What Your Stale Job Backlog Is Costing You

Every Open Job Is an Open Wound

A job sitting open past 90 days needs a documented reason and the relevant financial records. It may be active long-duration work, warranty follow-up or an administrative status gap. The effect may not appear on a single P&L line. The questions are concrete: are costs supported, are any contract-eligible invoices missing, and does the accounting treatment match the actual work? Open field status alone answers none of them.

Most contractors don't think about job duration as a financial metric. They think about it as an operational one, if they think about it at all. The earlier article reported an internal analysis of 2.1 million completed jobs. Its source extract and eligible company denominator are not reproduced here, and no predictive or causal margin model is supplied. Treat duration as a diagnostic input, not proof of realized margin.

The Benchmarks

The earlier article's duration summary is retained as unreproduced historical context, not a current verified Level Index distribution or industry-wide benchmark. These figures cannot establish a trade comparison or billing/cash effect, and are not the share of a company's current open backlog:

The first thing to notice: median is 10 days, average is 36 days. That's a massive long tail dragging the average up. Long-duration jobs can have legitimate scope and progress billing. The retained distribution does not measure their margin or the reason they remain open.

For a hypothetical 2,000-job population at an assumed 11.2% past 90 days, the count is 224. That is a planning illustration using the historical input, not a claim that each job accumulates cost or delays billing.

The Variance Is Staggering

Not every contractor has the same backlog problem. The retained planning profiles span different work mixes and review practices:

The table is a retained operator-planning comparison, not reproduced company percentiles. A short-duration profile and a 25%+ long-open profile need different review populations; neither establishes margin or financial health without costs, invoices and remaining-scope evidence.

The published cross-trade research does not establish a plumbing-versus-HVAC split or a causal explanation for this spread. Work mix, status conventions, active project scope and review ownership are candidate explanations to test. A planning case at 25%+ over 90 days does not alone prove process failure.

What Stale Jobs Actually Cost You

1. Trapped WIP

Where approved accounting policy capitalizes eligible job costs in WIP or inventory, inspect the supported asset and expense timing. Other contractors expense costs as incurred and use WIP as a billings-versus-earned schedule. An open operational job is not automatically a balance-sheet asset or an overstatement. Confirm the method with the accountant.

In a fictional planning example, 200 open jobs with $500 each of supported capitalized cost would total $100K. Those are invented inputs, not proof that 200 long-open jobs carry that asset or cash exposure. Reconcile the actual cost, billings, recognition and bank records before using the result.

2. Labor Overruns Nobody Catches

Separately, the Level Index labor-hours metric covers 315,393 jobs with both budgeted and actual hours across 1,391 companies: 40% exceed budgeted labor hours, and 18.3% exceed 150%. This is not a long-open-job subset and does not establish a duration-overrun relationship. Unreviewed jobs can hide additional hours to check. A tech goes back for a follow-up. Another one logs time to troubleshoot a callback. The PM spends an hour on the phone with the customer sorting out a punch list.

Check the actual hours, approvals and remaining scope at each review; short duration alone does not prove that overruns are noticed, and long duration does not prove that nobody is reviewing.

3. Missed and Delayed Billing

The verified Level Index closeout metric is work complete to recorded financial job closeout: median 1.7 days, P75 9.6 and P90 35.8, with n=555 completed-job observations. It does not establish invoice issuance or cash-collection timing. The earlier 14-day average is not supplied in the current canonical metric.

An earlier internal count reported about 288,000 completed jobs with logged hours and no linked invoice. That unreproduced count has no established eligible job/company denominator here and is not confirmed missed revenue. Investigate warranty, agreement/bundled work, internal jobs and invoices attached to a different job before deciding whether a bill was omitted. Separately, a fictional job lasting 120 days, followed by 36 days to its final invoice and 30 days to payment, would take 186 days to final cash. Those assumptions do not combine empirical percentiles into an observed path, and progress billing may produce cash earlier. If you're not tracking billing speed on a per-job basis, these slow-closers are invisible.

4. Phantom Margin

An open or closed marker does not establish complete costs or recognized margin. Compare estimates with the approved cost and revenue population at the cutoff, using a supported job-costing reconciliation. The longer a job stays open, the more important it is to check that the displayed margin reflects approved costs, change orders and warranty work. Additional labor, unreturned materials, warranty visits, scope additions without change orders. All of these erode the actual margin, but the estimated margin doesn't update unless someone forces the reconciliation.

I've seen contractors with healthy-looking job margins in their system discover, after a backlog cleanup, that a dozen long-running jobs were actually breakeven or negative. In those cases, the backlog was hiding the losses.

Why Jobs Go Stale

These are operator hypotheses to check, not measured cause frequencies:

1. No formal closeout process. The tech finishes the work but the job stays open because nobody owns the closeout step. The office assumes the tech will close it. The tech assumes the office handles it. The job sits in limbo.

