You get the behaviour you pay for: designing a sales comp plan that doesn't backfire

1. Before you start

A sales compensation plan is a set of instructions written in money. Whatever behaviour it pays for, it will get — including behaviour nobody intended. A tiny example. Pay a rep a flat bonus for every new customer signed, and nothing for whether that customer stays. A rep with two prospects — one a careful buyer who will use the product for years, one a poor fit who will sign today and leave in three months — is being paid, by your own plan, to treat them as identical. Both are “one new customer.” You did not intend to buy churn, but you wrote a rule that pays the same for it, so churn is what shows up. That gap — between the behaviour you meant to buy and the behaviour your plan actually pays for — is the whole subject of this course.

Three honest statements before you start:

This is a Decide course. You read a situation, learn the ideas, and make a call. You do not write or run any code, and there is no calculator here — the reasoning is qualitative.
The companies below, Harbor and Tideway, are composite — invented SaaS firms built from ordinary, illustrative figures so the reasoning is clean. No number here is a claim about any real company, and no real firm is the subject.
This is not a certification. It shows, to you, that you can read a comp plan for the behaviour it will actually produce and defend a redesign.

You need no arithmetic beyond comparing a few figures in your head. The difficulty is judgment: seeing, before you ship a plan, the behaviour it will quietly reward.

2. The Situation

Harbor is a composite B2B SaaS company selling workflow software, around $40M ARR, and its VP of Sales is rewriting the account-executive comp plan for next year. The draft pays commission only on new-logo bookings, adds a steep accelerator above quota, and caps commission once a rep passes 150% of target — on paper, tidy and affordable. The question that lands on your desk: ship this plan, or is it quietly wired to produce exactly the behaviour Harbor is trying to escape — reps who sandbag, cherry-pick, discount to close, and walk away the moment the signature is dry?

The trap is that a comp plan is judged on a spreadsheet, where every rep is assumed to sell more of the right thing. Reps instead read it for the easiest path to the biggest cheque — and find paths the designer never pictured.

3. What you’ll be able to do

After this course you will be able to:

Read a proposed comp plan for its incentive surface — not “is the payout fair?” but “what is the easiest behaviour this rewards, and is that the behaviour we actually want?”
Name the failure mode a plan feature invites — sandbagging and pull-forward from accelerators and period boundaries, cherry-picking and reflex discounting from volume-and-speed incentives, and churn from paying on the signature, not the relationship.
Explain why a commission cap controls budget but de-motivates and drives out the very reps you most want to keep, and say when a cap is still the right call.
Make and defend a redesign that aligns the plan to the real goal — durable revenue, not raw logos — and name the one measure whose absence is doing the most damage.

4. Prerequisites & time box

Prerequisites: none beyond knowing what a sales commission is (a rep is paid a base salary plus a variable amount tied to what they sell) and that a SaaS business lives on revenue that recurs — a customer who churns quickly can cost more to win than they pay back. No spreadsheet or setup — the Decide hall is read-and-decide in the browser; see its how-to-read page if this is your first concept course. No prior Decide course is assumed.

Time box: about 25 minutes of reading (measured), plus real thinking time on the call in section 7. That is at the 25-minute cap for a concept course.

Difficulty: 6 / 8 — a manager-level decision: several factors move at once and the naive reading points the wrong way, so you have to reason past it.

Free-tier honesty: no signups, no paid tools, requires_gpu: false. Nothing here costs money to learn.

5. The case & where the numbers come from

Harbor is a composite SaaS company. Its plan mechanics and figures — the salary split, quota, commission rate, accelerator, cap — are in-course illustrative assumptions, chosen to be typical of B2B SaaS and clearly labelled as such, not drawn from or claimed about any real firm. The ideas used to reason about them — Goodhart’s law, the principal-agent problem, perverse incentives, performance-related pay, churn — are standard and cited in section 11. There is no arithmetic you must reproduce; the figures exist only to make each incentive concrete.

