Why Your Best Reps Keep a Private Commission Spreadsheet
Shadow accounting is the most reliable signal in a sales organization. Here is what it costs you, why commission math is so easy to get wrong, and the three-question test that tells you whether your process is trustworthy.
Ask a room of salespeople whether they check their own commission by hand. Nobody raises a hand. Then one guy admits it, and suddenly half the room does.
They've all got a spreadsheet. Their own. Every closed deal, their read of the comp plan, and a running total they compare against whatever hits on payday. Some of them have kept it for four years.
There's a name for this. It's called shadow accounting, and it's the most reliable signal in a sales organization.
It doesn't mean your reps think you're stealing. It means they can't reproduce your math. A number you can't reproduce is a number you have to take on faith, and nobody takes their own pay on faith.
"Why is my commission different from what I calculated?"
That message shows up around the fifth of the month, and it's always some version of the same four questions.
Why is my number different from what I calculated?
Was the Henderson job in there?
Why'd I get paid on that sale but not this one?
Where did this adjustment come from?
Every one of those is fair. Every one of them eats an hour of a manager's day, usually spent reopening the same spreadsheet and tracing backwards. And while that's happening, your best closer has stopped selling and started auditing.
That's the real cost, and it never shows up on a line item. Your highest-paid people spend the first week of every month doing accounting work, badly, on data they only half have.
Why is commission math so hard to get right?
Because a commission isn't a number. It's a derivation, and there are a dozen legitimate ways it can land.
Take one $10,000 roof.
At 10% of revenue, the rep makes $1,000. At 10% of gross profit on a job with $4,000 of cost, the same rep makes $600. Split 30/70 between a setter and a closer, that $600 becomes $180 and $420. If the rep crossed a tier that month and the plan pays retroactively, the whole period re-rates upward. If the customer put a third down and the plan has a deposit kicker, add two points. If the customer got 20% off and the plan penalizes discounting, take some back. If a regional manager earns an override, the job costs you more than the rep ever sees. If it cancels in March, part of it comes back.
Same roof. Nine different right answers.
Now put that on a statement that says Commission: $843.27 and nothing else.
That's not transparency. That's an assertion. Your rep has no way to agree or disagree with it except by feel, and feel is exactly what produces a private spreadsheet.
Is it really that easy to get wrong?
Yes, and somebody measured it. Raymond Panko at the University of Hawaii has spent decades auditing real business spreadsheets. Across field audits covering more than a hundred of them, roughly 88% contained errors.
That's not a competence problem. It matches human error rates on any complex logical task — people make undetected mistakes at a few percent per action, in every field, always. A spreadsheet has no mechanism to catch them.
The trust damage got measured too. In a 2026 survey of 1,003 operations professionals run by DOSS, an operations software company surveying its own market, 35% of spreadsheet incidents cost somebody trust, credibility, or job performance.
And corrections are normal, not scandalous. An EY survey commissioned by the payroll company Paycom found the average organization makes 15 corrections per pay period.
Fifteen fixes a period is a functioning business. Fifteen fixes a period with no record of what changed or why is how you end up with a sales floor keeping private ledgers.
What should a commission statement actually show?
Not more numbers. Dumping a 400-row export on a rep isn't transparency, it's homework.
What ends shadow accounting is traceability. A rep should be able to walk a payment backwards to the sales that produced it and the rules that got applied, without asking a human.
Watch what that does to the conversation. Right now you get:
"I think I'm missing money."
You can't answer that. There's nothing in it to check. You'll spend two hours proving a negative and he'll walk out half-convinced anyway.
With a traceable statement you get:
"This sale got credited to me, but the deposit kicker didn't fire."
That's a great problem. It's specific. You can check it in ninety seconds. It resolves the same way for everybody. And you're both looking at the same document.
Transparency doesn't end disagreements. It turns them from arguments about trust into questions about configuration. That trade is worth almost anything.
What happens if you find an error after payroll goes out?
This is the one that does the most damage, and almost nobody plans for it.
The month closes. The money goes out. Somebody finds a mistake.
The obvious move is to open last month's file and fix the formula. Thirty seconds. And now you've got two versions of last month with no way to tell them apart. The file you sent to payroll says one thing. The sheet says another. Six weeks later somebody asks why a rep got $4,180 when the sheet reads $4,410, and nobody can answer — not because it's hidden, but because it got written over.
That's how commission records stop being evidence. The record was editable, so it got edited, and now it describes a month that never happened.
The rule that fixes it is one sentence. A closed month never changes, and corrections land in the next open month with a reason attached.
You end up with two records that are both true. What you paid in July. What you corrected in August, and why. Neither one has to lie to protect the other.
This matters most in the trades, because the facts arrive after the money does. Sale in June. Install in July. Final material invoice in August. Financing clears in September, or it doesn't. If you pay on gross profit, the profit is genuinely unknown the day you pay it. Corrections aren't a failure mode in that world. They're the job.
How do you know if your process is trustworthy?
Run this tonight. Pick a month you closed at least a quarter ago, and answer three questions.
One. Can you produce the exact figure you paid each person, as of the day you paid it — not as of today?
Two. Can you produce every correction made to that month afterward, each one with a reason and a name attached?
Three. Can a rep see both of those without asking you?
Three yeses and your records are evidence. Any no and you're running on memory. Memory isn't a control. It just feels like one until the first month somebody disagrees with you in writing.
Does KickSplit run payroll?
No, and that's on purpose. We do the math and hand your payroll company a file. The second we touch the money we're a payroll company, and you already have one.
What we do is the part that actually breaks.
The plan engine handles the shapes that kill spreadsheets. Percentage of revenue or gross profit, margin bands, flat per sale, trailing residuals. Tier ladders that pay retroactively or progressively. Deposit and deal-size kickers, product rates, discount penalties, caps, floors, draws, clawbacks, splits, manager overrides three different ways, new-hire ramps. You configure all of it yourself. No engineer.
Lock a period and it's sealed. Nobody edits it after that. Not an admin, not you, not somebody with database access. Corrections go forward into the next open period as their own line, and the app won't save one without a written reason. The original calculation stays put. Both numbers show up on the statement and in the export.
Reps see their own statement itemized, with the sales behind every line. If one looks wrong, they dispute that line — not the whole check. And an open dispute blocks the month from closing. Not "shouldn't close." Can't. The button doesn't work.
On our top two tiers, a won deal never gets re-keyed. It lands in a review queue with a preview of exactly what would be created. An admin approves it. Then it's a commissionable sale carrying its own cost, discount, product type, territory and deposit percentage.
Two more things, before you have to ask. Resolving a dispute doesn't move money by itself — the block is the valuable part, and the actual correction is a separate, deliberate action. And clawback stays off until you switch it on. Nothing is reversing commission behind your back.
The bottom line
Shadow accounting isn't an attitude problem and it isn't a personnel problem. It's a rational response to a number nobody can reproduce.
Make the math traceable. Make the closed month permanent. The private spreadsheet has nothing left to check.
KickSplit is a CRM and commission platform for commission-driven sales teams — lead to sale to a payroll-ready export, in one place. See how it works.