Scoring…
Score a bid first — the band is built around that bid.
Bars to the left of today are hours the crews actually worked. Bars after are the hours still to run on jobs already sold, spread at the pace each is burning. The line is the achievable run rate for that calendar month — Warren works ~2.5× the hours in August that it does in January.
Green is hours already sold. Blue would be pipeline hours scaled by win probability — empty until the pipeline source of record is decided. The line is what the crews can work in that month; anywhere the stack sits below it is capacity not yet sold.
A second opinion on the numbers in a bid, benchmarked against what completed Warren jobs actually did. It does not replace a project manager's read on scope, access or the customer. It catches what a spreadsheet cannot see: a labor allowance sitting below the historical median for that material scope, no rental on a job size where half of jobs incur some, a peak cash draw over 15% of contract.
| Field | Where it comes from | Watch for |
|---|---|---|
| Contract price | The number you intend to submit | Sets the size band, which drives most benchmarks |
| Labor, burden included | Cost class 1 in the estimate | Burden belongs inside this figure; the estimate carries it there, the actuals split it out |
| Estimated labor hours | Man-hours, not man-days | Drives expected labor cost directly |
| Materials | Cost class 2 | Exclude anything the manufacturer bills the customer for directly — bid Warren's scope only |
| Rental | Cost class 5 | Enter zero if none planned; zero on a job over $40K of cost triggers an imputation |
| Misc | Cost class 7, excluding the overhead allocation | Overhead is entered as a percent of labor and recalculated on expected labor |
| Project manager / lead foreman | Whoever will run it | Their measured hours record moves expected hours, shrunk toward neutral by how many jobs they have behind them |
| Season | When the bulk of the work happens, not when it is signed | Moves margin risk and the comparables cell, not expected hours |
| Start month, duration | Best estimate | Drives the cash walk, the margin curve and the labor rate at the job's midpoint |
| Cash terms | Contract terms | Retainage, owner and vendor lags drive peak cash need |
| Output | Read it as |
|---|---|
| Expected final margin | The middle of a range, not a point. Jobs land within about half the comparables' interquartile spread either side of it. |
| Risk points | How many things look wrong with the bid. 0–2 clear · 3–5 review the flags · 6+ senior review. |
| Confidence | How much to trust the margin being predicted. Independent of risk: a clean, simple job in work Warren has rarely done can be low risk and low confidence at the same time. |
| Cost bridge | Each cost line as estimated and as history says it will land, with the ratio or rate that got there and the snapshot field it came from. |
| Labor reality check | Estimated hours × the book ratio × the PM and foreman effects = expected hours, priced at the burdened rate expected at the job's midpoint. |
| Margin by month | The margin you would report each month if costs land where history says. It hugs the bid margin early and pulls away late — a quarter of the move by halfway, 44% in the final quarter. |
| Stochastic band | On its own tab: the same line with a shaded range one standard deviation either side, built from how widely jobs of this size and season finished around their expected margin, tightened or widened for the PM's consistency. Two jobs in three finish inside it. Read it for the size of the downside, not as a promise of the upside. |
| Cash walk | Arithmetic, not inference: payroll out as incurred, vendors on the lag, billing net of retainage on the owner's lag, retainage released after completion. The cumulative line lands on expected gross profit. |
| Escalation | Fires when any one is true: risk ≥ 6 · contract ≥ $1.5M · expected margin below the 22% floor · confidence below 35 · risk ≥ 3 on confidence below 55. Advisory until adopted as policy; Warren's standing control is a second-PM review over $250K. |
Every benchmark is rebuilt from Foundation (Reporting_Warren) and WREN into a dated snapshot; the header shows which one this page is using. Nothing is typed in. The live book is rebuilt the same way. The one shape still borrowed from the original tool is the square-of-percent-complete margin path, measured on 236 jobs with monthly history that the snapshot does not yet hold.
| Term | Meaning here |
|---|---|
| Panel | Completed Warren jobs over $10K, finished 2020 to the snapshot date, with an estimate, labor hours, cost and billing on record. Rule PN-1. |
| Funded | The original estimate plus approved change-order scope. Foundation never revises a budget after award, so a job that grew reads as over-hours unless the change orders are credited back. Status A change orders only; monthly re-forecasts (status E) are not new work. |
| Book ratio (hours) | Actual hours ÷ funded hours, geometric mean across the panel. Below 1.00 means jobs finish under their hours estimate on average. |
| PM / foreman effect | How a person's jobs land against their own estimates, relative to the book, shrunk toward 1.00 by n ÷ (n + 18) so a short record moves the number less. Applied together, so a foreman who always works for a tight-estimating PM is not blamed for the estimates. |
| Burdened rate | Wages plus employer burden per hour, from job_history classes 1 and 6 over timecard hours, for the latest complete half-year, escalated at the measured annual rate to the job's midpoint. |
| Cost-class ratio | Actual cost ÷ funded cost for a class across the panel. Aggregate is dollar-weighted; median is the middle job. Material pass-through jobs (billed direct by the manufacturer) are set aside. |
| Rental imputation | When zero rental is estimated on a job over $40K of cost, the median rental share that zero-rental jobs actually incurred is added. |
| Bid margin / final margin | Bid: (contract − original estimate) ÷ contract. Final: (billed − actual cost) ÷ billed. Move: final minus bid, in points. Fade: a move below zero. |
| Size band × season cell | Five contract-size bands × four seasons. The cell's job count and interquartile range of margin move drive the comparables, the confidence score and the band width. |
| Season | Where the labor hours landed: Shoulder Nov–Mar · Spring Apr–May · Summer Jun–Aug · Fall Sep–Oct. |
| Standard deviation | One number for how widely jobs like this one have scattered around their expected margin: interquartile range ÷ 1.349. Roughly two jobs in three land within one of them either side of the middle. |
| PM consistency | A PM's median absolute hours miss against the book's, shrunk by n ÷ (n + 20) and clamped 0.75–1.60. Tightens or widens the outcome spread and the band identically. |
| Implied labor factor | (Price − 1.10 × non-labor cost) ÷ labor. What the price implies labor must be covered by after every other line is marked up 10%. Projected uses expected costs. |
| Live book | Active jobs over $10K with funded hours (CO-credited), hours used, trailing-90-day burn and months left. Jobs under 40 funded hours have no labor estimate and show no percentage. |
| Run rate | Achievable hours per calendar month: the mean of complete months on record for that month, up to three years. |