How to Track Labor Hours Against the Estimate

Why this matters

Labor is where estimates miss most, and it is the one bucket you can still do something about while the job is running. Material is committed the moment it is ordered. Hours are spent one at a time, and a job that is going to run 15% over usually shows that in the first day, to anyone who is measuring the right thing.

Most shops measure the wrong thing. They look at hours consumed against total budget and feel fine, because they are 60% through the budget and it is only day two. Consumed hours against total budget tells you nothing on its own. The number that matters is hours consumed against work actually completed, and it is a different number.

Step 1: Convert the estimate into a labor budget the crew can see

The frozen estimate holds an hours figure. Turn it into a budget the people burning it can read: total hours, split by phase, printed or on the work order. Not the price, not the margin, just the hours.

Shops hesitate here, worried that a visible budget makes crews rush or pad. In practice the opposite happens. A crew given no target logs honest hours and has no way to tell you they are in trouble until the job is over. A crew given a phase budget will tell you at lunch on day one that the rough is not going to fit, which is worth more than any close-out report.

If you genuinely do not trust a specific crew with the number, that is a personnel question, and hiding the budget from everyone to manage one person costs you the early warning on every job.

Step 2: Split the budget by phase, not just a total

A single total is useless in flight, because there is no way to know whether you are ahead or behind until the end. Split the total into three to six phases that have a clear "this is done" line: prep, rough, set and connect, commission and clean, for example. Every phase needs a completion test a tech can answer yes or no to.

Three to six is the working range. Two phases give you one checkpoint, which is too late. Ten phases turn logging into a chore and the crew starts guessing which bucket a task belongs in, which corrupts the data you are collecting.

The split also does something the total cannot: it tells you where the hours went, not just that they went. A job that lands 12% over is a mystery. A job that lands 12% over because the rough phase ran 30% over while everything else landed clean is a specific, fixable finding.

Step 3: Log at the moment, to the phase, with rules everyone shares

Three rules, non-negotiable, written down once:

• Log at the transition, not at the end of the week. Hours reconstructed on Friday for Monday are estimates of an estimate. They regress toward the budget because that is the number the person remembers, and your variance quietly collapses to zero.
• Round to a consistent increment, 0.25 hours is the usual choice. Finer than that is false precision on a manual log. Coarser than that hides a real half hour on a short job.
• One phase code per block of time. If a tech genuinely split an hour across two phases, they pick the dominant one. Forcing precision the log cannot support just produces slower, worse logging.

Travel, warranty return trips, and truck stock follow whatever definitions your close-out procedure already set. What matters here is that the in-flight log uses the same ones, or your mid-job projection will not match your close-out variance and nobody will trust either.

Step 4: Put checkpoints where a correction is still possible

Set the checkpoint at each phase boundary, plus one hard checkpoint at roughly 50% of budgeted hours on any job long enough to have a middle.

The 50% mark is chosen for a specific reason: it is the last point at which a correction can still change the outcome. A problem found at 80% of budget can only be absorbed or billed, and by then the hours are spent. A problem found at 50% still has half the job left to recover in, or half the job's worth of scope left to write a change order against.

On a one-day job, the checkpoint is lunch. That sounds informal and it works: "where are we against the four hours" at midday is a real checkpoint that has saved more jobs than any report.

Step 5: Compare consumed hours to earned hours, not to elapsed time

This is the step that does the work. At each checkpoint, compute two numbers:

• Consumed: hours actually logged so far.
• Earned: the budgeted hours for the phases that are genuinely complete. Not partially complete. If a phase is half done, count half its budget only if you can honestly assess it, and default to counting completed phases only.

Divide consumed by earned. That ratio is your burn rate. The sibling article on handling a job running over calls the same quantity the burn ratio and states it as percent-consumed over percent-complete; the two are algebraically the same number, so a reader meeting both is not looking at two techniques. It projects: multiply the remaining budget by the same ratio, add the consumed hours, and you have a projected total. A burn rate of 1.10 means you are spending 10% more hours than the plan for the work you have finished, and unless something changes, the whole job lands about 10% over.

The reason to project rather than to look at raw consumption is that raw consumption always looks fine early and always looks alarming late. The burn ratio reads the same at 20% through as it does at 80% through, which is what makes it an early warning instead of a postmortem.

Step 6: Act on the checkpoint, and know which of the three moves you are making

A projection past your action threshold (a projected overrun above 10% of budget is a workable default) forces one of three moves. Name which one you are making, because they have different consequences:

• Write the change order now. Only valid if the overrun is added scope. The moment you identify work outside the bid, it gets documented and priced before the hours accumulate, not after. Retroactive change orders are the hardest conversation in the trade and most shops eat them.
• Fix the sequence for the remaining phases. Valid when the cause is method: wrong staging, wrong tool on site, work being done in an order that forces re-handling. This is the only move that actually improves the projection, and the improvement applies only to the phases still ahead of you.
• Absorb it and record the cause. Valid when the conditions are simply harder than bid and no scope was added. You will not recover the hours. The action is to protect the rest of the job and write the cause line while it is fresh, because this job's overrun is next quarter's template correction.

