Sales Compensation Guide: Designing a Plan That Changes Behaviour

A commission plan is not a reward mechanism. It is an instruction. Whatever you pay for is what your team will do more of, and whatever you do not pay for will happen only when someone is feeling generous. That is the entire theory, and almost every plan failure is a case of paying for one thing while asking for another.

So the design question is never what is competitive. It is: what behaviour do we need more of over the next twelve months, and does this plan pay for exactly that? If you need new logos, paying a flat rate on all revenue including renewals will get you renewal chasing, because renewals are easier. If you need retention, paying only on new business will get you a team that hands over customers and never looks back. If you need a specific product sold, a single blended rate will get you whichever product is easiest to sell.

Four decisions, in order

Every plan, however elaborate, is four decisions. How much pay is at risk, which sets how strongly the plan can influence anything. What the variable pay is measured against, which sets what it influences. When commission is earned, which sets how much risk sits with the rep. And when it is paid, which sets how immediately the feedback lands. Answer them in that order and most of the rest follows. Answer them out of order, or leave one implicit, and you get a plan that is argued about every month.

The honest principle is that the variable share should track how much control the individual has over the result. A rep who finds, works, and closes their own deals has high control, so a substantial variable share is fair and motivating. A rep who works inbound leads allocated by someone else, on a product with a six-month procurement cycle, has far less control, and loading their pay onto that outcome mostly transfers business risk to a person who cannot manage it.

Two practical tests. First, the bad-luck test: could a rep do everything you would want, and still take a serious financial hit through no fault of their own? If so, the variable share is too high. Second, the indifference test: at target, is the variable component large enough that a rep would change how they spend a Tuesday to earn it? If not, the plan is decoration and you should stop pretending it drives anything.

Role matters more than seniority here. New business roles usually carry more variable than account management. Roles that book meetings rather than close them should be paid on the thing they control, which is qualified meetings that progressed, not on eventual revenue they never touched again.

Most unachievable quotas come from the same arithmetic: annual revenue target, divided by number of reps, divided by twelve. That number is a wish. It contains no information about whether a person can produce it.

Build it from capacity instead

Start from the selling days in the period after holidays, training, and administration. Estimate how many real conversations one person can hold in a selling day, using your own call logs rather than an aspiration. Apply your own historical conversion from conversation to opportunity, and from opportunity to closed deal. Multiply by your actual average deal size, not the largest deal you have ever done.

What you now have is the plausible output of a competent person working the process you actually run. Compare it with the target. If capacity comfortably exceeds target, your quota can sit above capacity-derived expectations and still be fair. If capacity falls short, you have learned something important: the target requires either more people, a better conversion rate, or a larger average deal, and no quota number will conjure any of those.

Publish the working

Reps argue with numbers that arrive from nowhere and accept numbers whose derivation they can inspect, even when the number is demanding. Show the calculation in the quota conversation. It also has a useful side effect: it forces you to look at your own conversion rates honestly, and those are usually where the real leverage sits.

Read the distribution, not the average

At the end of a period, look at the spread of attainment across the team rather than the mean. A healthy spread has most people clustered somewhere around target with a tail either side. A bimodal spread, where everyone is either far below or far above, usually means the quota is not calibrated to the territory or the lead flow, and the difference between reps is mostly the difference between the patches they were given.

Crediting is where plans quietly fail, because the plan describes the rate and forgets to describe who the rate applies to. Six cases cover almost everything, and each one needs a sentence.

The specific choices matter less than having made them in advance. A rule decided while an argument is running will always look like a decision about the people rather than about the principle, whichever way it goes.

Three different moments, and conflating them causes most disputes. Commission is earned when the plan says the obligation crystallises, which might be at signature, at invoice, or at cash collection. It may then be held in part against cancellation or non-payment. It is paid in a specific payroll run.

For small companies a workable pattern is to earn at invoice, release most of the commission in the following payroll, and hold a minority portion until the customer has paid or passed an initial period. This keeps the feedback loop short enough to influence behaviour while protecting against the deal that unwinds in month two.

