How do you design a sales compensation plan that drives the right behavior?

What a comp plan actually is, and why it drives behavior faster than any playbook

A sales compensation plan is a contract that tells a rep, in dollars, what the company wants them to do. Everything else — the enablement deck, the QBR scorecard, the manager's coaching cadence — is advice. The comp plan is the only document with a payout attached, so when advice and comp disagree, comp wins every time. That is the whole reason plan design belongs in a RevOps function rather than in HR or finance alone: it is a behavior-routing system disguised as a payroll artifact.

The architecture is consistent across most B2B software and services organizations. Five pieces do the work. On-target earnings (OTE) is the total a rep earns at 100% attainment. The base/variable split decides how much of that OTE is guaranteed. The quota multiplier sets how much revenue a rep must produce relative to what they cost. The accelerator curve decides what happens on either side of the 100% line. And the crediting and clawback policy decides when money is truly earned. Change any one of these and the sales floor's behavior shifts within a quarter, usually faster than leadership expects.

The base/variable split is the clearest expression of causality. An account executive who personally sources, runs, and closes a deal has near-direct influence over the outcome, so a 50/50 split is the common convention — half the pay is at risk because half the result is genuinely attributable. A sales development rep, whose job ends when an opportunity is accepted, typically sits closer to 60/40 base-heavy, because the eventual close is out of their hands. A customer success manager owning renewals and expansion usually runs 70/30, since renewal outcomes are heavily shaped by product quality, support experience, and buying-committee turnover that no CSM controls. A RevOps manager, one full degree removed from the revenue event, tends to sit at 80/20 or 85/15 with an MBO-style bonus rather than a transactional commission. The pattern is not arbitrary: the more of the outcome the person genuinely causes, the more of their pay you can responsibly put at risk.

Quota multiplier is the unit-economics guardrail. Expressed as quota divided by OTE, it answers "how many dollars of production does this seat need to generate to be worth its fully loaded cost?" For quota-carrying AEs, a band of roughly 4–6x OTE is the durable convention. Below 4x, commission plus benefits plus management overhead eats too much of the gross margin on each deal and the plan structurally loses money. Above 6x, the number stops reading as a stretch goal and starts reading as a fiction — reps disengage, sandbag into the next period, or leave. SDR quotas are usually expressed in pipeline rather than closed revenue, and CSM quotas in a book-of-business multiple rather than new logo.

The behavioral consequence of getting the split and multiplier right is that the plan becomes self-enforcing. Reps read it, understand what it rewards, and go do that. The consequence of getting it wrong is that reps also read it, understand what it rewards, and go do that — which is precisely the problem when the plan rewards the wrong thing.

The step-by-step design process, from role model to signed plan

Comp design is a sequence, not a negotiation. Skip a step and you will discover the omission in January, when it is expensive to fix.

Step one: define the revenue model before the pay model. You cannot design compensation for a motion you have not described. Write down the segments, the average contract value in each, the average cycle length, the win rate, and the ratio of new logo to expansion revenue. A team where 70% of revenue comes from expansion needs a fundamentally different plan than a team where 90% is new logo, even if both sell the same product.

Step two: assign each role a causal metric. One primary metric per role. AEs get closed-won new ARR. SDRs get sales-qualified opportunities accepted by the AE team — accepted, not merely booked, because acceptance is the quality filter that stops meeting-stuffing. CSMs get gross retention plus expansion, often as two components with separate rates. Solutions engineers, where they carry variable pay, are usually tied to the AE team's attainment rather than to individual deals, because splitting a deal between two owners creates arguments that outlast the deal.

Step three: set OTE from market data, then derive quota from OTE. Do it in that order. Anchoring quota first and backing into OTE produces numbers that look fine in a spreadsheet and are uncompetitive in a hiring loop. Once OTE is set, apply the multiplier band for the segment: enterprise seats with long cycles usually sit at the lower end of 4–5x, SMB and transactional seats at the higher 5–6x end, because velocity compensates for smaller deals.

Step four: build the payout curve. Establish the base commission rate as target variable divided by quota. Then define the tiers. Below roughly 50% attainment, a decelerator — half rate or, in aggressive plans, zero — prevents the plan from funding chronic underperformance. Between 50% and 100%, the base rate holds flat. Above 100%, accelerators kick in, commonly in the 1.5x–3x range, with an additional kicker for exceptional overperformance. The point of the curve is that the marginal dollar of effort is worth more at 105% attainment than at 65%.

