Glossary · Game Theory
Principal-Agent Problem
also: agency problem · agency theory · principal–agent problem · moral hazard · multitasking problem · incentive design
Definition
The principal-agent problem arises when one party delegates work to another whose actions are imperfectly observable and whose interests differ. Jensen and Meckling (1976) and Holmström (1979) formalized the resulting trade-off between risk sharing and incentives; sales compensation, agency fee structures, and attribution disputes are its everyday forms inside a growth organization.
Because effort is hidden, the principal can only contract on outcomes, and outcomes are noisy, so strong incentives load risk onto the agent and weak incentives invite shirking. Holmström's informativeness principle says every signal that carries information about effort should enter the contract, which is the theoretical case for relative performance evaluation. Holmström and Milgrom (1991) added the multitasking problem: when only some tasks are measurable, paying for those tasks pulls effort away from the unmeasured ones, which is why high-powered commission plans corrode customer quality and why agencies paid on media spend overspend. Baker (1992) showed that gaming follows whenever the performance measure diverges from the principal's real objective, a pattern later popularized as Goodhart's law.
Essays on this concept
- Marketing Strategy
Principal-Agent Problems Inside the Growth Organization: Incentive Design for Sales, Agencies, and Attribution
A growth organization is a stack of principal-agent contracts: CMO to sales, company to agency, team to attribution model. All three fail the same three ways, and the fixes are contract design, not culture.
- Marketing Strategy
The Strategy-Execution Gap in Growth Teams: Why OKRs Fail and How Input Metrics Fix Them
Your Q1 OKR was 'increase activation rate by 15%.' It's March and you're at 3%. The problem isn't execution, it's that activation rate is an output. You can't execute on an output. Input metrics bridge the gap between strategy and daily action.
- Marketing Strategy
Bargaining Theory for Enterprise Deals: Outside Options, Patience, and the Split-the-Difference Fallacy
Nash and Rubinstein converge on one split; experiments show outside options matter only when they bind; across millions of field bargains people split the difference anyway. What that reframes for enterprise deals.
- Game Theory
Mechanism Design for Marketplaces: Incentive Compatibility as a Product Requirement
Ratings, reviews, rankings, and fees are mechanisms, and users play them. eBay feedback ran 99.3% positive while 3.39% of sales went bad. Incentive compatibility is the test, and it belongs in the spec.
- Behavioral Economics
Hyperbolic Discounting and Subscription Fatigue: A Quantitative Framework for Churn Prediction
How time-inconsistent preferences explain why subscribers cancel, and a mathematical framework that predicts churn windows before they open.
- Marketing Engineering
Multi-Touch Attribution Is Broken, A Causal Inference Approach Using Directed Acyclic Graphs
MTA models overestimate retargeting by 340% and underestimate display by 62%. The fix isn't better heuristics, it's abandoning correlational attribution entirely in favor of causal graphs.
Related concepts
Authoritative references