Right 99% of the time, which is wrong for someone every month
Bazoom’s sales teams close a lot of deals, fast, across two brands. At that volume a spreadsheet that is right 99% of the time still gets somebody’s commission wrong every single month. And that somebody finds out at payroll, the worst possible moment.
Each error cost more trust than money. A rep who has been paid wrong once checks everything twice afterwards, and stops believing the plan is on their side.
“One wrong payslip costs you a quarter of trust with that rep. We had one most months.”
Mathias Sjøln Madsen, CFO, Heymate & Bazoom
But more checking by finance only moved the problem
The first response was more control: a second reviewer, a longer close. Errors still got through, because finance was checking its own numbers against its own assumptions. The CFO’s real problem was not the calculation. It was that there was no moment, before money moved, where the rep and finance had looked at the same number and agreed.
Without that moment, every payout was finance’s word against the rep’s memory of the quarter.
So Bazoom gave every payout a moment of agreement before payroll
Prowi added the moment. At period end each rep sees their statement in the app, deal by deal, and confirms it. Anything that looks off is raised there, on the deal, while it is a two-minute fix. Then finance approves the lot and exports the payroll file.
It sounds procedural. In practice it changed the tone. Commission stopped being a monthly negotiation and became something everyone had already agreed on.
“The monthly approval flow makes commission transparent, and settled correctly, every time.”
Mathias Sjøln Madsen, CFO
Which is why the month now closes in an afternoon, for both brands
Heymate and Bazoom have different plans and different rhythms. Finance runs both through the same approval flow and the same export, and the close no longer stretches into the next month.
The number the CFO quotes is the one that stopped happening: corrections after payroll, over the last year. None.