Harbor’s draft AE plan (all figures illustrative):

Plan element	Draft (illustrative)
Base salary	$80,000
On-target earnings (OTE)	$160,000 (so $80,000 is variable, at target)
Annual quota	$800,000 in new-logo bookings (ACV)
Commission up to quota	~10% of new-logo ACV
Accelerator above quota	~15% of ACV above quota
Cap	commission stops accruing past 150% of quota
Renewals and expansion	pay the AE nothing
Customer retention / churn	not in the plan at all

Those last two rows — renewals and expansion pay nothing, retention is absent — do as much to shape behaviour as anything the plan pays for.

6. The Concepts

You get the behaviour you pay for

A comp plan is not a reward you hand out after the fact; it is a forward instruction the rep optimises against every day. So the first question about any plan is not “is it generous?” but “what is the easiest way for a rational rep to maximise their pay under it — and is that the behaviour we want?” Finding the easy path is the rep’s job, not a character flaw. If it and the goal point the same way, the plan works. If they diverge, the plan pulls the rep away from the goal, and no exhortation in the kick-off meeting pulls them back — the money is louder.

Two named ideas sit underneath this. The first is the principal-agent problem: when a principal (Harbor) hires an agent (the rep) to act on its behalf, the two do not automatically want the same things, and the principal cannot watch every move. The comp plan is the main lever for lining the agent’s self-interest up with the principal’s goal — do it well and the rep needs little supervision; do it badly and the rep works hard at the wrong thing.

The second is Goodhart’s law: when a measure becomes a target, it ceases to be a good measure. New-logo bookings are a fine description of a healthy sales team. The moment bookings become the thing that pays, reps optimise the number itself — and it can be pushed up by behaviour that leaves the business worse off (bad-fit logos, discounting to inflate close rates, timing deals for the accelerator). This is why “we’ll just pay on the KPI we care about” is not the safe move it sounds: paying on a proxy corrupts the proxy. The rest of this section is four specific ways Harbor’s draft lets that happen.

Accelerators, and the timing games they invite

An accelerator pays a higher commission rate on bookings above quota — Harbor’s jumps from about 10% to about 15% past $800k. But it turns the exact moment a deal is booked into money, and reps respond by managing the timing of deals, not just the winning of them.

Two timing games follow, and Harbor’s plan invites both. The first is pull-forward: as a period closes, a rep near quota has a sharp incentive to drag deals over the line now — even at a worse price — to reach the accelerator before the clock resets. That is where much year-end discounting comes from: not customer hardball, but a rep racing a deadline. The second is sandbagging: a rep who has already blown past the accelerator, or sees no way to reach it this period, holds finished deals back — to bank an easier start next period, or to dodge a quota that ratchets up if they over-deliver. Either way the plan has taught the rep to optimise the calendar. The tell is a booking pattern that spikes at every period boundary and collapses right after — a rhythm that reflects the comp cliff, not customer demand.

The design lesson is that sharp edges in a plan create games at the edges. The steeper the accelerator and the harder the period cutoff, the more the payoff turns on which side of a date a deal lands — so the more effort flows into moving dates instead of moving the business.

Caps, and why they de-motivate top reps

A cap stops commission accruing past some attainment — Harbor’s draft freezes payout at 150% of quota. Finance likes caps because they make cost predictable and cap a “runaway” rep’s earnings. But look at what a cap says to the person best at the job. To a rep who can sell far past quota, it reads: past here, more selling earns you nothing. The rational responses are the ones you least want. They coast once the cap is in sight. They sandbag the overflow into next period, so the company sees the revenue late or not at all. Or, worst, they leave for a competitor whose plan is uncapped — taking the top of your performance distribution with them.

That last point is the sting. Sales performance is not evenly spread — a small share of reps produce an outsized share of revenue, so the cap bites only on your best and is invisible to everyone else. It saves a bounded, visible amount of commission budget while risking an unbounded, delayed amount of lost bookings and talent.

