It is a sequence, not a percentage
Most brokerages describe compensation as a split — "our agents are on 80/20". That describes one step of a calculation with at least four, and the other three are where the disagreements come from.
The sequence, in the order it has to be applied:
1. Gross. The finder fee on the funded amount, plus any promotional or volume basis points that apply to this deal.
2. Off the top. A network or head-office fee, where the brokerage rolls up to a franchisor or a larger entity. This can be a percentage of gross, a flat amount, or both — and critically, it comes out before anything is split.
3. The split. The agent's percentage applied to what remains. Or their post-cap percentage, if they have already crossed their company-dollar cap this period.
4. Off the bottom. Transaction fee, desk fee, underwriting fee, E&O — applied after the split, to the agent's share.
What is left is agent net.
Why the order matters so much
Take a fee of a few hundred dollars on a deal. Applied before the split, the brokerage and the agent share it in proportion. Applied after, the agent bears all of it.
Same fee, same deal, same split percentage, materially different cheque. Multiply that across a year and you have the source of most commission disputes — not disagreement about the rate, but about the sequence.
This is the error spreadsheets make most reliably, because a spreadsheet has no opinion about order. Whoever built it encoded the sequence they had in mind that day, and it is rarely written down anywhere else.
Where manual calculation breaks down
Company-dollar caps
A cap means the agent's split improves once they have contributed a set amount to the brokerage in a period. The complication is that the cap is crossed mid-deal, not mid-month. The deal that crosses it is split partly at one rate and partly at another, or the whole deal moves to the new rate depending on your policy — and your policy has to actually exist, in writing, because the arithmetic differs.
After the cap, both the split and often the transaction fee change. If that switch is being applied by hand, it is being applied late.
Graduated tiers
A tier ladder steps the agent up as they reach thresholds. Three questions decide the arithmetic, and brokerages genuinely differ on all three:
- Is the tier measured on transaction count or funded volume? Some ladders mix both, so an agent moves up on whichever they reach first.
- Does a new tier apply only to subsequent deals, or retroactively to everything in the period?
- When does the change take effect — immediately, at the start of the next period, or on the agent's anniversary?
That last one is the qualification period, and it is the difference between "a good month" and "last quarter's payouts get rewritten".
More than one person on a deal
A lead agent, an associate, an assistant and a referrer can all take a share of the same file. Each allocation is defensible on its own; the failure mode is that they sum to more than the deal generated, and nobody notices until the payouts are added together at month end.
An automated calculation can refuse to over-distribute a file. A spreadsheet cannot, because each row looks fine in isolation.
Clawbacks
A funded deal that later unwinds has to be recovered. Recovering it as a single deduction from one cheque is technically correct and practically a problem. Recovering it across future cycles, with the reason attached and visible on the statement, is the version that does not cost you an agent.
The part that matters most
An agent who cannot see how their number was reached will not trust it, and the conversation you end up having is about whether the spreadsheet is right rather than about a specific line.
A statement that shows the arithmetic — gross, what came off the top, the split applied, what came off the bottom — turns a dispute into a five-minute check of one figure. That is worth more than the automation itself, because it changes what the disagreement is about.
Making it survive scale
Ten agents on one structure can be run by hand by someone careful. Thirty agents on four structures, with two on caps, one mid-tier and a referral arrangement between two of them, cannot — not reliably, and not by one person who is also doing other work.
The threshold is not brokerage size, it is structure count. The moment agents are on genuinely different arrangements, the calculation has to come from the funded file rather than from a monthly rebuild.
Mortgage360 calculates all of the above from the deal itself — multi-participant allocations with an over-allocation guard, mixed-metric tier ladders, qualification periods, caps with post-cap splits and fees, referral arrangements, and clawbacks recovered across cycles. The commissions page has the detail, and commission payroll covers how it reaches the people being paid.
For the wider question of how brokerages set these structures in the first place, see mortgage brokerage commission splits in Canada.