Subcontractor performance scoring: build your bench

Ask any 30-year residential GC the most important asset they own, and they will tell you it is the rolodex. Not the projects, not the brand, not the truck — the bench of subs who will show up on time, do good work, and not nickel-and-dime the change orders. That bench is built over decades of trial-and-error and is mostly held in the GC's head as a set of intuitions about who is good for what.

The intuition is real but it is also lossy. A GC who books 30 framers across 30 projects might genuinely know which one is the most reliable. The same GC asked about their tile setters across 18 projects will be much fuzzier. Most GCs have 40-80 subs across all trades and cannot consistently rank them. Subcontractor performance scoring is the discipline of making that ranking explicit, tracked, and based on measurable inputs.

The four metrics that matter

Resist the urge to invent ten metrics. Four is enough, and using more dilutes signal. The four that matter for residential:

1. On-time start rate

Of the last five jobs you booked this sub, on how many of them did they actually show up the day they said they would? A sub who starts on the booked day on 4 of 5 jobs (80%) is dramatically better than one who starts on the booked day on 2 of 5 jobs (40%). The second sub is costing you 1-3 days of schedule per job in deferred downstream work.

Anonymized data across 18 GCs and 312 jobs in a recent informal survey showed:

• Top quartile framers: 91% on-time start.
• Bottom quartile framers: 41% on-time start.
• Top quartile painters: 87%.
• Bottom quartile painters: 38%.
• Top quartile drywallers: 88%.
• Bottom quartile drywallers: 32%.

The variance is enormous and it is consistent within a trade. A GC who systematically replaces their bottom-quartile sub in each trade gains 8-15 days of schedule per project. That is the single highest-ROI move on a residential bench.

2. Punch item count

The number of punch list items attributable to this sub at close-out, normalized by job size. A tile setter who consistently generates 14 punch items per 1,000 sq ft is doing different work than one who generates 4 per 1,000 sq ft. The latter saves the GC a full week of back-and-forth at close-out.

Be careful with this metric on small samples. A single job with bad weather or homeowner-driven scope changes can produce an unrepresentative punch count. Use a 3-5 job rolling average, not a single project.

3. Communication responsiveness

How long does it take this sub to respond to a substantive question? Not a casual text, but an RFI, a change order, a schedule confirmation. The metric is average time-to-response in business hours.

A sub who replies in 4 hours is a different operator than one who replies in 38 hours. The slower sub adds 1-2 days of cycle time to every back-and-forth on the job, and those add up — across 15 substantive communications, that is 15-30 hours of schedule slippage on a single trade.

4. Change order discipline

The ratio of unexpected post-completion charges to the original contract value. A sub who runs 0-2% over baseline on change orders is one you can budget around. A sub who runs 12-18% over baseline is one whose bids you cannot trust.

This is not the same as "the cheapest bidder." A sub who bids low but consistently runs 15% over is more expensive than a sub who bids 8% higher but lands at the bid. The first sub also generates the most administrative friction, the most homeowner conversations, and the most disputed line items.

The math on dropping your bottom quartile

Let's do the math on a realistic residential GC running 6 projects a year, each $450K hard cost. Assume the GC currently uses the bottom-quartile sub for one major trade per project, unknowingly — they just have not tracked it explicitly.

Per project cost of using a bottom-quartile sub instead of a top- quartile one:

• 10 days additional schedule × $1,800/day field overhead = $18,000.
• 14 extra punch items × $250 avg labor + materials = $3,500.
• 11% change-order cost over baseline on a $35K sub contract = $3,850.
• Late-response cycle costs (deferred decisions, etc.) = $2,500.

Total per-project cost of staying with the bottom-quartile sub: roughly $27,850. Across 6 projects per year, that is $167,000. The cost of identifying and replacing the bottom-quartile sub is zero, if you have the data. The cost of building the data is essentially zero if you score every job.

The math is not subtle. A residential GC who tracks four metrics across their bench and quietly drops their bottom 20% of subs every two years will outperform a peer who doesn't by $100K+ per year, with no other changes.

How to start scoring without overhead

You do not need a system to start this. You need a spreadsheet and 5 minutes per job at close-out. Four columns, plus a sub name column, plus a job ID. Fill it in immediately after every punch close, while memory is fresh.

After 3 jobs the numbers are noisy. After 5 jobs you start to see patterns. After 10 jobs you have actionable data, and you can make sub-replacement decisions with confidence. The discipline is the work — the tool is irrelevant, but the consistency isn't.

Conversations with subs about the data

The right thing to do with sub performance data is share it with your subs. A sub who knows they are tracked on these four dimensions, who can see their own scores, and who understands that improvement keeps them on your bench is a sub who improves. Most subs have never received feedback this specific — the conversation is welcomed, not resented.

The wrong thing to do is hold the data secretly and just drop subs who score poorly. They will never know what happened, and you will lose subs who could have improved with a 15-minute conversation. Lead with transparency.

How Buildra fits

Buildra automatically tracks the four metrics in the background as you run projects on the platform — on-time starts from the schedule, punch counts from the punch module, response times from the communication threads, change-order discipline from the change-order module. By your fifth project, the dashboard shows you a ranking of your bench within each trade. Sub conversations come pre-loaded with their own scorecard, so the discussion is based on data rather than impression.