Designing auctions matters because small rule changes change incentives and revenue.
You roll out a new commission plan for your sales team: reps earn 12% on any deal over $5,000 ARR, but zero on smaller ones. Within a month, your Close Rate on deals under $5,000 drops to near zero - reps are either combining small customers into forced packages to clear the threshold or ignoring them entirely. Revenue from the $2,000-$4,999 segment falls 40%. You didn't change the product, the market, or the team. You changed one rule, and behavior followed.
Incentives are the rules - explicit or implicit - that make a specific action more or less attractive to the people choosing it. When you design Pricing, Commissions, or any system where humans make choices, the incentive structure determines what actually happens - often in ways you didn't intend.
An incentive is anything that changes the Expected Value of a choice from someone's perspective. That's it.
When a salesperson decides which lead to call next, they're running a rough calculation - consciously or not: what do I get if this works, what does it cost me to try, and what's my next-best Outside Option? The answers to those questions are shaped by the rules you set - compensation plans, performance targets, recognition, penalties.
Incentives aren't just money. Time is an incentive. Status is an incentive. Avoiding pain is an incentive. But for Operators running a P&L, the ones that matter most are the ones you can design: Pricing structures, Commissions, bonus criteria, penalty clauses, and the rules of any auction or Allocation system you build.
The core insight is simple but easy to forget: people optimize for the metric you reward, not the outcome you want. This is closely related to Goodhart's Law - once a measure becomes a target, it stops being a good measure - but incentives is the broader concept. Goodhart's Law is what happens when you get incentive design wrong.
Every line on your P&L is the result of someone's decisions - your customers' decisions to buy, your team's decisions about where to spend effort, your vendors' decisions about what to charge. You don't control those decisions directly. You control the incentive structures those decisions happen inside.
Three places this hits your Operating Statement hardest:
The P&L doesn't show you incentive misalignment directly. It shows you the symptoms - declining Close Rate, rising selling costs, unexpected Churn. Diagnosing those symptoms often comes back to asking: what behavior did our rules actually reward?
Incentive design has three moving parts:
1. The Expected Payoff per action - What does the person get for each possible choice? This connects directly to Expected Value. A rep facing two leads calculates (roughly): probability of closing times the Commissions they'd earn, minus the opportunity cost of time spent. If your commission plan pays $0 on deals under $5,000, the Expected Value of working a $3,000 lead is literally zero - no matter how easy the close.
2. Informational Advantage - What does the person know when they choose? In an auction, if Buyers can see each other's bids, they behave differently than if bids are sealed. In a sales team, if reps can see Pipeline Volume by account Allocation, they lobby for reassignment instead of working harder. What you let people see changes what they do.
3. The Outside Option - What happens if the person walks away? A salesperson whose Total Compensation is entirely variable - 100% Commissions, no guaranteed fixed portion - takes different risks than one with a predictable income floor. A Buyer in an auction with a strong alternative supplier bids less aggressively. The Outside Option sets the floor.
When you change any of these three, behavior shifts - sometimes dramatically. Small rule changes can flip outcomes because incentives interact. A bonus for hitting a performance target (Expected Payoff) combined with visible Pipeline Volume (Informational Advantage) combined with easy lateral moves inside the company (Outside Option) can create a system where your best reps game account Allocation rather than sell.
This is why auction theory is a serious field. The difference between a first-price auction (you pay what you bid) and a second-price auction (you pay the second-highest bid) seems small, but it changes incentives completely. In a first-price auction, rational Buyers shade their bids below their true value - this is Bid Shading. In a second-price auction, the Dominant Strategy is to bid your true value. Same items, same Buyers, different rules, different behavior.
Think about incentive design any time you're:
The general decision rule: if humans make choices inside your system, you have an incentive design problem whether you know it or not.
Your SaaS company has 200 customers and $1.2M ARR. A sales team of 5 reps closes about 15 deals/month across three segments: Enterprise ($10K+ ARR, 3 deals/mo team-wide), Mid-market ($3K-$9K ARR, 5 deals/mo), and SMB ($500-$2K ARR, 7 deals/mo). Average deal size: $12,000 ARR for Enterprise, $5,000 for Mid-market, $1,000 for SMB. You switch from a flat 8% commission on all deals to 12% on deals over $5K ARR and 0% on deals at or below $5K.
Before the change: monthly Commissions per rep across the deal mix. Each rep's share of 15 team deals: 0.6 Enterprise, 1.0 Mid-market, 1.4 SMB. Commission = (0.6 × $12,000 × 8%) + (1.0 × $5,000 × 8%) + (1.4 × $1,000 × 8%) = $576 + $400 + $112 = $1,088 per rep/month. Team total: $5,440/month.
After the change: SMB deals pay $0 commission. Mid-market deals at the $5,000 average sit right on the threshold - most fall at or below it and pay nothing. Enterprise deals jump to $1,440 each (12% × $12,000). The rational rep ignores SMB entirely and redirects effort toward pushing Mid-market prospects above the $5K line.
