# The Live-Demand Application Batch Standard, and What Fails the Reality Check

*You can grade a whole batch of target postings pass or fail against fixed thresholds, so two people scoring the same list drop the same dead entries before tailoring.*

- Canonical URL: https://www.refolk.ai/candidates/guides/live-demand-application-batch-standard
- Pillar: Reading the market
- Format: Standard
- Published: 2026-09-27
- Last reviewed: 2026-09-27
- Reading time: 16 min
- Keywords: is this job posting worth applying to, how to spot ghost jobs before applying, reposted job listing still worth applying, old job posting 30 days apply, check if job posting is live

## Key takeaways

- In Greenhouse platform data, 18 to 22% of jobs posted in any quarter are classified as ghost jobs, and a defensible batch contamination cap of about 20% mirrors that lowest ATS-grounded figure.
- The careers page beats the board because aggregator updates lag 24 to 72 hours when a role closes and the career page changes first, so board timestamps routinely lie about live status.
- Age thresholds must be role-typed: 30 to 45 days for typical fills but 50 to 62 days for engineering, or a blanket cut kills genuine live technical reqs.
- A 2% observed response rate against roughly 30% ghost jobs implies about 2.9% among real open roles, so a low callback rate mostly measures contamination, not your resume.
- A repost is only a meaningful ghost signal when content changed, title or salary shifted, or a new requisition number appears with no recruiter contact, since auto-refresh is routine.
- In Refolk's index, recruiter-to-engineer density is nearly identical across markets (US 0.40, Germany 0.44 recruiters per software engineer), so recruiter scarcity does not explain application silence.

Before you spend a week tailoring resumes and writing cover letters, grade the whole list of postings you plan to hit. This guide is the batch-level definition of done for a job seeker deciding where to aim their tailoring hours: fixed thresholds for freshness, repost contamination, employer activity, and competition realism, stated so two people scoring the same list reach the same verdict. Most market guides score one listing at a time. This one grades the committed set as a unit and tells you whether the week's effort is aimed at live demand before any of it is spent.

The reason to do this at the batch level is simple. Tailoring is your scarcest resource, and dead listings do not announce themselves. If a fifth of your list is ghost or filled, you will spend a fifth of your best hours on applications no human will ever read, and then read your low response rate as a verdict on your resume. It is not. It is a verdict on your list.

## What "live demand" means, and why grade the batch not the posting

Live demand means the posting maps to a real, currently open requisition where a human will review applicants within days. Grading the batch, rather than each posting in isolation, is what lets you set a maximum contamination rate and know whether the week is worth starting.

A single-posting red-flag check answers "should I apply to this one?" A batch standard answers a different, higher-leverage question: "of the twenty I committed to, is the surviving set clean enough that my tailoring hours are well spent?" The math is the point. In Greenhouse platform data, 18 to 22% of jobs posted in any given quarter are classified as ghost jobs, and 70% of Greenhouse clients posted at least one ghost job in Q2 2024. A LiveCareer survey of 918 HR professionals in March 2025 found 93% say their employer posts ghost jobs, with 45% admitting they do so regularly. Contamination is not an edge case. It is the ambient condition, and it is why the visible response rate is so grim.

**2.9% - Implied response rate among real, open roles**

A 2% observed rate against roughly 30% ghost jobs implies about 2.9% among live roles, because dead listings return 0% and drag the average down.

That single number reframes the whole job. The industry-wide response rate is about 2 to 3%: for every 100 applications, 97 to 98 receive an automated rejection or no response, and only 2 to 3 lead to human contact. But dead listings return zero by definition, so they pull the average down. Strip them out and the real-role rate is close to 3%. Grading the batch first raises your effective yield without changing a word of your resume. That is the entire thesis of this standard.

> A low response rate mostly measures the contamination in your list, not the quality of your candidacy.

## The pass thresholds: what a batch must clear

A batch passes when its surviving set clears four fixed thresholds: contamination under roughly 20%, freshness graded by role type, no unexplained repost or requisition churn, and no employer carrying a layoff or freeze against many open reqs. Two people applying these thresholds to the same list should drop the same rows.

