# The Demand-Check Resume Standard, and What Fails a Pass

*You can grade a finished resume pass or fail against live hiring demand for your target role, using posting-frequency thresholds two people would apply the same way.*

- Canonical URL: https://www.refolk.ai/candidates/guides/demand-check-resume-standard
- Pillar: Reading the market
- Format: Standard
- Published: 2026-09-22
- Last reviewed: 2026-09-22
- Reading time: 16 min

You have a finished resume. Before you send it, you want to know one thing: does it aim at roles the market is actively hiring for, or at a shrinking niche you happen to be good at? This guide gives you a pass/fail standard for exactly that check, keyed to how often a skill appears in real postings rather than any vendor's opinion of what is hot. Run it once, and two people grading the same resume against the same postings will reach the same verdict.

Most guides in the resume-and-market space help you build or rewrite a resume toward demand. This one assumes the writing is done. It is the gradeable checklist you run after the fact: countable thresholds, an evidence bar, and a decline check, all applied inside a fixed search area.

## What "demand-aligned" means, stated so two graders agree

A resume is demand-aligned when every must-have skill for the target title is present with evidence, and no listed skill sits below the dead-weight line, inside one defined search area. That sentence is the whole standard. Everything else in this guide is how to make each clause countable.

The trap is grading on feel. "This looks like a strong data resume" is not a standard, because the next reader might disagree. The fix is to convert every judgement into a number pulled from real postings. The core measure is **skill rate**: the number of postings for your target role that mention a given skill, divided by the total postings you sampled. It is the same formula the Burning Glass Institute uses in its skills-first methodology, where skill rate is the number of postings mentioning a skill divided by total postings for that role.

Skill rate makes the must-have/differentiator split a math operation, not an opinion. The Burning Glass Institute pairs skill rate with **TF-IDF** - term frequency of a skill within a role's postings multiplied by inverse document frequency, which downweights skills that show up across many roles - and weights them 30% skill rate, 70% TF-IDF to curate 60 to 65 skills per role from an initial list of about 70. You do not need their full pipeline. You need the property it gives you: two graders using skill rate over the same sample of postings converge on the same list. That two-person-agreement property is what turns a resume review into a standard.

> Skill rate turns a resume review from an argument into an arithmetic you can both check.

## The three signals a demand check reads, and what each proves

A demand check reads three signals: how often a skill appears in live postings, whether the target title still posts at all, and whether the occupation is growing or shrinking. Each proves something specific, and each lies in a specific way if you read it alone.

| Signal | What it proves | How it lies |
|---|---|---|
| Skill rate over 20 postings | The market repeatedly asks for this skill | A single posting inflates a niche skill into a "must-have" |
| Live posting count per title | The title is a category employers post | Strong past presence hides near-zero current demand |
| BLS projection direction | The occupation is growing or shrinking | Field-level decline masks a healthy task mix inside it |

Read together, these signals cross-check each other. A skill with high skill rate inside a title that still posts, in an occupation projected to grow, is a safe must-have. A skill with high skill rate inside a title that no longer posts is a warning, not a win.

One measure you must not use as a demand signal is raw professional headcount - the number of people who list a skill or hold a title. Headcount describes supply, not open roles.

> **Watch out:** Headcount is not demand
>
> The number of people who hold a skill counts supply, not open roles. Refolk's index lists 584,322 US professionals with Python against 30,068 with COBOL, but neither figure tells you what employers are posting for. Always pair headcount with live posting volume.

## Why headcount and demand diverge, and the gap is the point

Headcount and demand are different numbers, and the distance between them is what a demand check exists to catch. A skill can be crowded with supply and still be weak on your resume as a differentiator; a skill can be scarce and still command value against sticky legacy demand.

The clearest illustration comes from Refolk's index of professional profiles, which counts people, not postings.

| Skill | Country | Professionals | Multiple vs COBOL |
|---|---|---|---|
| Python | United States | 584,322 | 19.4x |
| COBOL | United States | 30,068 | 1.0x |

Python is held by 19.4 times as many US professionals as COBOL. Reading that as "Python is 19x the demand" is the error. It says Python is 19x the supply. On a resume, a skill that hundreds of thousands of people list is a weak differentiator by definition, even when demand for it is genuinely high - the crowd flattens its distinctiveness. COBOL, thin on supply against a legacy install base that will not migrate quickly, can be the scarcer, stronger line for the right role. The demand check separates "many people have this" from "the market posts for this," and only the second belongs in the grade.