SMB segment collapse: the team was adding 7 SMB deals/month at ~$1,000 ARR each. Post-change, that drops to ~2 incidental closes. The team stops adding roughly $5,000 in new ARR per month. After 12 months, the book is $60,000 ARR smaller than projected. But the Revenue shortfall in year one is not $60,000 - the loss accumulates gradually. Month one you're short $5,000 ARR (~$417/month of Revenue). Month six you're short $30,000 ARR (~$2,500/month). Actual Revenue lost in year one: approximately $32,500 - the sum of the monthly shortfalls as the ARR gap widens.
Mid-market distortion: reps pressure $3,000-$4,000 ARR prospects into $5,000+ commitments they don't need. Short-term ARR per deal rises, but Churn Rate in that segment increases from 5% to 15% annually because customers are overcommitted. On a Mid-market base of ~60 customers at $5,000 average ARR, that's 6 extra churned accounts - roughly $30,000 ARR lost, with about $15,000 of Revenue impact in year one as the Churn phases across the calendar.
Net P&L impact over 12 months: Revenue lost from SMB collapse: ~$32,500. Revenue lost from excess Mid-market Churn: ~$15,000. Commission costs increase because paying 12% on qualifying deals exceeds the old 8% rate, even with fewer total deals: approximately +$12,000 in selling costs. Net year-one damage: roughly -$60,000, before the selling costs of re-acquiring churned customers.
Insight: The commission change had a clear intent - focus on bigger deals - but ignored the Expected Payoff math at the boundary. The $5,000 threshold created a cliff where deals worth $4,999 ARR became literally worthless to the rep. Any time you put a hard threshold in an incentive system, expect behavior to cluster around gaming that threshold.
Same SaaS company from Example 1. After seeing the damage from the $5K cliff, you redesign Commissions using a graduated scale instead of a binary threshold: 6% on deals under $3K ARR, 8% on $3K-$7K, 10% on $7K-$12K, 12% on deals above $12K.
Under the gradient, a $1,000 SMB deal pays $60 commission (6%). That's less than the original flat 8% ($80) but infinitely more than $0. A rep won't prioritize SMB, but an easy close between meetings is still worth taking - the Expected Payoff is positive, just small.
Mid-market incentives are smooth. A $4,500 deal pays $360 (8%). A $5,500 deal pays $440 (8%). There's no cliff to game - moving a deal from $4,500 to $5,500 earns an extra $80 in Commissions, not an infinite jump from $0 to $660. Reps have no reason to pressure customers into packages they can't use.
The strongest incentive targets Enterprise ($12K+) at 12%, where the company's margins are highest and where Upsell conversations are natural, not forced. A $15,000 deal pays $1,800 - clearly the best use of a rep's time, without any threshold manipulation required.
Expected outcome: SMB segment stabilizes at ~5 deals/month (down from 7 but no collapse). Mid-market Churn returns to baseline because customers aren't overcommitted. Commission costs land between the old flat 8% and the broken cliff plan. The gradient preserves the strategic focus on larger deals without destroying a Revenue segment.
Insight: Gradients create preference without creating cliffs. The rep still earns more per hour on Enterprise deals, so effort naturally flows upward. But the SMB floor prevents the catastrophic abandonment that a binary threshold causes. When designing any incentive threshold - Commissions, bonuses, penalty triggers - ask whether a smooth gradient achieves the same directional pressure without the destructive boundary effects.
You run a marketplace that sells 100 ad slots per day via auction. Currently you use a first-price auction (winner pays their bid). Average winning bid is $2.00. You're considering switching to a second-price auction (winner pays the second-highest bid) and adding a $1.50 reserve price.
Under first-price rules, Buyers engage in Bid Shading. A Buyer who values a slot at $3.00 might bid $2.00, hoping that's enough to win without overpaying. Your Revenue: 100 slots × $2.00 average = $200/day.
Under second-price rules, the Dominant Strategy flips. That same Buyer now bids their true value of $3.00, because they know they'll only pay the second-highest bid. If the runner-up bid $1.80, the winner pays $1.80 - not $3.00.
Here's the subtlety most Operators miss: under standard assumptions, both auction formats tend to produce similar Revenue. In a first-price auction, Buyers bid lower but the winner overpays relative to the competition. In a second-price auction, Buyers bid higher but pay less than their bid. The format switch alone is not free money - it reshapes behavior without necessarily changing aggregate Revenue.
The reserve price is where the real gain lives. Without a floor, some slots sell for $0.50-$1.00 to the lone Buyer in thin auctions. The $1.50 reserve price eliminates those below-cost transactions entirely. Any slot without a bid above $1.50 goes unsold - you lose perhaps 5-10 low-value slots per day.
Combined outcome: Revenue rises to roughly $215-$220/day, a 7-10% increase. The gain comes primarily from the reserve price eliminating the worst-paying transactions, not from the format change. Fewer slots sell (maybe 90-95 instead of 100), but Revenue per sold slot rises enough to more than compensate.
