Crew duration history and predictions

Completed stage logs are rolled into per-crew, per-stage, per-building-size averages that can then estimate how long a new job's stage will take, with a confidence figure.

What it is

An insights engine that groups completed stages by crew, stage type and square-footage bucket, storing average, median, min, max, sample count and average variance; then offers a days estimate for a new job with confidence scaled by sample size, and generates readable comparison and delay insights.

Also called: how long does this crew take · estimate days from history · crew averages by size

See it
Crew duration history and predictions
Apr
May
Jun
Jul
Ronan Petsch — Equipment Storage
Framing
Ivy Brubaker — 30×40 Garage
Concrete
Ivy Brubaker — 40×64 Shop
Trim
Gideon Alt — Hobby Shop
Framing
Marlin Hoover — Riding Arena
Concrete
Framing crewConcrete crewTrim crew
An insight card list: 'Crew B averages 4.2 days for Framing — 2,500-3,500 sqft buildings · 7 projects · Avg +0.6 days vs planned'. Sample data — no customer information appears here.
How it works
  1. 1Building type strings like '40x60 barn' are parsed into width, length and type, then bucketed under 1,500 / 1,500-2,500 / 2,500-3,500 / 3,500-4,500 / 4,500+ sqft.
  2. 2recomputeSnapshots groups completed logs and upserts a performance snapshot per crew+stage+bucket.
  3. 3estimateDays looks for an exact crew/stage/bucket snapshot, falling back to a sample-weighted average across buckets at 70% confidence.
  4. 4Confidence is sample count capped at five.
  5. 5generateInsights produces per-crew averages, crew-versus-crew comparisons per stage and size, and average day impact per delay reason.
Why we built it

Predicting how long a crew will take from the last job they did is useless when the last job was half the size of this one. Stage history is bucketed by building size before crews are compared, so this crew is slow on framing means slow on framing for a building of this size, not slow because the last one was bigger. From that history it produces averages, predicted durations and the correlations worth acting on, on the same principle the performance view uses: raw duration flatters whoever happens to get the small jobs.

The problem
  • Durations estimated from memory rather than from history.
  • Crew comparisons distorted by building size.
  • No confidence signal on a small sample.
Sound familiar?
What you get
Schedules built from what this crew has actually done on buildings this size.
A stated confidence so a two-job average is not treated as fact.
Crew-versus-crew comparison at the stage level.

See it on your own jobs

Twenty minutes, your numbers, no slide deck. We’ll build one of your real buildings in front of you and send you the estimate link at the end — yours to keep either way.

or keep browsing features →