None of this makes a cap always wrong. A cap can be defensible when a payout would be a genuine windfall the rep did not create — a house account that renews itself, a one-off whale from a territory quirk, a data error. The honest way to handle those is a narrow, named exception (a windfall clause, per-deal review), not a blanket cap that treats every over-performer as a problem to contain. If your reason for the cap is “we might have to pay a great rep a lot,” that is not a reason to cap — that is the plan working.

Cherry-picking and the race to discount

When the plan pays on the count and speed of deals more than their quality, two behaviours follow that both look like “hustle” and both erode the business. The first is cherry-picking: faced with a hard, high-value, long-cycle account and several small, easy, fast-closing ones, a rep paid per booking against a quarterly quota will rationally chase the easy deals and let the strategic account drift. The plan has priced a difficult flagship and three trivial customers as roughly the same pay for very different work — so the rep does the easy work, and the company loses the accounts that would have mattered most.

The second is reflex discounting. If what pays is closing — hitting the quota gate, catching the accelerator, winning the leaderboard — then price becomes the rep’s cheapest tool for buying certainty and speed. A discount converts a maybe, later into a yes, now, and if the plan rewards the now, the rep spends the company’s margin to get it. This is sharpest when a plan or a sales contest rewards raw logo count without regard to price realised: a dollar of discount then costs the rep almost nothing and buys a faster close, so they discount freely and the price sags.

Both behaviours trace to the same root: the plan measures whether a deal closed, not whether it was a good deal at a fair price. Pay on quantity and speed alone, and you get quantity and speed — at the expense of quality and price.

Measure what matters: retention, not just new logos

Now the two absences from Harbor’s plan. New-logo bookings pay; renewals, expansion, and retention pay nothing. For a subscription business this is the most consequential line, because a SaaS company does not live on signatures — it lives on revenue that recurs. A customer who churns inside a few months can cost more to acquire than they ever pay back, so a “new logo” is good only if it stays. By paying solely for the signature, the plan tells the rep that what happens to the customer afterwards is not their problem — and so it isn’t. Reps sign marginal-fit accounts they suspect will churn, because the plan pays the same for a durable customer and a doomed one. The churn the company is fighting is, in part, a behaviour its own plan is buying.

This is the measure-what-matters principle, the constructive side of Goodhart’s law: if paying on a proxy corrupts the proxy, the discipline is to make what you pay for as close as you can to what you actually want. Harbor’s real goal is not logos; it is durable recurring revenue — often read as net revenue retention, the degree to which the existing base grows rather than leaks. A plan aligned to that goal pays for the behaviour that produces it: margin-aware new business (so discounting is not free), plus some stake in whether the customer stays and grows — a portion of pay tied to retention or expansion, or a clawback if a new customer churns inside a set window. The point is not any one mechanism; it is that the plan should reward durable revenue, so the rep who protects it is paid more than the one who books a signature and moves on.

The limits keep you honest, in both directions:

You cannot pay for everything. Every metric you add dilutes focus, and a plan a rep cannot understand cannot steer them — if they can’t see how today’s action maps to this month’s cheque, the incentive goes dead. Alignment means choosing the few right measures, not stacking on more. It is also not the only lever: coaching, deal review, and hiring for fit move the same behaviours.
Some gaming is harmless. The question is never “is the rep responding to the incentive?” (they always are) but “does the behaviour it produces help or hurt the real goal?” A plan so tightly policed that reps spend their time proving compliance is its own failure.

The lesson is blunt: a comp plan is read for the easiest path, so design it for the behaviour you actually want, and assume every gap will be found.

7. Your Call

You have seen how Harbor’s plan quietly buys behaviour it never meant to. Now a different decision lands on your desk.

Tideway is a composite vertical-SaaS company selling scheduling software to dental and veterinary clinics — a different company in a different corner of SaaS from Harbor. Its AEs are paid a flat commission on new bookings with a rich quarterly accelerator, and every quarter the company runs a sales contest: a cash bonus to whoever signs the most logos, counted one per clinic regardless of deal size. Two things worry leadership. Bookings spike every quarter-end and collapse in the first month of the next quarter, and reps privately admit holding deals for the deadline. And the average discount has ballooned while churn among the smallest clinics is climbing. Finance’s proposed fix: add a commission cap to “control the cost and stop the discounting frenzy,” and keep the logo contest running as the motivator.

