = 0, P = 0. It does not add parsed years, and minimum years are answered on the form. The verdict is Drop, and the LinkedIn headline should match.
• Founders, fractional executives, and advisors: the people the phrase clusters around are buying or selling experience directly. Score R honestly. It may be 1.
• Credential-led fields: where length of licensure is the qualification, R = 2 and the verdict is Keep. Colorado's law explicitly lets law firms ask how long a candidate has been licensed to practise.

The scoring procedure, step by step

Run these eight steps once for your base resume, then repeat steps 2, 5, and 8 for each posting. No source gives timings. The estimates below are my own rough guesses: about 30 minutes for the inventory, 10 minutes per posting for thresholds, 1 to 2 hours for the first full scoring pass, and several days of waiting for records.

Notes on the steps that people get wrong:

• Step 1 is the slowest part when your history is spread across old files. Refolk writes your resume from your own history, so the full record of dated lines already exists in one place before you start cutting.
• Step 4 exists because screens verify records, not your memory. Score V against what the records say, not what you remember.
• Step 6 is where sources disagree. One practitioner guide says to list older roles by title and company only, with no dates. Another says never to leave out employment dates. The scores settle it: undated entries are safe only when P = 0, meaning you clear the posting's minimum years without them. Otherwise use real dates or drop the role.
• Step 8 compares two numbers. The parsed total should match the figure you type into the minimum-years question, not the figure you feel you have earned.

If you keep older roles, keep them short and keep their dates exact. Avoid year-only ranges. A range like "2021-2021" could mean one month or eleven, and year-only dates may not match payroll records.

EARLIER EXPERIENCE
Senior Analyst, Company Name, City | Mon YYYY to Mon YYYY
Analyst, Company Name, City | Mon YYYY to Mon YYYY
Associate, Company Name, City | Mon YYYY to Mon YYYY

How this goes wrong: failure modes and false positives

Most age-proofing failures come from a verdict that holds on the resume but breaks at another checkpoint. The seven below are the common ones. Each comes with the check that catches it before an employer does.

A verdict that holds on the resume but breaks on the form, the check, or LinkedIn is not a verdict.

1. Grad year dropped, but the form asks for it. The resume leaves the date off, then the employer's form in a state with no ban requires it. The omission now reads as concealment. This is why C = 2 overrides age signal. Check: open the employer's form before you finalize, and score C from the form, not from what you assume.
2. Undated Earlier Experience against a 15+ year knockout. You answer "20 years" on the form, but the ATS parses 12 because the undated block added nothing. Check: run the step 8 parse test and compare it to your form answer. If they disagree, P was really 2 and the verdict should have been Keep with dates.
3. Compress by merging roles or rounding to years. Two roles at one employer become one line, or month-year becomes year-only. The screen flags dates that differ from payroll records, and the report does not separate rounding from lying. Check: compare every kept line to your The Work Number file. Any edit that changes a recorded fact is V = 2, which forbids Compress.
4. LinkedIn left at the full timeline (false positive). The resume shows 15 years while Recruiter computes your span from a start date in the 1990s, and the Year of Graduation filter still applies. You believe you are hidden from age filters when you are not. Check: compare your earliest LinkedIn start date and education dates with the resume.
5. LinkedIn cut too far to match the resume. You trim LinkedIn hard, and Recruiter then counts you as a 3-year professional and filters you out of senior searches. Check: decide which seniority band you want to appear in and set your earliest LinkedIn start date to fit it. Consistency means telling the same story, not having identical line counts.
6. Legacy tool kept only in an old role. The tool is relevant, but because it sits only under a dated early role, the system reads it as stale. Check: does the tool also appear under a current role, truthfully? If not, Compress means moving it out of the skills list, not keeping it as a highlighted skill.
7. "25+ years" dropped in a credential-led field. Drop is the wrong verdict where length of licensure is the qualification. Colorado specifically allows law firms to ask how long a candidate has been licensed. Check: if the posting or the profession measures you by years of practice, R = 2 and rule 1 keeps it.

One more limit. The age-signal weights here rest on one strong experiment family plus practitioner lists of tells. The pooled estimates from the Batinovic et al. meta-analysis by age band (40 to 49, 50 to 59, 60 to 65, 66 and over) did not come back in my research. If you need band-specific weights, pull them from the paper rather than guessing.

Keeping the verdicts current

The verdicts are not permanent. R and P change with every posting, C changes with every form and every LinkedIn edit, and state law changes over time. Re-score each time one of those inputs changes, and re-run the parse test after every edit.

• Per posting: re-score R and P against that posting's minimum years and named tools. A role that was a Drop for one posting can be a Keep for another. Refolk tailors your resume to each posting and scores how well you actually fit, which saves you rebuilding the scoring sheet from scratch for every application.
• Per form: open each employer's form before you submit and re-score C on any graduation date or birth date question.
• Per jurisdiction: re-check the state labor department's guidance when you target a new state. Do not rely on a list, including the table above.
• Per records update: when you change jobs, pull a fresh The Work Number file so V is scored against current records.