**19.4x - US professionals listing Python vs COBOL in Refolk's index**

A supply gap, not a demand gap - the distinction the whole standard turns on.

## Why geography has to be fixed before any threshold

Fix the search area first, because the same title can be five times as common in one market as another, and any skill-rate threshold you compute is only valid inside the area you sampled. A resume "aligned" in one country can be aimed at a shrinking niche in another.

Refolk's index shows the size of the effect for one title across two markets.

| Title | Country | Professionals | Ratio vs Germany |
|---|---|---|---|
| Data Scientist | United States | 26,751 | 5.2x |
| Data Scientist | Germany | 5,118 | 1.0x |

The same title runs 5.2 times larger in the US than in Germany. Import US thresholds into a German search and you will grade against the wrong denominator. This is why step one of the procedure is a fixed search area, not a nice-to-have: the 20-posting sample, the skill rates, and the pass/fail line are all computed inside it. Change the geography and you must recompute everything.

#### What a demand check is computed inside

1. **Search area** - The metro, country, or remote scope you will actually apply within
2. **Target title** - A live posting category inside that area, not a dead one
3. **20-posting sample** - The evidence base for every skill rate you compute
4. **Skill rate** - The countable measure that grades each skill

*Every threshold is valid only within the layer above it, so fix the outer layers first.*

## The procedure: from finished resume to a graded verdict

Run these eight steps in order. Titles come first, then postings, then skills, then the grade. The order matters: define target titles before extracting skills, because a skill list built from a single attractive vacancy inflates niche skills into false must-haves.

#### Grade a finished resume against demand

1. **Define target titles and geography** - Write 3 to 5 target titles plus alternatives, and fix one search area (metro, country, or remote). Done = a written list and a defined area.
2. **Confirm each title is a live posting category** - Check the current posting count per title in the search area and discard any that return near zero. Done = a countable live-posting number per title.
3. **Pull a 20-posting sample per surviving title** - Collect 20 current, realistically applicable job descriptions as raw text. Ten can mislead you; twenty lets you count what actually repeats.
4. **Compute skill rate for each skill** - Count how many of the 20 postings name each skill, giving every skill an appears-in-X-of-20 figure. Done = a skill-rate count per skill.
5. **Classify each skill: must-have, differentiator, or dead weight** - Tag high-frequency skills as must-have, distinctive-but-not-universal as differentiator, near-zero as dead weight, applying TF-IDF logic so a universal skill is not mistaken for a differentiator.
6. **Grade the finished resume against the classification** - Mark each must-have present-with-evidence or absent, and flag every listed skill below the dead-weight line. Done = pass/fail per must-have plus flags.
7. **Attach evidence to each must-have** - Back each must-have with a work sample, quantified outcome, or verifiable reference, matching evidence type to skill type. Done = no bare claims on must-haves.
8. **Run the decline check on the target occupation** - Map the title to its SOC code and check the BLS projection direction. A projected double-digit decline triggers a retarget decision.

### Setting the pass/fail line for a skill

No public source publishes a percentage floor below which a listed skill should be cut, so you construct it from skill rate rather than cite it. Over a 20-posting sample, use these bands as a defensible default:

- **Must-have:** appears in a clear majority of the sample. These are the skills the resume must carry with evidence.
- **Differentiator:** appears in a distinctive minority and is distinctive to this role rather than every role. High TF-IDF, moderate skill rate. Keep these; they are where you win.
- **Dead weight:** appears in one posting or none. A skill here is occupying prime resume space it has not earned. Flag it.

The average posting requires 12 to 18 specific competencies, and only about 15% of candidate-role matches exceed 75%, so you are not aiming to carry every skill in the sample. You are aiming to carry the must-haves with proof and drop the dead weight.

> **Rule:** A must-have must recur
>
> No skill qualifies as a must-have on the strength of a single posting. Require it to appear across the 20-posting sample. One posting is a vacancy, not a market.

## How the demand check goes wrong

The demand check fails in seven predictable ways. Each produces a false positive - a resume that looks aligned but is not - and each has a specific check that catches it. This is the part of the standard worth the most attention, because a grade that passes a bad resume is worse than no grade.

| Failure mode | False positive it produces | The check that catches it |
|---|---|---|
| Single-posting targeting | A "must-have" that appears once | Require recurrence across the 20-posting sample |
| Frequency without distinctiveness | Leading the resume with a universal skill | Apply TF-IDF logic: distinctive to this role, or to every role? |
| Declared skill, no evidence | Passes a keyword scan, fails a skills screen | Attach a work sample or outcome to every must-have |
| Title that no longer posts | High self-match to a dying category | Confirm live posting count and cross-check BLS direction |
| Headcount read as demand | 584k Python professionals read as high demand | Pair headcount with live posting volume |
| Geography blindness | US thresholds imported into a German search | Recompute skill rate inside the actual search area |
| Assessment mismatch | Portfolio offered for a safety-critical skill | Match evidence type to skill type |

Two of these deserve extra weight because they pass silently.