How this differs from the taught case (the transfer): a different company and sub-sector (Tideway, a vertical-SaaS vendor to clinics, not Harbor’s horizontal workflow SaaS); different figures and plan mechanics (a flat commission plus a most-logos cash contest, not Harbor’s accelerator-and-cap-on-ACV draft); a different kind of decision (diagnose and fix a live plan already misbehaving, and judge a proposed fix, rather than approve-or-reject a fresh draft); and an added constraint the taught case did not have — an existing sales contest paying on raw logo count that cannot simply be ignored. The core concept is the same: you get the behaviour you pay for.

8. Self-check

Before you write the memo, make sure you can say each of these in one line:

For a given plan feature (an accelerator, a cap, a logo contest), what is the easiest behaviour it rewards — and is that what the company actually wants?
Why does a commission cap save a bounded, visible amount while risking an unbounded, delayed one, and who does it bite?
What is Tideway’s real goal, which measure’s absence does the most damage, and what change would you make first?

If any is fuzzy, reread section 6: those five ideas are the whole course.

9. Stretch

Push the thinking further on your own:

Harbor’s plan pays AEs nothing on expansion, yet most SaaS growth comes from the existing base. Design the smallest change that would make a rep care about a customer’s second year — and name the new behaviour it might accidentally invite. (Every fix is a plan with its own easiest path; find that path before you ship it.)
The genuinely hard one: you may add exactly one measure to a pure new-logo plan, because the reps must still understand their plan. Which single measure buys the most alignment for a SaaS business, what does it still fail to protect, and why does it beat the two runners-up?

10. Ship it — your decision memo

Write a one-page memo to Tideway’s leadership. State the call (do not add the cap; the discounting and the quarter-end spike come from the contest and the deadline, so fix what the plan measures — replace the raw-logo contest with pay on margin-aware bookings plus a stake in retention, and smooth the period cliff). Show the reasoning in two or three lines (the plan buys the churn and the discounting because it rewards logo count and speed while paying nothing for durability; a cap would freeze payout on your best reps without touching either cause). Name what you rejected (the cap, and doubling the contest prize) and why, and the one thing that would change your mind (evidence the discounting is customer-driven, not deadline-driven). This memo is your own argued claim — not a credential.

11. Sources

Harbor and Tideway, and every figure attached to them, are composite and illustrative — constructed to be typical of B2B SaaS for clean teaching, not drawn from or claimed about any real company. The ideas used to reason about them are standard; references below.

Concept / claim	Source (publisher)	URL	Accessed
”When a measure becomes a target, it ceases to be a good measure”	Wikipedia — Goodhart’s law	https://en.wikipedia.org/wiki/Goodhart%27s_law	2026-07-20
Aligning an agent’s incentives with a principal’s goal under imperfect monitoring	Wikipedia — Principal–agent problem	https://en.wikipedia.org/wiki/Principal%E2%80%93agent_problem	2026-07-20
Incentives that produce behaviour their designers did not intend	Wikipedia — Perverse incentive	https://en.wikipedia.org/wiki/Perverse_incentive	2026-07-20
Pay tied to measured performance, and its effects	Wikipedia — Performance-related pay	https://en.wikipedia.org/wiki/Performance-related_pay	2026-07-20
How a variable, quota-based sales incentive is structured	Wikipedia — Sales management	https://en.wikipedia.org/wiki/Sales_management	2026-07-20
Why a churning customer can cost more than they return in a subscription business	Wikipedia — Churn rate	https://en.wikipedia.org/wiki/Churn_rate	2026-07-20

Next up

That’s the top of the Sales & Revenue track. Browse all courses → to pick your next call.