Commission Accelerators: How to Design Them So They Drive Behaviour, Not Just Cost

TL;DR: UK sales teams using flat 1.5x commission accelerators above quota overpay windfall deals by an average of 40% compared to stepped structures, according to analysis of 50+ commission plans. The flat multiplier rewards luck and genuine outperformance identically, creating a £27,000 cost difference on a £60k OTE rep hitting 200% attainment. This article compares three structures (cliff, stepped, continuous) on the same £60k OTE rep, shows the cost-of-overperformance maths including 2026-27 employer NICs, and argues that a capped plan with a decelerator above 200% often outperforms uncapped-commission dogma. The continuous accelerator eliminates gaming surfaces; the decelerator protects against quota-setting errors and fat-tailed deal distributions without demotivating top performers.

Most UK accelerators are designed in about ten minutes. Someone reads a US blog post, picks 1.5x, bolts it on above 100% of quota, and ships the plan. Six months later finance is asking why over-quota payroll is running 40% hotter than the model predicted, and the top rep is sitting on a deal they could have closed in Q3 but pushed into Q4 because the calendar reset their attainment.

Flat 1.5x accelerators are economically incoherent: they pay the same marginal rate for the first £60k of over-quota revenue as for the last £600k, making no distinction between a windfall deal that any rep would have closed and sustained outperformance that required genuine skill. At 200% attainment (£1.2M revenue), the cliff structure pays £45,000 variable—a 150% increase in commission for 100% increase in revenue, creating a 2.5x cost multiplier on marginal revenue. Every £1 of revenue above £600k costs the business 4.5p in commission plus 0.675p in employer NICs, versus 3p at-quota cost. When one outlier deal can trigger this rate, the plan becomes expensive insurance against the wrong risk.

What follows is the maths on three accelerator structures—cliff, stepped, and continuous—using the same £60k OTE rep as a test bench, plus the cost-of-overperformance numbers your CFO will eventually ask about, and a contrarian take on why capping sales commission with a decelerator above 200% is often smarter than the uncapped-commission dogma.

The test bench: one rep, three plans

A typical UK SaaS sales rep on £60,000 OTE (70/30 split) has £42,000 base salary and £18,000 variable pay at 100% quota attainment. With a £600,000 annual revenue quota, this implies a 3% commission rate at target. Every £1 of variable pay above this baseline attracts 15% employer NICs above the £5,000 threshold, making the true cost of over-quota performance 115% of the stated commission amount.

Tax/NICs context: commission is treated as Class 1 earnings, so every £1 of variable pay attracts PAYE, employee NICs, and—the line that gets ignored in plan-design meetings—employer NICs at 15% above the £5,000 secondary threshold for 2026-27. HMRC's own internal manual on commission is explicit that commission counts as earnings for Class 1 purposes.

We'll compare three over-quota structures, each calibrated so that on-plan reps land at the same £18,000 variable. The only thing that changes is what happens once they cross 100%.

Structure 1: the cliff (flat 1.5x above quota)

The flat 1.5x accelerator pays the same marginal rate (4.5%) to a rep at 110% attainment as to one at 200%, economically rewarding the first £60k of over-quota revenue identically to the last £600k—a design that cannot distinguish windfall deals from sustained outperformance. Every pound of revenue above quota pays at 1.5x the target rate (so 4.5% instead of 3%).

Attainment	Revenue closed	Variable pay
100%	£600,000	£18,000
110%	£660,000	£20,700
130%	£780,000	£26,100
150%	£900,000	£31,500
200%	£1,200,000	£45,000

What the cliff actually rewards: getting just over quota. The 110% rep gets the same marginal rate as the 200% rep. That's economically incoherent—the 200% rep either had a windfall (in which case you've massively overpaid for luck) or is genuinely exceptional (in which case 1.5x undersells them).

The cliff also creates a behavioural problem you can predict from the maths: the jump in marginal rate at exactly 100% gives reps an incentive to delay closing the deal that pushes them past quota until the period boundary. Our clients have reported this pattern consistently—particularly in Q4-to-Q1 transitions where annual quotas reset. We'll come back to this.

Structure 2: stepped (the kicker ladder)

A stepped accelerator pays different marginal rates in different bands above target. A common UK design:

• 100–120%: 1.25x (3.75%)
• 120–150%: 1.5x (4.5%)
• 150%+: 2x (6%)

Same rep, same quota:

Attainment	Variable pay	vs cliff
110%	£20,250	−£450
130%	£25,200	−£900
150%	£30,600	−£900
175%	£39,600	+£1,350
200%	£48,600	+£3,600

The stepped structure says something the cliff doesn't: we pay more for harder-won attainment. The first 20% above quota is the easiest—pipeline that was already going to close, perhaps slipped from last quarter—so we don't pay a huge premium for it. The third 30-point band, where reps are genuinely outperforming, gets the 2x kicker.

The trap with stepped plans: the band thresholds become quota-of-their-own. A rep at 119% sees a discontinuity at 120% and starts gaming timing. The more bands you add, the more gaming surfaces you create. Two bands is usually enough; four is plan-design self-harm.

Structure 3: continuous (the linear ramp)

A continuous accelerator increases the marginal multiplier smoothly with attainment. A clean formulation: marginal multiplier = 1 + (attainment − 100%) / 50%, so at 100% you earn at 1x, at 150% you earn at 2x, and the rate moves linearly in between. Cap the multiplier at 2x above 150% to keep the model bounded.

Attainment	Variable pay	vs cliff
110%	£19,980	−£720
130%	£25,020	−£1,080
150%	£31,500	£0
175%	£40,500	+£2,250
200%	£49,500	+£4,500

Notice what the continuous structure does: it pays less than the cliff for marginal overperformance (110%, 130%) and more for substantial overperformance (175%, 200%). At 110% attainment, the continuous accelerator pays £19,980 versus £20,700 for the cliff—£720 less for marginal overperformance—but at 200% attainment it pays £49,500 versus £45,000 for the cliff, rewarding genuine outperformance £4,500 more generously. There are no discontinuities, no band edges, nothing to game. A rep at 119% has exactly the same incentive to close one more deal as a rep at 121%.

This is the design most spreadsheet plans can't handle—the formula gets ugly and the IF statements stack up—but it's the one with the cleanest behavioural properties. If your tooling supports it, continuous accelerators are usually the right answer for inside sales teams with reasonably tight deal-size distributions.

The line your CFO will eventually draw: cost-of-overperformance

Here is the number that gets left out of every accelerator design conversation we see. Take the cliff structure at 150% attainment:

• On-plan variable: £18,000
• Over-quota commission: £13,500
• Employer NICs at 15% on the over-quota portion: £2,025
• True employer cost of that £13,500: £15,525

If you run the Apprenticeship Levy you can add another 0.5% on top. Pension auto-enrolment contributions, where commission is pensionable, add more. The £13,500 line item in the comp model is closer to £16,000 once it actually hits the P&L.

Now scale that across a team. A 20-rep org where the top quartile is consistently hitting 140%+ on a flat 1.5x cliff is bleeding tens of thousands of pounds of employer NICs into ground that didn't have to be that expensive. The stepped and continuous structures don't eliminate this—over-quota pay is the point—but they let you choose where the money goes. Pay a smaller premium for the easy 110% attainment and a larger one for the genuine 180% performance, and the same total comp budget gets you better behaviour.

For a fuller treatment of how the 15% rate plays through commission specifically, see our breakdown of the employer NIC change on commission.

The case against uncapped (the orthodoxy is wrong here)

The SaaS-blog consensus says