Here is where the contamination cap comes from. The estimates cluster by method, and the honest move is to anchor on the lowest ATS-grounded figure rather than the alarming self-report ones.

| Method / source | Ghost-rate estimate |
|---|---|
| Greenhouse ATS platform data | 18-22% |
| Ashby ATS data | ~18% |
| JOLTS hiring-outcome gap | ~30% |
| Employer self-report (Resume Builder) | ~40% posted at least one; 30% live now |

A defensible batch cap of about 20% mirrors the lowest ATS-grounded number. That is deliberate. Employer self-report runs to 40% posting at least one fake listing and 30% currently advertising a role that is not real, but self-report and platform classification measure different things. Use 20% as the line the surviving set must clear, and treat anything above it as a signal that you graded too loosely or picked a contaminated sector.

Sector choice moves this baseline more than any tactic you can apply. Ghost rates cluster by industry: construction runs 38%, art 34%, and legal 29%. An identical batch discipline yields very different pass rates across fields. If you are aiming at construction postings, expect to cut harder and start with a longer list.

> **Rule:** The 20% contamination line
>
> A batch passes only when the surviving set's estimated ghost rate sits at or below roughly 20%, the lowest ATS-grounded figure. If more than one row in five looks dead after grading, do not start tailoring; widen or re-source the list first.

## Timestamp against the source, not the board

The single highest-value check is comparing the board's posting date to the company careers-page or ATS date, because aggregators sync on a documented delay. When a company closes a role in Greenhouse, Workday, or Lever, the update can take 24 to 72 hours to appear on the aggregator if it appears at all, and the company career page changes first.

This is why a board that says "posted 2 days ago" can be showing you a filled role. The board timestamp proves the aggregator re-indexed; it does not prove the requisition is open. The careers-page date is authoritative because it is where the change lands first.

#### Where a closed role stays visible

1. **Role closes in ATS** - The requisition is filled and marked closed in Greenhouse, Workday, or Lever
2. **Careers page updates** - The company site drops or hides the listing first
3. **Aggregator lags** - The board keeps showing the role for 24 to 72 hours, if it updates at all
4. **You see "fresh"** - The board timestamp reads recent while the role is already dead

*A role can be filled on the careers page while still showing as fresh on the aggregator for up to three days.*

When you timestamp, record the source-verified age in your tracker, not the board's claim. If a role appears on a third-party board but not on the company site at all, treat it as likely pulled after filling and verify before you spend anything on it.

Refolk exists partly to remove the manual half of this. When I tailor an application, I read the posting and score how well your history actually fits it, which is only worth doing on a posting you have already confirmed is live. [Refolk](/candidates) does the fit scoring so your verified hours go into the applications most likely to land.

## Grade freshness by role type, or you will kill live reqs

Freshness must be graded against the function, not a blanket age cut. Most legitimate openings fill within 30 to 45 days, but engineering roles average 50 to 62 days, so a single 30-day rule would wrongly kill genuine live technical reqs.

The dominant rule of thumb anchors on time-to-fill, and that number moves by sector and by year. SHRM's average time to fill fell from 48 days in 2023 to 41 days in 2024, with a 44-day benchmark reported for 2024. But averages hide the spread. Tech startups hire quickly, while regulated industries and government roles can run for months, and engineering roles specifically average 50 to 62 days per LinkedIn data. A listing active for two months or longer with no updates is a genuine warning sign for most roles, but a 55-day engineering req is often still live.

> **Watch out:** Do not fail technical and regulated reqs on age alone
>
> A blanket 30-day cut is the most common way this standard misfires. It drops live engineering roles that legitimately run 50 to 62 days and regulated roles that run for months. Tag age by role type before you cut, or you will delete your best real openings.