Step five: write the crediting rules before anyone asks. Split credit on co-sold deals, treatment of multi-year contracts (annualized versus total contract value), what happens to a deal when a rep leaves mid-cycle, ramp draws for new hires, and the clawback window if a customer churns early. Every one of these will come up. Deciding them in advance in a document is a governance exercise; deciding them in the moment is a fight.

Step six: model the plan against last year's actual results. Take the prior period's real attainment distribution and run it through the new plan. What would you have paid? What would your top rep have earned? What would your median rep have earned? If the model produces a commission expense that leadership will not approve, fix it now, not after the plans are signed.

Step seven: roll it out with time to absorb it. Announce 30–60 days before the period starts. Manager one-on-ones, not just an all-hands. Then hold open office hours for the first several weeks of the new period, because the questions that surface in week two are the ones that would otherwise become disputes in month four.

Costs, timelines, and the ranges that keep the math honest

Every comp decision has a number attached, and the numbers constrain each other. Understanding the ranges is what separates a plan that survives a board review from one that gets rewritten in March.

Commission cost as a percentage of revenue is the top-line constraint. Mature organizations with efficient motions generally run lower than growth-stage organizations buying market share, and a plan that drifts materially above the range leadership signed up for is a signal that either quotas are too low, pricing is too soft, or discounting is unmanaged. Track it monthly rather than annually; by the time an annual number is wrong, three quarters of payouts have already gone out the door.

Ramp is the most commonly underfunded line item. A new AE in an enterprise motion with a six-to-nine month cycle cannot produce closed revenue in their first quarter, and pretending otherwise just means their first plan is a guaranteed miss. The standard fix is a ramped quota — a fraction of full quota in month one through three, more in months four through six, full quota thereafter — combined with a non-recoverable draw so the rep has income while the pipeline builds. A recoverable draw, where the company claws the money back out of future commissions, is cheaper on paper and much more expensive in attrition. New reps who spend their second year paying off their first year rarely stay for a third.

Timeline for a comp cycle runs roughly a quarter. Design and modeling in the first month, finance and leadership approval in the second, rollout and manager conversations in the third, launch on day one of the new period. Compressing this into three weeks in late December is a recurring organizational habit and a reliably bad one; it produces plans nobody had time to stress-test and rollouts where managers are learning the plan alongside their reps.

Payout cadence matters more to behavior than most leaders assume. Monthly payouts on closed-won deals keep the feedback loop tight and the plan psychologically real. Quarterly payouts smooth administrative burden and align with bookings recognition but weaken the connection between the action and the reward. Many organizations split the difference — monthly commission on closed deals, quarterly on attainment-based accelerators once the period's attainment is actually knowable.

Clawback windows typically run 90 to 180 days for early churn, and their purpose is behavioral rather than financial. The dollars recovered are usually modest. What the window actually does is stop a rep from closing a customer they know is a bad fit, because the commission does not vest until the customer has stayed long enough to prove the fit was real. Set the window to roughly match the point at which onboarding failure becomes visible.

Tooling is a real line item once the team passes a threshold. Spreadsheet-based commission calculation is survivable for a handful of reps and becomes a liability somewhere in the low double digits, at which point the failure mode is not cost but trust: a miscalculated commission statement damages the plan's credibility more than a slightly-below-market OTE does. Dedicated incentive compensation platforms — CaptivateIQ, Spiff, Xactly, QuotaPath among the established names — automate calculation, give reps real-time visibility into earnings, and produce an audit trail. Pricing is generally per-payee per-month and scales with team size and plan complexity; get a quote rather than assuming, since the range across vendors is wide.

Dispute rate is the underrated cost. Every commission dispute consumes a manager's time, a RevOps analyst's time, and a chunk of the rep's selling week. Plans with fewer moving parts generate fewer disputes, which is one of the practical arguments for simplicity that has nothing to do with motivation theory.

Where teams get it wrong, and why the failures repeat

The failure modes in comp design are remarkably consistent across companies, industries, and decades. They repeat because each one is locally rational and globally destructive.

