The margin problem in most field service operations is not in the estimate.
The estimate is usually fine. The scope is priced correctly. The labor rate reflects the market. The materials are costed within normal range. The job, on paper, makes money.
The margin problem is in the gap between what the estimate assumed and what the field actually did.
That gap has a name. It is called rework. And most field service operators do not measure it — not because they do not care about margin, but because their systems are not set up to surface it. "Complete" is a status in the work order system. It is not a verification. The job is marked done when the PM gets a call from the crew. The cost of the second trip, the third follow-up, and the material re-order does not appear next to the original job. It appears somewhere else in the P&L, as a service call or a supply order or an overtime charge, with no clean line back to the job that caused it.
By the time the period closes, the rework has been absorbed and distributed across so many line items that no one can find it. The operation looks like it has a cost problem. What it actually has is a first-time-completion problem.
What rework actually costs
Rework in field service is not just the direct labor and materials of the second trip. It is the full chain of costs that the second trip generates.
The technician who goes back is not running a new job. They are running a lost-margin job — a job where the revenue has already been recognized, the customer has already been invoiced, and every hour spent on that site is pure cost. There is no scope expansion, no upsell, no additional billing. The trip exists because the first trip did not finish.
That technician's time is also not available for a paying job. Field capacity is finite. When a tech is on a callback, a new work order is either waiting, being reassigned to a subcontractor at margin cost, or generating overtime for another tech. The direct cost of the callback is the tip of the problem. The opportunity cost is usually larger.
Customer relationships absorb rework quietly for a while. Then they do not. A second trip is an inconvenience. A third trip is a pattern. A fourth trip is a contract conversation. The revenue at risk from rework is not visible in the P&L until the renewal conversation goes badly.
None of this appears in most field service reporting. The job was completed. The customer did not formally escalate. The tech ran the callback under a service order that was absorbed into overhead. The estimate-to-actual on the original job looks fine because the rework was never attached to it.
How to find it
The first step is not a new system. It is a question the existing system can answer if you ask it correctly.
For every job marked complete in the last 90 days: was a service order opened on the same site, by the same customer, within 30 days of close? If yes, that is a potential callback. Pull the list.
This is not a perfect rework report. Some of those service orders are unrelated to the original job. Some are legitimate follow-on work the customer requested. But the list gives you a place to start, and in most field service operations, the size of the list is immediately useful information.
Sort it by original job. How many callbacks per job type? How many per crew member? How many per subcontractor? How many per project manager? The distribution almost always tells a story. Rework is not evenly distributed across the operation. It clusters around specific jobs, specific crews, and specific process failures that are repeated.
That clustering is the operational insight. The question is not "how much did we spend on rework this quarter." The question is "which three things are generating most of the rework, and what is the operational cause of each one."
What the fix usually is
In most engagements, the rework clusters around the same two or three failure points.
The first is scoping. The job was sold with an assumption that did not match the site. The crew arrived and found a different layout, missing infrastructure, or scope the customer described differently than what the estimator understood. The first trip addressed what the crew could address. The second trip addressed what they found when they got there.
The fix is a site confirmation step — a walk or a form or a photo review — that happens between contract and schedule, not after the crew is on site. Define "ready" before you schedule, not after you dispatch.
The second is closeout. The job was done, mostly. Some items were left open — a punch list that someone was going to come back to, a test that needed a customer contact who was not available, a document that needed to be signed and was not. The job got marked complete because the significant work was done, and the open items were small enough that no one prioritized a return trip. Until the customer called.
The fix is a closeout gate. The job cannot be marked complete until the punch list is confirmed empty and the closeout packet — sign-off, photos, warranty, next steps — is submitted. This is an operational rule enforced in the work order system, not a conversation about being more careful.
The third is subcontractor management. The callback pattern clusters around specific subs who regularly leave open items, do not follow the closeout process, or do work that requires a return trip to correct. Without a scorecard that tracks callbacks by subcontractor, the operations manager is running on reputation and instinct. The rework is invisible because it is not attributed.
The measurement is the intervention
In most field service operations, introducing a first-time completion rate — a single metric that tracks whether a job required any return trips within 30 days — changes behavior before anything else changes.
The crews know the number exists. The PMs know the number is tracked by crew. The subcontractors know the number is part of how they are evaluated for future work. That visibility alone — not new software, not new training, not new process — reduces the rate of "mostly done" closes.
Rework drops when it is visible and attributed. It is high when it is invisible and absorbed.
Before deploying any AI on job costing, scheduling, or dispatch: know your first-time completion rate by crew, by job type, and by subcontractor. That number tells you more about where your operation is bleeding margin than any report your accounting system can run.
Fix the standard. Count the rework. Then decide what to automate.