Estimate Accuracy — Sold vs Invoiced vs Worked

ABQ Plumbing Heating & Cooling · jobs and projects completed in the last 12 months · sold hours and material cost against the work · full drill-through
Loading…

How far off are estimates?

Each job's worked hours divided by its sold hours. The middle band is within ±15% — on target.

Hours — worked ÷ soldMaterial — procured ÷ estimatedOn target ±15%
Rolling three months · one scale for both, 1.0× is exact

By department

Grouped by the department that sold the estimate. Click a row for its estimators and jobs.

By business unit

Estimators

Credited to whoever sold the largest estimate on the job or project. Score is provisional — a 1–10 band on the average miss. Click a card for the trend, the recap and every job behind it.

ImprovingRegressingSteady / too few recentcircle size = jobs

What the company can do about chronic shortfalls

Generated from the live numbers on the default view (2+ sold hours, install work only, anomalies excluded) each time the page loads. It does not change with the filters above.

Every labor line item

Worked hours are shared across a job's line items in proportion to their sold hours. "Clean reads" are jobs where one item is at least 80% of the sold hours — the most trustworthy evidence for that item. Click a row for the jobs.

Labor added to invoices that was never estimated

Service lines on the invoice with no link back to an estimate line. Repeat offenders are scope the estimate keeps missing — except diagnostic fees on service calls, which precede the estimate by design.

Every job and project

Sorted by the largest miss in hours. Click any row to verify it: estimates, jobs, worked hours, invoice lines and material, each linked to ServiceTitan.

About this build. Every job or project completed in the rolling 12 months ending two weeks ago that carried a sold estimate — the two-week gap lets payroll hours catch up. A job with no project is measured on its own; a job on a project is measured with every job on that project, so a sale made on one visit and installed on another is judged as one piece of work. Figures are recomputed in Peaka every six hours.

Three hour figures
Sold is estimate line hours × quantity, from every sold estimate on the job or project, change orders included. Invoiced is the same on the invoice, split into lines carried over from the estimate and lines added afterwards. Worked is paid timesheet hours from payroll. Invoice sold hours are stored per unit in ServiceTitan, so they are multiplied by quantity here.
Why sales visits are excluded by default
On a project, the job where the estimate was written is the sales visit. Those hours are selling time, not install labor — roughly 30% of all paid hours on projects. Counting them makes estimates look about a third short when the install labor itself runs close to the sold hours. Switch to "+ sales visits" to see the difference. On a single service job, worked hours include the diagnosis before the estimate was written, so single-job ratios run high by design.
Material cost
Estimated material is the cost on the estimate's material and equipment lines. Actual is what was procured for the work: purchase orders plus transfers, excluding permits, subcontractors and equipment rental. Invoice material cost is shown but not used, because most equipment lines on invoices carry $0 cost.
What this cannot separate
An overrun can be a scoping miss or a slow crew. At this level the two look the same, so an estimator's number includes the installers who did the work. The line-item view is the better guide to where the price book itself is wrong.
Anomalies
Jobs with over 40 paid hours and more than five times the sold hours are flagged and excluded by default — they are jobs used as a timesheet bucket, not estimating misses. Extra invoice lines over 100 hours on a small job, such as a membership renewal SKU, are also zeroed.
Recaps
Generated from each estimator's live numbers — bias, spread, materials, where their overruns concentrate and their trend — on the default view, every time the page loads. They describe patterns, not people; read them alongside the drill-down.
Scores
Provisional 1–10 bands on the average miss: 20% or less = 10, then 25, 30, 35, 40, 50, 60, 75 and 100%. Not yet a company-approved scale — change the bands once a card exists.