Capping commissions. A cap is a message to the top decile of the sales team — the people who typically produce a disproportionate share of revenue — that the company would rather limit their upside than pay for outsized results. The stated rationale is always cost control. The actual effect is that the best rep stops selling once the cap is in sight, parks the remaining pipeline in the next period, and starts taking recruiter calls. If the concern is that a rep will earn "too much" on an anomalous whale deal, the correct instrument is a windfall clause negotiated in advance for deals above a defined size, not a blanket ceiling that punishes consistent overperformance.

Paying on activities instead of outcomes. Put dollars on calls made or demos booked and reps will produce calls and demos — of declining quality, precisely calibrated to clear the threshold. Activity metrics are legitimate coaching inputs and legitimate leading indicators for forecasting. They are terrible payout metrics because they are trivially gameable and they measure motion rather than result. The partial exception is the SDR role, where the outcome genuinely is an opportunity handed off; even there, pay on opportunities *accepted* by the receiving AE, so the AE team functions as the quality gate.

Changing the plan mid-period without grandfathering. This is the most reliable way to trigger resignations. A rep who prospected an account in February under one set of rules, and closes it in August under a worse set, does not experience that as a policy update. They experience it as money taken from them. Even when leadership has a genuine reason to change — a loophole is being exploited, the market moved, a product line was discontinued — the pipeline in flight must be honored under the rules that were in force when the work started. The trust cost of skipping grandfathering consistently exceeds the dollar cost of honoring it.

Too many variables. Multi-component plans that weight new logo, expansion, multi-product attach, margin, and a strategic MBO all at once look sophisticated in the deck and produce paralysis on the floor. When a rep cannot tell you what they earn on a given deal without opening a spreadsheet, the plan has stopped functioning as a behavioral signal and become an accounting artifact. Two or three components is a workable ceiling for most quota-carrying roles.

Comping CSMs on renewals they did not influence. If a CSM inherits a book and a renewal lands in their second week, paying full commission on it rewards presence rather than work — and sets a precedent that is hard to unwind. A vesting window of roughly a quarter before renewal credit begins, with full credit once the CSM has genuinely owned the account, keeps the incentive pointed at expansion and save work rather than at calendar luck.

Ignoring team dynamics in account-based motions. In enterprise selling, where a pod of AE, SE, SDR, and CSM works one account together, a purely individual plan creates hoarding: reps protect accounts instead of pulling in the colleague who would actually advance the deal. Tying a modest slice of variable pay — commonly in the range of ten to twenty percent — to pod or team attainment restores the collaboration incentive without diluting individual accountability into a group bonus nobody feels.

Designing comp in isolation from territory design. This is the failure that hides best. A perfectly balanced plan applied to wildly unbalanced territories produces the appearance of a rep quality problem. If one rep's patch contains three times the addressable accounts of another's, no accelerator curve will make the resulting attainment distribution meaningful. Territory design, quota allocation, and comp design are one project, not three, and they belong to the same owner.

Not measuring the plan in flight. Annual review is a governance cadence, not a monitoring cadence. Watch attainment distribution monthly: what share of reps are above 80%, above 100%, above 120%. A healthy plan produces a broad middle with a meaningful over-performing tail. If almost nobody clears 80%, the quotas are wrong or the territories are, and reps will conclude the plan is theater. If nearly everyone clears 120%, the quotas are too soft and finance will notice before you do. Also watch voluntary attrition among top performers in the two quarters after any plan change — that is the clearest signal that something in the redesign broke trust.

A decision framework: matching plan shape to motion, cycle, and stage

There is no single correct plan, only a plan correctly matched to a motion. A few decision axes do most of the work.

Cycle length drives the split. Short transactional cycles support a more aggressive variable share, because a rep sees the payout loop close within weeks and the income is genuinely controllable. Long enterprise cycles argue for a more base-heavy split, because a rep who cannot close anything for two quarters still has rent to pay, and a plan that ignores that fact selects for reps who can afford to gamble rather than for reps who can sell. Applying one split across a team that contains both motions is the most common segmentation error.

Revenue mix drives the metric. If growth comes primarily from new logos, weight the AE plan toward new ARR and keep expansion in the CSM plan. If growth comes primarily from the installed base, the CSM plan needs real variable weight and real expansion targets, not a token retention bonus. In product-led motions where self-serve accounts convert and then get handed to a sales-assisted team, decide explicitly whether the rep gets full credit for revenue the product generated on its own — and be honest about the answer, because reps will figure out the true rule regardless of what the document says.