**Frequency without distinctiveness.** A skill can appear in all 20 postings and still be the wrong thing to lead with. "Communication" appears in nearly every role in every field. High skill rate, near-zero distinctiveness. If you lead your resume with a skill that is universal rather than distinctive to the role, you have spent your best real estate on a line that separates you from nobody. TF-IDF exists precisely to downweight these. Ask of every high-frequency skill: is it distinctive to this role, or to every role?

**Declared skill, no evidence.** A resume can match every keyword and still fail, because the evidence bar has moved to the screening stage.

## The evidence bar: why a keyword match is not a pass

A must-have skill passes only when it is backed by evidence a skills-based screener would accept: a work sample, a quantified outcome, or a competency-specific reference. A bare declaration passes a keyword scan and fails the screen.

The market has shifted under resumes here. NACE reports skills-based hiring used in 87% of interviews and 65% of screens, while GPA screening fell from 73% of employers to 42%. Almost two-thirds of employers use skills-based methods to identify candidates, and LinkedIn reports companies running the most skills-based searches are 12% more likely to make a quality hire. The consequence for your resume: the reader is increasingly looking past the claim to the proof.

Named practitioners converge on what counts as proof. Valid evidence includes work samples, structured interview responses, and task-based assessments. For entry-level roles, portfolio work, community projects, and documented design challenges are treated as valid signals. Competency-focused reference checks now ask a referee to describe a project where the candidate used a specific skill to achieve an outcome, rather than offering a general endorsement.

The one rule underneath all of it is **evidence fit**: the proof must resemble the capability closely enough to be informative. A presentation skill should not be inferred only from a multiple-choice test; a safety-critical competence should not be inferred only from a polished portfolio. Match the evidence type to the skill type, or the evidence fails validation even when it exists.

**87% - NACE Job Outlook: interviews using skills-based hiring methods**

Screening runs 65%, so the evidence bar sits before the interview, not inside it.

**Evidence line for a must-have skill**

```
Skill: [must-have skill name]
Proof type: work sample | quantified outcome | competency reference
Evidence: [what you can show - shipped artifact, metric moved, referee who saw it]
Where it lives: [portfolio link, resume bullet, or reference name]
Fit check: does this proof resemble the skill closely enough to be informative? Y/N
```

*Attach one of these to every must-have. Pick the row whose proof resembles the skill.*

Where attaching this evidence and computing skill rates by hand gets tedious, tooling helps. [Refolk](/candidates) writes your resume from your own history, tailors it to each posting, and scores how well you fit, which surfaces the must-haves you are carrying without proof before a screener does.

## Running the decline check without abandoning your field

Map your target title to its Standard Occupational Classification code, check the BLS Employment Projections direction, and treat a projected double-digit decline as a retarget trigger. But retarget the task mix first, not the whole field, because decline concentrates in tasks.

BLS publishes projections for 832 detailed occupations. The decline signal is stark and specific.

| Occupation | Projected change 2024-34 | Absolute change |
|---|---|---|
| Word processors / typists | -36.1% | -14,400 |
| Data entry clerks | -25.9% | (not stated) |
| Cashiers | (not stated) | -313,600 |
| Office assistants | (not stated) | -177,800 |
| Bookkeepers | (not stated) | -94,300 |

Data entry clerks are projected down 25.9%, the largest percentage decline of any occupation, and typists down 36.1%. Meanwhile data scientists are projected up 34.6%, among the fastest growing. The pattern is that automation targets routine tasks, not domains. A resume that fails a demand check usually fails on task-level obsolescence: the data-entry portion of an administrative career, not administration itself. The fix is to retarget the task mix - move the resume's weight toward the durable, growing tasks in the same domain - before concluding the field is gone.

