CompanyGraph
Why a decision must be evaluated in the changed environment it creates, not only by its immediate arithmetic.
The first result changes the next conditions
A first-order effect is the direct result of an action: a price cut lowers price per unit, a layoff lowers payroll, or a sales target increases the number of accounts opened. A second-order effect appears when that result changes what other people do. Customers switch or buy more, competitors match the price, employees alter what they report, suppliers renegotiate, or regulators intervene.
The distinction is not “one step versus an infinite chain.” It is a reminder that the company is not the only actor. A cost saving that leaves a team understaffed can create delays and rework. A promotion that increases volume can exhaust inventory and degrade service. A new metric can improve the measured number while moving effort away from the underlying purpose.
Incentives can move the measurement away from the work
Targets and scorecards are especially exposed because people learn which actions are rewarded. In 2016, the Consumer Financial Protection Bureau said Wells Fargo employees opened unauthorized deposit and credit-card accounts to meet sales goals. The direct effect of the targets was more reported products. The later effects included customer harm, remediation, enforcement, and damage to the bank's ability to trust its own sales numbers. The enforcement record does not prove that every target causes misconduct; it documents how a local metric and pressure system can produce a result that defeats the intended customer relationship. The CFPB enforcement release describes the alleged account-opening practice and the penalty.
The practical test is whether the measured output remains connected to the underlying service. Look for complaints, reversals, rework, employee turnover, audit findings, and customer retention alongside the target. A metric that rises while those indicators worsen may be showing displacement rather than improvement.
Competition changes the payoff
A company can model a new product or lower price as if competitors were passive. If rivals can copy the move, the initial margin or share gain may be competed away. If they respond with capacity cuts, bundling, or a different distribution route, the final market can look unlike the original plan. Customers also adapt: they stockpile before a price increase, delay purchases before a product launch, or switch when a policy changes.
Platform policies make the timing visible. A fee or ranking change affects sellers directly, then changes advertising, product selection, prices, and whether sellers build a second channel. In its Amazon case, the FTC alleged that marketplace fees and visibility rules influenced sellers' access to buyers. The allegation is not a final finding, but it is a concrete reminder that a platform decision can change the behavior that created the platform's value. The FTC's case summary describes the alleged seller responses and dependence.
Time separates the saving from the cost
Second-order effects often arrive after the period in which the first-order benefit is reported. Deferring maintenance lowers current expense, but failures, outages, or replacement costs may arrive later. Reducing research can improve a quarter while narrowing the product pipeline. A layoff can reduce payroll immediately while removing the people who know how to solve recurring defects.
The evidence is usually distributed: maintenance records, delivery times, quality complaints, employee departures, renewal rates, and cash spending may sit in different systems. A quarterly margin can show the immediate result but not every deferred burden.
Actions can reinforce or offset themselves
Some decisions create favorable loops. Better reliability can reduce support calls, improve retention, and provide cash for further reliability work. Other decisions create adverse loops. Service cuts increase complaints, complaints increase churn, churn reduces cash, and lower cash forces further service cuts. The loop is not guaranteed; it becomes plausible when each link is observed in the company's actual process.
Acquisitions illustrate both possibilities. The first-order model may include revenue synergies and overlapping costs. The next effects include customer uncertainty, key-employee departures, duplicated systems, and management time diverted from the existing business. A deal can still work, but the integration evidence must be tracked separately from the purchase announcement.
How to use the concept without pretending to predict
For any material decision, write down the direct effect, then ask who is affected, what each actor can change, and what new incentive or constraint the decision creates. Trace one or two plausible responses rather than inventing an unlimited narrative. Identify the earliest observable signal for each response and the person with authority and resources to correct it.
Use scenarios and ranges. Ask what happens if competitors match, customers delay, employees optimize the metric, suppliers pass through the cost, or regulators impose a new condition. Then compare the scenario with later evidence. Second-order reasoning improves the questions an investor asks; it does not convert uncertainty into a forecast.