Company stage drives tolerance for complexity. Early-stage teams should run the simplest plan that can possibly work: one metric, one rate, one accelerator. There is not enough data to calibrate anything more elaborate, and the flexibility to change next year is worth more than precision this year. Mature organizations with clean attribution can support multi-component plans, segment-specific multipliers, and strategic overlays — but they should still ask, for each component, what behavior it is buying.

Data quality is a hard constraint, not a preference. If your CRM cannot cleanly attribute a deal to a rep, you cannot pay on it — not because of principle but because every ambiguous case becomes a dispute, and disputes compound. This is why almost nobody successfully compensates on "influenced revenue": it is defensible analytically and indefensible politically. Fix the attribution before you fix the plan, and treat comp-grade data hygiene as an ongoing RevOps obligation rather than a one-time cleanup.

Strategic priorities get a component, not a rewrite. When leadership wants to push a new product line, or shift upmarket, or improve multi-year contract mix, the instinct is to redesign the whole plan around the new priority. The better instrument is usually a spiff or a bounded component layered on top of a stable core — time-boxed, clearly scoped, and easy to retire. It buys the behavior change without destabilizing the plan that is already working.

Downstream effects: what a comp plan does to forecasting, pricing, and the rest of the business

Comp design leaks into systems that have nothing obviously to do with payroll, and RevOps teams that treat the plan as a self-contained artifact get surprised by the second-order effects.

Forecast accuracy is partly a comp artifact. Quarterly-quota plans with steep accelerators produce sawtooth forecasting: deals pulled forward at period end to clear a threshold, and deals pushed into the next period once the threshold is unreachable. That is rational rep behavior, not dishonesty, and it means the forecast distortion is designed in. Annual quotas with quarterly checkpoints flatten some of the sawtooth. If your forecast is reliably wrong in the same direction at the same point each quarter, look at the payout curve before you look at the reps.

Discounting behavior tracks the crediting rule. Pay commission on gross revenue and reps optimize for closing, which means discounting to close. Introduce a margin component or a discount-tiered rate — where a deal closed at list pays a higher rate than a heavily discounted one — and discounting drops measurably. The trade-off is complexity, and it is only worth taking on when discounting is genuinely a problem rather than a hypothetical one.

Contract structure follows the credit definition. If multi-year deals are credited at total contract value, reps will push three-year terms hard, which may be exactly what the business wants or may be a cash-flow problem depending on billing terms. If credited at annualized value, the incentive to push multi-year disappears. Decide which behavior you want, then set the crediting rule to produce it — rather than discovering the behavior your existing rule produced.

Hiring and ramp planning depend on plan realism. A quota nobody hits makes every new hire look like a bad hire, which corrupts the hiring post-mortem and often leads to firing recruiters instead of fixing quotas. Conversely, quotas that are too soft mask genuinely weak hires until the cost is significant.

Cross-functional friction is often comp-shaped. The classic marketing-versus-sales fight over lead quality is frequently a comp artifact: if SDRs are paid on meetings booked, they will book weak meetings, AEs will reject them, and both teams will blame the other's judgment rather than the payout metric. Similarly, sales-versus-customer-success friction over which team owns expansion usually resolves the moment the crediting rule is written down and both plans reference the same definition.

Adjacent functions can borrow the same logic. Partner and channel teams, professional services organizations with utilization or attach targets, and renewals desks all face the same design question — which behavior are we buying, and can we measure it cleanly enough to pay on it. The base/variable causality principle transfers directly: the further a role sits from the revenue event, the smaller the variable share and the more sense a milestone or MBO structure makes relative to a transactional rate.

The through-line is that a compensation plan is the highest-leverage behavioral instrument most revenue organizations have, and it is usually the least deliberately designed. Treat it as a product with users, requirements, and a release cycle. Model it before you ship it, monitor it while it runs, and change it on a predictable schedule with the pipeline in flight protected. Do that, and the plan quietly does the management work that memos and dashboards never quite manage.

Related questions

How often should the plan change?

Once a year, on a fixed schedule, with the new plan communicated 30–60 days before the period starts. Mid-period changes should be rare, narrowly scoped, and always grandfather deals already in the pipeline under the prior rules.

Should SDRs be paid on meetings booked?