#### Retarget decision after the decline check

Horizontal axis runs from Occupation declining to Occupation growing. Vertical axis runs from Judgement-heavy tasks to Routine tasks.

| Quadrant | What it means |
| --- | --- |
| Declining, judgement-heavy | Reframe toward the durable tasks; the field is thinning but your work is not the target |
| Growing, judgement-heavy | Grade as-is; you are aimed where demand is |
| Declining, routine | Retarget urgently; both field and task mix are shrinking |
| Growing, routine | Move your task weight upward before the growth outpaces the routine work |

*Cross the occupation's projection direction with whether your task mix is routine or judgement-heavy.*

This is where Refolk's index earns its keep. Because it counts people by title, skill, and geography, you can see how thin or crowded your target is before you commit the resume to it, and check whether the profile you are aiming to become actually exists in your market.

Ask me this: `US professionals whose current title is data entry clerk moving into RPA or automation roles` - [run the search](https://www.refolk.ai/start?q=US%20professionals%20whose%20current%20title%20is%20data%20entry%20clerk%20moving%20into%20RPA%20or%20automation%20roles).

*Returns real people who escaped a declining occupation by retargeting the task mix, so you can see the durable path out of a shrinking title.*

## The final grade: a checklist you can adopt as policy

Run this checklist against the finished resume. Every item is a yes/no that two graders would answer the same way. If any must-have or evidence item fails, the resume fails the demand check and goes back for one edit, not a rewrite.

#### Demand check, pass/fail

- [ ] The search area is fixed to one metro, country, or remote scope, in writing
- [ ] Each target title returned a countable, non-trivial live posting count in that area
- [ ] A 20-posting sample was pulled per surviving title
- [ ] Every skill in the sample has an appears-in-X-of-20 skill-rate count
- [ ] Each skill is tagged must-have, differentiator, or dead weight using skill rate plus TF-IDF logic
- [ ] Every must-have is present on the resume, or the gap is deliberate and noted
- [ ] No must-have is carried on a single posting alone
- [ ] Every must-have carries a work sample, quantified outcome, or verifiable reference
- [ ] Every piece of evidence resembles the skill it proves (evidence fit)
- [ ] No listed skill falls below the dead-weight line without justification
- [ ] Headcount was never read as demand; live postings confirmed it
- [ ] The target SOC was checked against BLS direction; no unaddressed double-digit decline

### Keeping the grade current

A demand check has a shelf life, because postings, thresholds, and projections all move. Re-run steps two through six whenever you change search area, add a target title, or notice your reply rate drop. The 20-posting sample is cheap to refresh and is the fastest way to tell whether the market has moved or your resume has drifted. Re-run the decline check annually, when BLS updates its projections, or immediately if you hear that a chunk of your day-to-day tasks is being automated in your field. The standard does not expire; the numbers inside it do, and the whole design is so that refreshing the numbers is a 90-minute job rather than a rebuild.

## Frequently asked questions

### Is my resume aligned with job market demand?

It is aligned when every must-have skill for your target title is present with evidence and no listed skill falls below the dead-weight line. Define the target title first, confirm it posts in your search area, pull a 20-posting sample, and compute how many postings name each skill. A resume that carries the recurring skills with proof, and drops the ones that never appear, passes the demand check.

### How many postings should a skill appear in before I keep it on my resume?

No public source sets a universal floor, so build one from the skill-rate method. Over a 20-posting sample, a skill that appears in most postings is a must-have, one that appears in a distinctive minority is a differentiator, and one that appears in one or none is dead weight. Require recurrence: a skill named in a single posting is a false positive, not a requirement.

### How do I tell if my target role is declining?

Map your title to its Standard Occupational Classification code and check the BLS Employment Projections direction. Projected double-digit declines are the signal. For 2024-34, data entry clerks are down 25.9% and word processors and typists down 36.1%, while data scientists grow 34.6%. Decline usually hits the task mix rather than the whole field, so retarget the tasks before abandoning the domain.

### Does a high number of professionals with a skill mean high demand?

No. Headcount counts people who hold a skill, not open roles. Refolk's index lists 584,322 US professionals with Python against 30,068 with COBOL, but that gap describes supply, not what employers are posting for. A crowded skill can be a weak differentiator, and a scarce one can retain value against sticky legacy demand. Pair headcount with live posting volume before you conclude anything.

### Why does a keyword-matched resume still get rejected?

Because the evidence bar now sits at the screening stage. NACE shows skills-based methods in 87% of interviews and 65% of screens, while GPA screening fell from 73% to 42%. A resume can match every keyword and still fail when a screener asks for a work sample or a competency-specific reference. Attach a work sample, a quantified outcome, or a verifiable reference to every must-have.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/demand-check-resume-standard*