Set your thresholds like this and hold to them so two graders agree: general and commercial roles stale at 45 days, engineering and technical roles stale at 62 days, and regulated or government roles get flagged borderline rather than cut and pushed to the live-verification step instead. The point is not a single magic number. It is a fixed, role-typed line you can defend to a second reader.

## Read repost and employer signals for what they prove

A repost is only a ghost signal when content changed or a new requisition number appears with no recruiter contact; a plain refresh proves nothing. Employer signals are read the opposite way: a company that announced layoffs or a freeze while carrying dozens of open listings is a strong ghost signal on its own.

Auto-refresh on a fixed schedule is routine. LinkedIn roles open longer than 30 days can be auto-reposted under enterprise agreements, which means "reposted" is often just the board doing its job. Treat a repost as neutral by default. It becomes meaningful only when the title, seniority, or salary band changed, or when a fresh requisition number appears alongside recruiter silence. That combination suggests a role being re-shopped rather than filled.

The employer-activity check is where you catch the fresh-looking posting that is actually dead. Cross-reference the employer's headcount trend and recent news against its open-req count. A "fresh" posting at a company in a public hiring freeze is a false positive that age and repost checks will both miss. The freeze context is the tell.

Two things to keep straight while you grade, because mixing them corrupts your read:

- **Response rate** measures replies to candidate outreach: how often you hear back.
- **Callback rate** measures how many applicants a hiring team advances. These are different metrics, and blending recruiter-advance numbers with candidate-reply numbers will make you misjudge a channel.

And silence is not always the employer choosing to ignore you. In Refolk's index, recruiter-to-engineer density is nearly identical across markets, at 0.40 recruiters per software engineer in the US and 0.44 in Germany. Recruiter scarcity does not explain the black hole. Overload does: recruiter workload rose to 588 applications in Q3 2024, a 26% increase year over year, and the average posting receives over 200 applications. The people are there. They are buried.

| Pair | Market A | Market B | A:B ratio (derived) |
|---|---|---|---|
| Recruiters | US 131,068 | UK 9,200 | 14.2x |
| Software Engineers | US 329,692 | Germany 21,159 | 15.6x |
| Recruiters per SWE (derived) | US 0.40 | Germany 0.44 | - |

## The batch grading procedure

Run these eight steps in order over your committed list. The first six grade each row; the seventh applies the batch verdict; the eighth verifies only the survivors, so you never spend a live-verification check on a row you already cut.

#### Grade the batch before you tailor

1. **Assemble the committed batch** - List every posting you intend to tailor this cycle with URL, employer, source board, title, and first-seen date. Done means one tracker, one row per posting.
2. **Timestamp against the ATS or careers page** - Open the company careers page or ATS link and record the true post or refresh date, not the board's, because aggregators lag 24 to 72 hours and the career page changes first.
3. **Grade freshness by role type** - Flag anything over the age threshold for its function: 30 to 45 days for typical fills, 50 to 62 days for engineering. Tag each row fresh, borderline, or stale.
4. **Detect repost and evergreen contamination** - Compare the current listing to any prior version and check for requisition-number churn. Flag reposts only when content changed or a new req number appears with recruiter silence.
5. **Check employer hiring activity** - Cross-reference headcount trend and recent news against open-req count. Flag any employer with recent layoffs or a freeze plus many open reqs.
6. **Score competition realism** - Estimate applicant-pool depth against the role pool. Flag oversupplied-market roles for lower expected yield, to set expectations, not to auto-cut.
7. **Apply pass/fail thresholds to the batch** - Compute the batch contamination rate and drop failed rows. The surviving set must clear the roughly 20% max-contamination line.
8. **Verify survivors are live before tailoring** - Confirm each survivor on-source and, where possible, confirm engagement, since real roles show action within days. Record a live-verification note.

