The Search-Worthy Target Standard, Graded Before You Commit

You have settled on a target: a role title, a level, a place, maybe an industry. Before you point weeks of applications at it, you want to know whether there is enough of a real market to justify the aim. This guide is the go/no-go gate you apply on a Tuesday to a target you just wrote down. It grades that target against a fixed market-evidence bar and returns one of three verdicts - pass, adjust, or reject - built on a deduplicated live-opening count, a refresh rate, level depth, a pay floor, and a modal-requirement fit gate.

The point of a standard is that two people grade the same case the same way. "Set realistic goals" is not a standard. A deduplicated count of fresh, fit-passing openings at your rung, compared against your own offer math, is.

What "search-worthy" means, as a gradable bar

A target is search-worthy when enough deduplicated, fresh, fit-passing live openings exist at your specific rung and pay floor to sustain the application volume your offer math requires. Everything below is machinery for turning that sentence into a number and a verdict.

The bar has six gates. Each one can independently sink a target, and each has a documented failure mode where it lies to you.

Two honest limits before you start. First, no published source sets a specific numeric viable-opening count at a stated level and metro; that threshold is not publicly established, so I present the volume figures as a derived planning floor, clearly labelled. Second, the 7-day freshness share is a proposed standard, not a cited fact. Where I lean on evidence rather than convention, I say so.

Why the raw count lies, and how dedup fixes it

The first thing that fails a target is the count you started with, because it is almost never a count of real openings. Jobs appear 3 to 5 times across platforms, and Lightcast reports its two-step deduplication removes up to 80% of all jobs it collects. An ungraded number is not merely noisy; it can be five times too high, which is exactly the gap between a pass and a reject.

The mechanism has two sources. Aggregators re-list the same advertisement from multiple boards. And companies repost expired listings without removing the original, so a listing that expires, gets renewed, and leaves the old version alive shows up as the same job twice with different posting dates.

The fix is to match on the description, not the title plus company. Description-level matching catches staffing-agency reposts that a title-plus-company match misses. Collapse duplicates into a single listing while preserving all the source URLs, so you see one job with four links, not four jobs. Lightcast checks normalized fields - title, company, location - across 60 days of data to identify duplicates.

There are two windows in play, and they do different jobs. Use the 60-day window to find duplicates, because a repost from two months ago is still the same job. Use the shorter 30-day window to judge freshness, because a job that has sat live for a month is a likely ghost. Do not collapse the two. A listing can be a valid duplicate to merge and a stale listing to drop.

Grading level depth, where headcount hides rung thinness

A target can look deep on total headcount and be thin at your specific rung, and this is the failure that catches experienced seekers who assume "big market" means "big at my level." You grade depth by splitting the deduplicated count by rung before you judge anything.

Refolk's index gives a clean supply proxy for how a family thins by level. For the US Data Analyst family, the senior rung is a small slice of the whole.

Read that as candidate-supply depth by rung, a proxy for where the target thins out on the demand side too. In Refolk's index, senior is only 22.5% of the family. That is one problem. The second stacks on top: among senior-level postings, roughly 1 in 5 is a ghost. So the senior rung is thin and noisier at once, and a headcount-deep target can fail at your own level for two independent reasons.

A headcount-deep target can fail at your own level for two independent reasons stacked, thinness and ghost density.

The recalibration is direct. If your rung is thin, dropping one level often recovers depth because the non-senior slice is more than three times larger. If the thinness is real and the level is fixed, that is an adjust verdict pointing you at geography, covered next.

Geography is the biggest lever, so grade it early

Geography moves supply by roughly four times at the same title, which makes a metro or country constraint a larger lever on target viability than most level or skill tweaks. If your target fails on depth, relaxing where before relaxing what usually recovers more openings.

At the same "Data Analyst" title, the US pool is 4.03 times the UK's in Refolk's index. That is the scale of a country move. A metro constraint inside one country compresses the pool further. So when a target reads thin, test whether widening the geography clears the depth gate before you touch level or industry, because it is the cheapest large win available.

This is where a supply lookup saves hours. Instead of hand-counting career pages across metros, you can ask Refolk for the exact live, deduplicated slice you want and see the depth before you commit.

Freshness and ghosts, the discount you must apply

Even a deduplicated count overstates the market, because a large share of listings are stale or were never meant to be filled. Ghost and never-filled listings run somewhere between 18% and 27% of all online job listings, with Ashby's 18% never-filled figure the strongest lower bound for well-run teams. A common detection rule treats any job posted more than 30 days ago as a likely ghost, based on typical hiring timelines.

The structural version of this is the openings-to-hire gap. In one month employers posted 7.4 million openings but made only 5.2 million hires, leaving more than 2.2 million jobs that never materialized, and the openings-to-hire ratio has fallen from 1.8 to 1 at the peak to around 1.4 to 1. Roughly a third of the apparent market never converts. That is not seasonal; it is the discount you apply to any raw count.

The figures in that funnel illustrate the collapse ratio Lightcast documents; the point is the shape, not a promise about your specific target. Grade freshness as the share of your rung's real openings posted in the last 7 days. There is no cited pass/fail share, so treat any threshold as a proposed standard: I use "a healthy target refreshes a visible minority of its openings every week." A target with a fresh-share near zero is a backlog wearing the costume of an active market.