Clawbacks deserve care. A clawback that recovers commission on a customer who cancels in the first weeks is reasonable and most reps accept it. A clawback that reaches back six or twelve months turns the plan into something reps feel they can never rely on, and it tends to encourage exactly the defensive, low-risk selling you did not want. Where you use one, cap the window, state it in the plan, and apply it consistently rather than only when the amount is large.

Take a three-person new business team selling a mid-priced service with a six-week sales cycle. Historical figures from the CRM show each rep holds roughly eight substantive conversations a day, that about one in six becomes a qualified opportunity, and that roughly one in four opportunities closes. There are around sixty selling days in the quarter after holidays and training.

Sixty days at eight conversations gives 480 conversations. At one in six, that is eighty opportunities. At one in four, twenty closed deals. Multiply by the actual average deal size and you have a defensible quarterly quota for one rep, and you have it with the arithmetic visible.

Now test the plan against behaviour. If you pay a flat rate on everything including renewals, the fastest route to that quota is renewals, and new logo growth will stall. So the plan pays the standard rate on new business, a reduced rate on renewals, and an uplift on the one product line the business needs to establish. Three rates, each with a one-line reason.

Crediting: split credit where two reps genuinely worked a deal, recorded on the opportunity at the time rather than reconstructed later. Timing: earned at invoice, four fifths released in the next payroll, one fifth after payment. Ramp: new joiners at a reduced quota for the first two months, since the cycle is six weeks and expecting full production earlier is arithmetic nonsense.

At the end of the quarter, one rep lands at 130 per cent, one at 96, one at 61. Before concluding anything about the third rep, check the inputs: conversations per day, opportunities created, and average deal size. If their conversation volume matches the others but conversion is far lower, that is a skill or a territory problem. If conversation volume is half the others, that is an activity problem. Those two situations need completely different responses, and the attainment number alone cannot distinguish them.

The plan is too complicated to hold in your head

If a rep cannot estimate their own commission on a deal while standing in front of a customer, the plan cannot influence what they do in that moment. Fix: reduce to a maximum of three measures. Complexity in a comp plan is almost always a sign that several people each got their metric added.

Changing the plan mid-period

Saves money once, costs trust for years. Fix: fix the period, announce changes only at boundaries, and explain the reasoning before the new period starts rather than in the same email as the new numbers.

Paying on things the rep does not control

A rep paid on gross margin who cannot set price is being paid on a lottery. Fix: pay on what they can change, and manage the rest through targets and coaching.

No visible running attainment

When people cannot see where they stand until payday, they either disengage or invent their own tracking, which then disagrees with yours. Fix: a live attainment view they can check any day, sourced from the same records the calculation uses.

Quota as a negotiating position

Setting quota high on the assumption that people will fall short is a strategy that works exactly once. Fix: set an honest quota and manage performance separately. Those are two different conversations and merging them corrupts both.

Ignoring the top performers leaving

The most expensive signal a plan can give, and the easiest to explain away. Fix: treat voluntary attrition among high attainers as a plan defect until proven otherwise, and ask directly in the exit conversation whether pay design contributed.

Check four things each quarter. Attainment spread: most of the team should be near target, with a tail either side. Cost of sales as a share of revenue, which should be stable or falling as the team matures. Dispute count and resolution time, which is the clearest proxy for whether the plan is understood. And timing of closed deals within the period, which reveals whether the plan is shaping when business arrives rather than how much.

One qualitative check is worth as much as the four numbers: ask each rep to explain their own plan back to you in two sentences. If they can, the plan is doing its job. If three people give three different explanations, the document needs rewriting before the rates do.

Every calculation above depends on records that already exist somewhere: conversations held, opportunities created, deal size, close dates, and who worked what. If those live in a spreadsheet updated at month-end, your quota setting is guesswork and your crediting is memory.

HelloGrowthCRM keeps activity, opportunity ownership, split credit, and close dates on the deal record as work happens, so attainment is a view rather than a monthly reconstruction, and a rep can check where they stand without asking. There is a free plan to start on, and paid access is $10/user/month billed annually.

Related reading: sales automation, lead management software, CRM for small business, CRM versus a spreadsheet, product features, use cases, and pricing.