Insight: You changed two rules and Revenue moved - but the gain traces almost entirely to the reserve price, not the format switch. This matters: before assuming a mechanism change will generate surplus, identify which specific rule drives the value. Auction theory exists because small rule differences interact in non-obvious ways, and misattributing the cause leads to overconfidence in future mechanism tweaks.
People optimize for the metric you reward, not the outcome you want. Design incentives around the behavior you actually need, then verify by asking: if someone gamed this ruthlessly, would I still be okay with the result?
Hard thresholds in incentive systems - commission cliffs, bonus tiers, penalty triggers - create boundary effects where behavior clusters around gaming the threshold. Use gradients instead of cliffs where possible: a graduated commission scale (Example 2) achieves directional focus without destroying the segments below the cutoff.
Every Pricing decision, compensation plan, and Allocation mechanism is an incentive design problem. If you're not designing the incentives deliberately, the system is still producing incentives - you just don't know what they are.
Designing incentives for the average performer and forgetting the edges. Your best and worst performers respond to incentives most aggressively. The rep about to miss their performance target and the rep who already exceeded it both behave very differently from the median - and they're the ones who break your system.
Assuming people won't optimize. Engineers especially make this mistake when building internal systems: 'nobody would actually game this.' They will. If the incentive exists, someone will find it. The question isn't whether - it's how fast and how much damage before you notice.
You run a customer support team. Current incentive: agents are measured on tickets resolved per hour. Your CSAT score has been dropping for three months while tickets-per-hour keeps climbing. Diagnose the incentive misalignment and propose a revised structure that protects both Throughput and CSAT. Include specific metrics and thresholds.
Hint: Think about what 'resolved' means from the agent's perspective vs. the customer's. What's the fastest way to mark a ticket resolved? What does that do to the customer experience? Then ask: can you design an Expected Payoff where the easiest path for the agent also produces the best outcome for the customer?
The misalignment: 'tickets resolved per hour' incentivizes closing tickets fast, not closing them well. Agents mark tickets resolved prematurely, customers reopen them or Churn silently. The metric goes up while actual service quality drops.
Revised structure: Measure net resolved tickets - a ticket only counts if it stays closed for 48 hours with no reopen and no follow-up complaint. This makes premature closure counterproductive (it costs you a reopen). Add a CSAT floor at 85% - below that threshold, the Throughput bonus zeroes out. This prevents the gaming path of 'close fast and hope nobody complains' because enough people will complain to drop you below floor.
Expected outcome: tickets-per-hour drops 15-20% initially (agents spend more time per ticket), but net resolution rate rises and CSAT recovers within a quarter. The Profit impact is positive because Service Recovery on poorly-handled tickets costs 3-5x what doing it right the first time costs.
A landlord offers two lease structures for the same commercial space: (A) $5,000/month flat rent, or (B) $2,000/month base plus 4% of your monthly Revenue. You expect your business to generate $100,000/month in Revenue in year one, growing 15% annually. Which lease should you choose, and at what Revenue level does your preference flip? Show the math.
Hint: Calculate total annual rent under both structures for year one and year two. Find the break-even Revenue where both structures cost the same. Then think about which structure creates better incentives for the landlord's behavior toward you as a tenant.
Year 1: Option A = $5,000 × 12 = $60,000. Option B = ($2,000 + 0.04 × $100,000) × 12 = $6,000 × 12 = $72,000. Option A saves you $12,000 in year one.
break-even: $5,000 = $2,000 + 0.04R, so $3,000 = 0.04R, so R = $75,000/month. Below $75K monthly Revenue, Option B is cheaper. Above $75K, Option A is cheaper.
Year 2 at 15% growth: Revenue = $115,000/month. Option A still $60,000/year. Option B = ($2,000 + $4,600) × 12 = $79,200. Option A saves $19,200.
Choose Option A if you expect Revenue above $75K/month. But consider the incentive angle: under Option B, the landlord earns more when you earn more. This aligns their incentives with yours - they're more likely to invest in the property, approve improvements, and renew your lease on good terms. Under Option A, they're indifferent to your success. If you're early-stage and uncertain about Revenue, Option B also gives you downside protection - your rent drops automatically if Revenue falls below expectations.
Incentives directly feeds into auction theory and Bid Shading - understanding why Buyers behave differently under different rules requires understanding incentive structures. It's the mechanism behind Goodhart's Law (the failure mode of badly designed incentives) and connects to Game Theory and Dominant Strategy, since equilibrium outcomes depend on what the rules reward. On the P&L side, incentives explain otherwise mysterious patterns in Commissions, Churn, Close Rate, and Cost Structure. When you design Pricing or Subscription Pricing, you're designing incentives for your customers. When you build Quality Gates or Exception Review processes, you're designing incentives for your team.
Disclaimer: This content is for educational and informational purposes only and does not constitute financial, investment, tax, or legal advice. It is not a recommendation to buy, sell, or hold any security or financial product. You should consult a qualified financial advisor, tax professional, or attorney before making financial decisions. Past performance is not indicative of future results. The author is not a registered investment advisor, broker-dealer, or financial planner.