No — pay on opportunities accepted by the AE team, not raw meetings set. Acceptance puts a quality gate on the metric. Paying on booked meetings reliably produces high-volume, low-quality calendar invites and a lead-quality argument between teams.

What is a reasonable clawback window?

Commonly 90 to 180 days, tuned to when onboarding failure becomes visible in your product. The purpose is behavioral — it discourages closing known-bad-fit customers — rather than financial, since recovered dollars are usually small.

When do we need dedicated comp software?

When manual calculation starts producing errors or consuming meaningful analyst time — often somewhere in the low double digits of payees. The trigger is trust, not cost: one wrong commission statement damages plan credibility more than a below-market OTE does.

How do we handle a rep who inherits a huge open pipeline?

Define it in the crediting rules before it happens. Common approaches include splitting credit with the departed rep's territory pool, applying a reduced rate on deals already at late stage, or crediting fully but adjusting the ramped quota upward to match.

FAQ

What's the right split between base salary and variable pay?

It depends on how directly the role causes the revenue event. Account executives who personally close deals commonly sit near 50/50. SDRs, whose work ends at handoff, tend toward 60/40. CSMs owning renewals and expansion typically run 70/30, since renewal outcomes depend heavily on product and support. RevOps and sales operations roles usually sit at 80/20 or 85/15 with an MBO component. Shift these by five to ten points to match cycle length and segment risk.

How do I set quotas that are fair but still stretch the team?

Set OTE first from market data, then derive quota as a multiple of it — roughly 4–6x for quota-carrying AEs. Lower in the band for long enterprise cycles, higher for fast transactional motions. Then sanity-check the number against territory potential and prior-year attainment distribution. A quota that is mathematically defensible but sits in a territory with insufficient addressable accounts is still an unfair quota.

Should I cap commissions to control cost?

Almost never. Caps tell top performers the company will not share upside, and they respond by parking deals and updating their résumé. Manage cost through quota design, territory balance, and the accelerator curve instead. If the real worry is a single anomalous mega-deal, negotiate a windfall provision for deals above a defined size in advance, rather than imposing a ceiling on everyone's consistent overperformance.

How simple does the plan need to be?

Simple enough that a rep can compute their payout on a given deal in their head. Two or three components is a practical ceiling for most quota-carrying roles. Complexity does not just confuse — it generates disputes, consumes manager time, and weakens the behavioral signal the plan exists to send. If explaining the plan requires a spreadsheet walkthrough, it is over-engineered.

What should I measure to know the plan is working?

Attainment distribution monthly — the share of reps above 80%, 100%, and 120% of quota. Commission cost as a percentage of revenue against the range finance approved. Voluntary attrition among top performers, especially in the two quarters after a plan change. Dispute volume, which is a direct readout on plan complexity and data quality. Adjust only at period boundaries, with clear communication.

Can I fix a bad plan mid-year?

You can, but do it surgically and always grandfather in-flight pipeline under the old rules. Narrow the change to the specific broken mechanic, communicate the reasoning openly, and give managers the numbers to walk each rep through the personal impact. Broad mid-year rewrites without grandfathering are the single most reliable trigger for a wave of resignations.

Sources

1. Harvard Business Review — https://hbr.org/2015/04/motivating-salespeople-what-really-works
2. Alexander Group — https://www.alexandergroup.com/insights/sales-compensation/
3. WorldatWork — https://worldatwork.org/
4. McKinsey & Company — https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights
5. Bain & Company — https://www.bain.com/insights/topics/sales-and-channel-effectiveness/
6. Xactly — https://www.xactlycorp.com/blog
7. CaptivateIQ — https://www.captivateiq.com/blog
8. SHRM — https://www.shrm.org/topics-tools/topics/compensation
9. Gartner — https://www.gartner.com/en/sales

Related on PULSE

• [How do you build a customer expansion playbook that drives 120%+ NRR?](/knowledge/q10872)
• [How do you run customer onboarding that actually drives retention?](/knowledge/q10856)
• [What is the operator playbook for a 25-minute weekly pipeline review that drives real forecast accuracy vs becoming theatre?](/knowledge/q9519)
• [What drives playbook adoption if your reps think the playbook is trash?](/knowledge/q539)
• [How do you reconcile bookings when Palantir prime contract timing drives your sub revenue recognition?](/knowledge/q10491)