2. Waiting on the customer. Punch list items, final inspections, customer sign-offs. The job is 98% done but stays open waiting for something outside your control. Completed work may not yet be invoiced; check contract eligibility and the progress-billing schedule.

3. Warranty or callback ambiguity. The original job spawns a callback. Instead of creating a new job for the warranty work, the tech logs time against the original job. The original job never closes because there's always "one more thing."

4. Nobody's looking. Check whether anyone owns the aging report and its exceptions. If nobody runs a report on jobs open longer than 90 days, nobody knows they exist. The jobs sit in the system, accumulating cost, and the only person who notices is the CFO doing year-end reconciliation.

5. The PM left. Project managers leave, retire, or switch roles. Their open jobs become orphans. Nobody inherits the backlog review. Six months later, there are 40 jobs assigned to someone who doesn't work there anymore.

How to Implement Backlog Hygiene

Weekly Backlog Review

Every Monday, run a report: all jobs open longer than 60 days. For each one, the assigned PM must answer three questions:

1. Is work still actively happening? If yes, is the budget tracking to plan?
2. If work is complete, why hasn't it closed? What's blocking closeout?
3. Has the customer been invoiced for completed work? If not, verify contract billing eligibility and send a progress invoice where permitted.

A short review may take roughly 30 minutes depending on volume and evidence quality. It can identify candidate status, billing and time-allocation exceptions; confirm the support before taking action.

90-Day Escalation Trigger

For a chosen 90-day review threshold, use alerts where the installed system supports them, or review a saved report. Escalate unexplained exceptions to the operations manager or controller while preserving the PM's scope explanation. Legitimate long-duration work should remain open.

The escalation requires a written justification: why is this job still open, what's the remaining scope, what's the expected close date? If the support is incomplete, retain the exception and assign an owner. Do not force-close a job merely to clean the dashboard.

Closeout Process: Make It a Defined Step

Closeout isn't "the job is done." Closeout is a defined checklist:

1. All labor hours reviewed and approved
2. All materials reconciled (purchased vs. used vs. returned)
3. Final walkthrough or customer sign-off complete
4. Final invoice sent
5. Job status changed to "closed" in the system
6. Budget vs. actual variance reviewed

Until all six steps are done, the job is open. And someone owns that checklist, typically the office admin or project coordinator, not the tech.

Separate Warranty Work

Create a new job for every warranty callback. Link it to the original job for tracking, but don't log warranty labor against the original job number. This keeps the original job's cost accurate and allows it to close on time. It also gives you a clean view of warranty cost as a percentage of revenue, a critical metric for understanding true job margin.

Set a Hard Cutoff

Pick a number. 120 days. 150 days. Whatever makes sense for your trade. Review every job past that cutoff in a monthly exception process. The authorized controller verifies remaining scope, costs, billing eligibility and approvals before any close or invoice action. Keep legitimate long-duration work and unresolved exceptions open with a named owner.

This sounds aggressive. It is. But the alternative is a backlog that grows forever, with each stale job making your financials a little less trustworthy.

---

The Bottom Line

The retained historical duration inputs are 10 days median, 36 days mean, 11.2% past 90 days and 4.1% past 180 days; they are not reproduced current benchmarks. Neither they nor the retained under-1% to 25%+ planning profiles identify a process-versus-trade cause. Use your own scoped aging report to establish the actual population and open-job reasons.

A stale job can expose incomplete cost, billing or status evidence, but the effect depends on its records and accounting policy. A weekly review, a selected aging trigger and a defined closeout process can surface missing cost, billing or status support. Whether cash is released depends on contract terms, eligible billings and collections.

If you've never run a report on jobs open longer than 90 days, do it this week. Document the population and the reason for each exception.

Q: How does Level help with backlog management? A: Where the installed systems expose supported job status, cost, budget and invoice records or exports, we build an aging view with original job IDs and source cutoffs. We flag 60-, 90- and 120-day review populations, mark missing cost or budget support as unknown, and have a person assess each exception before financial reliance. The first audit is free.

Q: What's a healthy percentage of jobs over 90 days? A: Level does not publish a verified over-90-day benchmark. The retained 11.2% historical input is unreproduced and has no established trade split here. Set your review threshold by work mix. Separate active progress-billed projects from jobs with no recent cost, billing or status activity, then review those with a named owner.

Q: Should I worry about jobs that are legitimately long-duration? A: Long jobs aren't the problem. Unreviewed long jobs are. A 9-month commercial buildout that's actively managed, progress-billed monthly, and on budget is fine. A 9-month job where nobody's checked the budget since month two and no invoice has gone out since month three is a cash flow emergency. The distinction is whether someone is actively managing the job's financial performance, not just its operational status.