The competition-realism step deserves a word, because it is the one people misuse. Its output is not a cut; it is an expectation. Response rate varies enormously by channel, and knowing where a posting lives tells you how many applications one reply is worth.

| Channel | Response rate | Apps per response (derived) |
|---|---|---|
| Indeed | 20-25% | ~4-5 |
| LinkedIn | 3-13% | ~8-33 |
| Company website | 2-5% | ~20-50 |
| Tech roles (any) | 0.5-2% | ~50-200 |

A company-website application at 2 to 5% is not a failing application. It is a channel where one reply costs 20 to 50 tries. Use this table to set the size of your surviving batch, not to grade individual rows out.

Ask me this: `Software engineering roles at US companies that have grown headcount in the last 90 days and are not in a hiring freeze.` - [run the search](https://www.refolk.ai/start?q=Software%20engineering%20roles%20at%20US%20companies%20that%20have%20grown%20headcount%20in%20the%20last%2090%20days%20and%20are%20not%20in%20a%20hiring%20freeze.).

*Returns companies whose headcount is actually rising, so the postings you grade start from live-demand employers rather than freeze-context ghosts.*

## How the reality check fails: false positives to guard against

This standard fails in two directions. It passes dead listings when you trust the wrong signal, and it cuts live reqs when you apply a signal too bluntly. The failure modes below are the most valuable part of the standard, because a check that overclaims is worse than no check.

| Failure mode | What it looks like | The check that catches it |
|---|---|---|
| Trusting the board timestamp | Board shows "posted 2 days ago" for a filled role | Use the careers-page or ATS date; the aggregator lags 24 to 72 hours |
| Treating every repost as dead | Dropping a live urgent req that auto-refreshed | Flag only content change or a new req number with recruiter silence |
| Failing technical reqs on age | Cutting a genuine 55-day engineering role | Apply role-specific thresholds; engineering averages 50 to 62 days |
| Missing layoff context | A "fresh" posting at a company in a freeze | Cross-reference layoff or freeze news against open-req count |
| Reading low response as personal | Assuming your resume is the problem | A 2% observed rate with 30% ghosts implies ~2.9% among real roles |
| Ignoring the 48-hour window | A real role you applied to on day 12, never reviewed | Apply within the first 48 hours; after 7 days it may not be seen |

Two of these deserve extra weight. The first, trusting the board timestamp, is the most common way a dead listing passes. The board is a re-index event, not a status check. Always drop to the source.

The last, the 48-hour window, is the failure that makes a genuinely live posting into a dead application. Postings receive the most applicants in the first 48 hours, and after seven days an application may never be seen, because recruiters stop reviewing once the pipeline fills. This is why grading the batch and then acting fast beats grading slowly. A posting can pass every liveness check and still be a wasted application if you arrive on day ten. Your grading discipline has to be fast enough to preserve the window.

> **Tip:** Grade fast, apply inside 48 hours
>
> The batch grade is only useful if it clears the same day you found the postings. Volume peaks in the first 48 hours, so a survivor you tailor a week later has already lost most of its odds. Timebox the grade to one sitting.

There is also a quiet false positive on the cutting side: dropping a real role in an oversupplied market because its expected yield looks low. Low yield is not the same as dead. A role with 200 applicants is competitive, not fake. Keep it if it passes the liveness checks, and simply price the effort accordingly.

## The pre-tailor checklist

Before you commit a single tailoring hour, run the surviving batch against this list. If any item fails, the row is not ready.

#### Verify before you tailor

- [ ] Every posting has a source-verified age from the careers page or ATS, not the board timestamp
- [ ] Age is graded against the role type, with engineering held to 50-62 days not 30
- [ ] Reposts are flagged only where content changed or a new req number appears with recruiter silence
- [ ] No survivor sits at an employer with recent layoffs or a freeze plus many open reqs
- [ ] The surviving set's estimated contamination is at or below roughly 20%
- [ ] Each survivor carries a live-verification note confirming on-source status and, where visible, engagement
- [ ] The grade was completed the same day, so you can apply inside the 48-hour window

#### From committed list to tailored applications

| Stage | Figure | Note |
| --- | --- | --- |
| Committed postings | 100 | everything you planned to tailor |
| Pass source-verified age | 78 | after removing stale and unverifiable rows |
| Pass repost and employer checks | 65 | after cutting churned reqs and freeze-context postings |
| Clear 20% contamination line | 60 | the surviving batch worth your tailoring hours |

*A batch of 100 committed postings narrows to a live, worth-tailoring set after each grading gate.*

The funnel figures are illustrative of the shape, not a measured outcome. Your own narrowing depends heavily on sector, because ghost rates cluster by field. Run the grade, record your actual pass rate, and let that number tell you whether your sourcing is healthy over time.

