Most ICM platforms answer one question: how much do we owe the rep?
It's a necessary question. But it's not the most important one.
The question that actually moves the number is different: what should that rep do today to generate the highest return — for themselves and for the organization?
That's Performance Optimization. And it cannot be calculated with a compensation system alone.
Why the calculation requires two data sources
A compensation system knows what it agreed to pay and what it already paid. It knows the plan rules, the periods, the quotas, the tiers. It can tell you how much a rep has accrued through yesterday.
A revenue system knows what was sold, to whom, when, at what margin, in what product. It can tell you the momentum of an account or the pace of a branch.
Separately, each system is useful. Together, they produce something neither can produce alone: the payout gradient of each rep.
The payout gradient answers a specific question: given this rep's current state — their attainment, their pipeline, their plan rules, the time left in the period — what is the next effort that pays them the most and generates the most for the organization?
It's not a prediction. It's not an estimate. It's a deterministic calculation that requires compensation data and revenue data to speak the same language, over the same period, from the same source of truth.
The five signals of the gradient
For each rep, the system ranks opportunities using five signals:
Boundary value — how close the rep is to the next commission tier. Crossing that threshold has a multiplier effect that can be worth more than closing a larger deal.
Effort-normalized value — the return adjusted for likely closing effort. An $80K opportunity that takes two months of work may pay less than three $20K deals closing this week.
Time feasibility — how much real time is left in the period and which opportunities can realistically close within that window.
Protect risk — what already-accrued commission is at risk of clawback if a prior sale reverses.
Component headroom — how much room is left in each plan component before accelerators or caps change the return on the next sale.
The system doesn't guess. If the data isn't sufficient to make the calculation with confidence, it shows nothing. The platform prefers silence over approximation.
The four surfaces
Performance Optimization is not a report. It's four ways to access the same calculation depending on who needs it:
Overview — for the administrator. See the attainment distribution across the entire team before publishing next period's plan. Make quota decisions with real data, not last year's outcomes.
Targets — for the administrator. Inline quota management with a full audit trail. Every adjustment documented before it impacts a rep.
Scenarios — for the administrator. Rehearsal mode. Simulate plan changes and see the projected impact on attainment distribution before applying anything. Nothing is saved until you decide.
Opportunity — for the rep and the manager. The rep sees their Three Moves: the three actions that pay the most right now, ranked by gradient. They see their earnings ring — what percentage of the period's potential they've already captured. They see the wayfinder heatmap — a heat map of their opportunities ordered by expected return. The manager sees the concentration risk of their team.
An activation condition
Performance Optimization requires the Compensation Agent and the Revenue Agent to be active simultaneously. Without revenue data, the gradient cannot be calculated. Without compensation data, there's no plan to calculate against.
It's the intelligence that emerges when the loop is closed.