One move that is not on the list: adding a body to catch up. Adding a second tech to a phase compresses calendar days, not labor hours, and usually increases total hours slightly through coordination and duplicated setup. If the customer needs it done by Thursday, add the hand and understand you are buying schedule with hours. If the problem is the hours themselves, another hand makes it worse.

A worked job, checkpoint by checkpoint

A four-phase install budgeted at 40.0 labor hours:

End of day two, the crew has finished prep and demo and finished rough. Nothing on set and connect yet. The log shows 24.5 hours consumed.

The comfortable read: 24.5 of 40.0 is 61% of the budget consumed, day two of a four-day job, plenty left. That read is wrong, and it is the single most common way a job gets to day four before anyone notices.

The right read: earned hours are 6.0 plus 16.0 equals 22.0, because those are the two phases genuinely complete. That is 55% of the budgeted work done. So the crew consumed 61% of the budget to earn 55% of it.

Burn rate is 24.5 divided by 22.0, or 1.114. Applied forward: the remaining budget is 40.0 minus 22.0 equals 18.0 hours, and at the same rate that becomes 18.0 times 1.114, about 20.1 hours. Projected total is 24.5 plus 20.1, about 44.6 hours against a 40.0 budget, roughly 11.4% over.

That clears the 10% action threshold, so the checkpoint forces a decision. The job owner walks the causes: no scope was added, the customer asked for nothing extra, but the rough took longer because the crew staged material at the truck and carried it in piece by piece across a long run.

That is a method cause, so the move is fix the sequence. Staging moves inside for the set and connect phase, and the two remaining phases get run with material at the work face instead of at the curb.

The job closes at 43.0 hours against the 40.0 budget, 7.5% over. The sequencing change recovered about 1.5 hours off the 44.6 projection, which is a real save and also an honest one: it did not undo the rough phase's overrun, it only stopped it from repeating in the last two phases. The close-out cause line reads: long carry from truck to work face, stage inside on this building type. That line is what turns one job's 7.5% into a checklist item that stops the next four.

What changes the answer

• Very short jobs. On anything under about four hours, phase splitting and burn ratios are overhead with no payoff. Track total hours against the estimate and read it at close. The in-flight machinery earns its keep on multi-day work.
• A crew you have not run this job type with before. The first job's burn rate is measuring the crew's learning as much as the estimate. Project from it, but do not correct the template from it.
• Time and materials. The customer absorbs the hours, so the projection is not protecting your margin. Run it anyway and use it as a customer-communication trigger: a T and M job tracking 30% past the verbal ballpark needs a phone call before the invoice, not after.
• Weather or access holds. Hours the crew spent waiting are real cost but they are not burn rate on the work. Log them to a hold code and keep them out of the earned-versus-consumed ratio, or a two-hour lockout will read as a method problem and you will go fix a sequence that was fine.

How to verify you got this right

• The phase budgets sum to the total. In the example, 6.0 plus 16.0 plus 12.0 plus 6.0 equals 40.0. If your phases do not sum, one phase is absorbing everything nobody could categorize and your phase-level findings are fiction.
• Consumed and earned are both stated with the same base. "24.5 consumed against 22.0 earned, out of a 40.0 hour budget." Percentages without the base attached are how a 55% and a 61% get read as the same number.
• The close-out total matches the sum of the phase logs. If they disagree, hours are being logged outside the phase structure, and your burn ratios were computed on partial data.
• Your projections get less wrong over time. Compare the 50% checkpoint projection against the final actual on ten jobs. If the projection is consistently optimistic, your phases are back-loaded: the later phases carry more of the real work than their budgets say, and the split needs rebalancing even though the total is right.

The failure mode to watch for is a crew that logs hours to keep the burn rate at 1.0. It looks like discipline and it is the death of the whole system. The tell is a job whose phase logs all land within a few minutes of budget and whose total still comes in over, because the unlogged time has to surface somewhere. When you see it, the fix is never a lecture about honest logging. It is removing whatever consequence taught them that an over number is dangerous to report.

References

• U.S. Small Business Administration: labor cost tracking and job costing guidance for small contractors.
• Standard construction practice for labor budgeting, phase coding, and earned-hours progress measurement.
• See related: How to Compare Estimated Against Actual on Every Job.
• See related: How to Capture Actual Costs Without Slowing the Crew.
• See related: The Variance Threshold Worth Investigating.