Parts 2 through 5 cleaned up the inputs: an honest baseline, budgets you can trust, production hours, and right-sized crews. Each one closes a way the program could pay for the wrong thing.

The ledger test checks whether it all holds together. Take last year's jobs, run them through this year's rules, and read the ledger the program would have written.

It's the cheapest mistake you'll ever catch. Every problem it finds is one you didn't pay for.

⚠️
Base pay isn't part of the test

The ledger test models the performance incentive only. Base hourly wages were paid in full for every hour worked last year, and they will be under the program too. The test asks one question: what would the incentive on top of that wage have paid, and was it earned?

How the test works

You need four inputs, and you already have most of them:

Run every job through the rules. Each one either pays or it doesn't. Add it up and you have a year of payouts that never happened — which is exactly what makes them safe to read.

Five things the ledger should tell you

A total payout number on its own tells you very little. These are the five readings we look at, and what a bad one usually traces back to.

CheckWhat a pass looks likeA fail usually means
Payout vs. real savingsPayouts are a share of hours actually savedSlack in the baseline (Part 2)
Share of jobs that payReachable on most jobs, automatic on noneTarget set too loose or too tight
Paid on money-losing jobsClose to zeroThe budget was wrong, not the crew (Part 3)
Spread across crewsFollows performance, not routesTravel or crew size leaking in (Part 4, Part 5)
Paid on callback jobsHeld back or pulledQuality standard isn't wired in

Two readings matter most. Money paid on jobs that lost money means the program would reward bad estimates. And payouts piling up on a few crews usually means the benchmark is measuring their routes, not their work.

What a failed first run looks like

A composite from our readiness work: a residential services company with nine crews ran the ledger test on its first draft of the rules.

Ledger reading (12 months)Draft rulesRevised rules
Jobs that would have paid78%54%
Total modeled incentive$212,000$118,000
Paid on money-losing jobs$41,000$3,000
Share to top 2 of 9 crews46%24%
Paid on jobs with callbacks$17,000$0 — held

The first run looked generous, and that was the problem. Nearly four in five jobs paid, and two crews took almost half the money.

Both of those crews ran the tightest in-town routes. And $41,000 went to jobs that lost money, because their budgets had been padded for years.

They didn't launch. Budgets were reset on the worst job types, the benchmark moved to production hours, crew sizes were normalized, and callback holdbacks were added.

The second run paid less, and every dollar traced to real savings. When crews saw the modeled numbers, the scoreboard matched what they'd seen on the job.

"If you can't explain every dollar in last year's ledger, you're not ready to write this year's checks."

Don't fix a bad ledger by cutting the split

When a first run comes back too expensive, the tempting fix is to shrink the crew's share of the bonus pool. The total drops, and the problem stays.

The money still flows to the wrong jobs and the wrong crews — just smaller amounts. Trace each bad reading back to the input it came from, fix that input, and run it again.

When to run it again

Run it before launch, and again after any rule change, pricing change, or large new contract. Once you're live, compare real payouts against what the ledger predicted.

A gap between modeled and real payouts is an early warning. It usually means something in the inputs has shifted.

Where Protiv fits

The ledger test is the last step of the readiness work our team runs with you, using the job and time data you already sync. Today it's a team-led process with growing platform support, and more of it moves into the product on a regular basis.

Once you're live, the safeguards you modeled run on every job: quality standards, penalties, and a held-bonus ledger that keeps payouts back until callback windows clear.

The whole series in six questions

Answer all six before launch, and the incentive pays for performance — not geography, headcount, or a budget that was wrong from the start.

  1. Why do most programs fail?
  2. Is your baseline honest?
  3. Wrong budget, or crew behind?
  4. Which hours are you measuring?
  5. How many people did the work?
  6. What would the program have paid?

Frequently asked questions

Is this a piece-rate pay model?

No. The base hourly wage is always paid in full for every hour worked, regardless of job outcome. The performance incentive is a bonus layered on top, earned only when a crew hits a quality standard and a schedule benchmark built from real job data. The ledger test models that incentive only — never base pay.

How much history do we need?

Twelve months covers a full season, which matters for most trades. With less, run the test on what you have and flag the months you're missing, since seasonal swings can hide there.

What if the test shows we'd have paid too much?

That's what the test is for. Trace the excess to its source — a padded budget, blended hours, an extra crew member — fix that input, and rerun. Cutting the crew's share hides the problem without fixing it.

What if it shows we'd have paid almost nothing?

A target nobody reaches won't change behavior, and crews will tune it out. Very low payouts usually point to budgets that are too tight or a benchmark built on hours the crew doesn't control.

Should crews see the ledger results?

That's your call, but it helps. Showing crews what a typical job would have paid, and why, is one of the clearest ways to build trust before launch.

Ready to run the ledger test on your own jobs?

A 30-minute conversation with our team walks through what your program would have paid last year — before it pays anything this year.

Schedule a demo →