The fit gate and the pay floor

Two gates decide whether the surviving openings are actually yours: whether you qualify, and whether they pay enough. Both can be graded from public postings.

The fit gate at ~80%

The practitioner rule is that if you meet around 80% of the qualifications and can point to evidence you would do the job well, you should apply. The exception is a hard gating credential like a license, which is pass/fail, not weighted. Extract the modal must-have requirements across the target's postings and keep only the openings where you clear roughly 80% and hold any hard credential.

The years-of-experience bar is often softer than it looks. The share of US postings with no years'-experience requirement rose from 60% to 70% between April 2023 and April 2024. But bars vary sharply by sector, so grade against your own target's postings, not the average.

If your target sits in a high-bar sector and you are short on years, the fit gate will thin the count hard. That is real signal, not a reason to bluff a keyword match.

The pay floor

Posted ranges are now reliable enough to test against in much of the US. Eleven states plus DC - California, Colorado, Hawaii, Illinois, Maryland, Massachusetts, Minnesota, New Jersey, New York, Vermont, and Washington - require a good-faith pay range in the posting, and the posted range is usually close to what the job really pays. Pull the modal advertised range across the target's postings; if the typical band sits below your walk-away number, reject.

Walk-away number: ____
Modal advertised band across postings: ____ to ____
Verdict: PASS if band top clears walk-away comfortably / ADJUST if it straddles / REJECT if band sits below walk-away

Discard meaningless bands. An overly broad range such as $50,000 to $200,000 may not comply with good-faith mandates and tells you nothing, so exclude it and use the modal band across compliant postings, not one outlier. Where the target is in a non-disclosure state, this gate is weaker; triangulate from public benchmarks and treat the verdict as provisional.

The procedure, start to verdict

Run these eight steps in order. Each has a "done" condition, so you know when to move on, and the whole sequence takes an afternoon.

The final step needs the offer math explicitly. The average seeker needs 32 to 200 applications to land one offer, with most studies pointing to 100-plus, and Ashby found applications per hire tripled from 2021 to 2024 and stayed above 300 through 2025. But offer probability peaks in a band: applicants to 21 to 80 positions have a 30.89% offer rate versus 20.36% for 81-plus. The channel changes the math by an order of magnitude.

A target that fails on volume at 2 to 3% conversion can pass at 10 to 20%, so the fit gate doubles as a volume multiplier. This is where tailoring stops being advice and becomes arithmetic: Refolk tailors each application to the posting and scores the fit, which is what moves you from the portal-blast row to the tailored row of that table.

Derived planning floor, labelled as derived and not published: if you need roughly 50 to 100 real applications for a credible offer shot and about 1 in 5 live openings is a ghost, the target needs materially more than 100 deduplicated, fresh, fit-passing openings at the right rung to sustain that volume without forcing you past the productive 21 to 80 band.

How this goes wrong: seven false positives

Every gate has a way of reading pass when the truth is reject. These are the failure modes to check by name, and this section is the most valuable part of the standard because a target that lies to you costs weeks.

• Undeduplicated count inflates the market. A five-listing role reads as five openings. Collapse by description-level match; if the source URLs point to one employer and role, it is one opening.
• Stale backlog reads as active hiring. A large count that is mostly months old. Date every listing, drop past the 30-day cutoff, and recompute.
• Ghost jobs pad the fit-passing count. Postings alive with no intent to fill. Confirm the role sits on the company's own career page; senior rungs run about 1 in 5 ghost.
• Headcount depth masks rung thinness. A deep market where your specific level has a handful of real openings. Split by rung before grading anything.
• Fit-gate keyword bluff. A vague keyword match reads as qualified. Grade against must-haves at ~80% and treat licenses as pass/fail, not weighted.
• Pay-floor dodge. A compliant-looking but meaningless range like $50,000 to $200,000. Discard non-good-faith ranges and use the modal band, not an outlier.
• Volume mirage. Enough raw openings, but too few survive dedup, freshness, fit, and pay to sustain 50 to 100 real applications. Run the offer-math comparison on the final filtered count, never the raw one.

Notice how these compound. The senior-rung example fails on three of the seven at once: it is thin on depth, dense with ghosts, and small enough that the survivors miss the volume floor. A single failure mode is an adjust signal. Three stacked is a reject.

Verify before you commit

Run this checklist against your written target. If any item is unchecked, you do not yet have a verdict, you have a raw count wearing one.

Keeping the verdict current

A verdict is a snapshot, and the market underneath it moves, so re-grade rather than trust an old pass. Re-run the freshness gate first, because it decays fastest: the openings that made your target search-worthy this week may be a stale backlog next month. Re-run dedup whenever you refresh, because reposts accumulate. The structural discounts - the 18% to 27% ghost share, the 1.4 to 1 openings-to-hire ratio - are stable enough to keep as fixed assumptions, so you do not need to re-derive them each time.

If a re-grade flips a target from pass to adjust, apply the levers in cost order: widen geography first, since it moves supply about four times at the same title, then drop a level to recover depth, then loosen the industry, and only then reconsider the role. Retraining into a different role is the most expensive lever and belongs last, after the cheaper adjustments have failed. The standard does not tell you to aim lower. It tells you which single constraint to relax so the next verdict comes back pass.