## Keeping the standard current

Re-check the mechanisms, not the values, because the numbers move but the levers do not. Two facts drive everything here: the aggregator lag that makes careers-page timestamps authoritative, and the role-typed fill times that set your freshness lines. Both can shift.

Re-anchor your freshness thresholds against a current time-to-fill benchmark each cycle. SHRM's figure already moved from 48 days in 2023 to 41 in 2024, so the "30 to 45 days" line is a starting point, not a constant. Pull the latest sector time-to-fill you can find and adjust the borderline and stale lines to match. The engineering exception at 50 to 62 days is the pattern to preserve even if the exact days change: technical and regulated roles run longer, and your standard must not punish them for it.

Re-check the aggregator-lag window too. If boards start syncing closes faster than 24 to 72 hours, the careers-page cross-check becomes cheaper but no less necessary. Until then, the careers page remains the single highest-value check, and the board timestamp remains the single most common lie. Keep the source check as your first gate and the rest of the standard holds.

One last discipline. Log your realized response rate against your graded batches, separated by channel using the response-rate table as your baseline. When Indeed applications land near 20 to 25% and company-website applications land near 2 to 5%, your grading is sound and your yield is a channel effect, not a batch failure. When even your Indeed survivors return near zero, the standard is telling you something is wrong upstream, in the sourcing or the timing, and it is time to re-run the whole grade before you write another cover letter.

## Frequently asked questions

### Is this job posting worth applying to if it is 40 days old?

It depends entirely on the role type. Most legitimate openings fill within 30 to 45 days, so a 40-day general posting is borderline and needs a live-verification check. But engineering roles average 50 to 62 days to fill, so a 40-day engineering req can be genuinely open. Grade age against the function, not a blanket cut, then confirm on the careers page before you invest tailoring hours.

### How do I spot ghost jobs before applying without special tools?

Run three free checks. First, compare the board timestamp to the company careers-page or ATS date, since aggregators lag 24 to 72 hours and the career page changes first. Second, check whether the employer announced layoffs or a freeze while carrying dozens of open reqs, which is a strong ghost signal. Third, look for a repost with changed content or a new requisition number and no recruiter contact. Two of three failing is enough to cut.

### Is a reposted job listing still worth applying to?

Often yes. Auto-refresh on a fixed schedule is routine, and LinkedIn roles open longer than 30 days can be auto-reposted under enterprise agreements, so a repost alone proves nothing. A repost only becomes a meaningful ghost signal when something changed in the listing, such as title, seniority, or salary band, or when a new requisition number appears alongside no recruiter contact. Treat a plain refresh as neutral.

### What ghost-job rate should I assume when grading a batch?

Use about 20% as a defensible working cap, which mirrors the lowest ATS-grounded figure. Greenhouse platform data classifies 18 to 22% of quarterly postings as ghost jobs, Ashby data runs near 18%, and JOLTS gaps imply around 30%. Employer self-report is higher, with four in 10 companies posting fake listings in 2024. Sector matters more than any tactic: construction runs 38%, art 34%, and legal 29%.

### Why is my application response rate so low even for real jobs?

A low rate mostly measures contamination, not your candidacy. The industry-wide average is about 2 to 3%, but dead listings return 0% and drag that visible average down. A 2% observed rate against roughly 30% ghost jobs implies about 2.9% among real open roles. Channel and timing also dominate: company websites run 2 to 5%, tech roles 0.5 to 2%, and postings get the most applicants in the first 48 hours.

---

*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/live-demand-application-batch-standard*