Case Study 26-1: The Wells Fargo Account Fraud Scandal — Goodhart's Law at Scale

Overview

Between 2002 and 2016, Wells Fargo employees opened approximately 3.5 million unauthorized bank accounts and credit cards in customers' names without their knowledge or consent. The scandal, which became public in 2016 and resulted in $3 billion in fines and settlements, the firing of over 5,300 employees, the forced retirement of CEO John Stumpf, and lasting reputational damage to one of America's largest banks, was not primarily the result of individual moral failures. It was the predictable consequence of a performance measurement and incentive system that made fraudulent behavior the rational response to impossible targets.

This case study examines the Wells Fargo scandal as a catastrophic illustration of Goodhart's Law, visibility asymmetry, and what happens when performance metrics become the architecture of control in a high-pressure sales environment.

Background: "Eight Is Great"

Wells Fargo's retail banking strategy under CEO John Stumpf was built around a concept called "cross-selling" — selling existing customers additional products (checking accounts, savings accounts, credit cards, mortgages, insurance, investment accounts). The theory was that customers who held more Wells Fargo products were more financially engaged and therefore more profitable over time.

Stumpf famously promoted the goal of selling each customer eight products — "Eight is great" was a company mantra — because he believed customers would use Wells Fargo as their primary financial institution if they had that many account relationships. Cross-sell ratio (the average number of products per customer) became the primary metric of retail banking success, tracked daily by branch managers, reported quarterly to investors, and used to evaluate everything from branch performance bonuses to individual teller compensation.

The targets were aggressive. Branch managers were expected to hit daily sales quotas and were required to report their numbers to regional managers multiple times per day. In some regions, managers held "jump into January" calls where branch employees competed to open the most accounts. Employees who missed targets faced humiliation, demotion, or termination. Those who hit targets received bonuses, praise, and promotion.

The Metric Becomes the Mission

The cross-sell ratio was a reasonable measure of customer engagement when it reflected genuine customer interest in Wells Fargo products. It became a catastrophic target when branch employees discovered that they could increase the ratio without actually selling anything — by opening accounts customers didn't ask for.

The mechanics were straightforward: an employee with access to a customer's existing account information could open additional accounts, fund them briefly from the customer's existing balance, and record the new accounts as sales. The customer might never notice if the accounts were dormant and the small transferred amounts were returned before the customer noticed — or if the accounts generated modest fees that were easy to overlook in a bank statement.

From the individual employee's perspective, the calculation was grimly rational: meet your quota through fraudulent means and keep your job, or refuse and be fired for missing targets. Former employees testified in congressional hearings that managers who knew about unauthorized account openings pressured employees to continue, and that internal hotlines for reporting fraud were ineffective. Some employees who reported fraud were fired for it — ostensibly for other reasons, but with suspicious timing.

Surveillance Architecture Analysis

The Wells Fargo scandal reveals a surveillance architecture that was extensive upward (management surveilling employee performance) and nearly blind downward (employees surveilling each other) and outward (the organization surveilling itself for integrity).

What was measured: Cross-sell ratio (products per customer), number of new account openings, credit card applications submitted, appointment settings, referrals to investment and mortgage divisions.

What was not measured — or not prioritized: Account usage rates (were the new accounts actually being used?), customer satisfaction with products they held, account closure rates, customer complaints about unauthorized accounts, employee morale and ethics violations.

This is Goodhart's Law in its institutional form: the measurement system was designed to capture sales activity, not customer benefit. Because customer benefit was not measured, it was not managed. Because compliance and integrity were measured primarily through self-reporting, they were only managed when self-reporting occurred. And because employees who reported fraud faced retaliation, self-reporting was suppressed.

The Visibility Problem

Wells Fargo's leadership — including Stumpf and regional executives who received regular cross-sell ratio reports — had extensive visibility into aggregate metrics. What they did not have (or claimed not to have) was visibility into the quality of those metrics: whether accounts were being legitimately opened, whether customers had consented, whether the cross-sell numbers represented genuine engagement.

The CFPB's 2016 consent order found that Wells Fargo had received approximately 700 complaints per year from customers about unauthorized accounts for years before the scandal became public. Those complaints existed in the organization's data — but they were routed to customer service rather than to the executive leadership receiving the cross-sell reports. The complaints were measurable data; they simply weren't measured in a way that connected to the performance dashboard.

This is visibility asymmetry in an organizational form: different parts of the organization had access to different pieces of the picture, and the pieces were not assembled in a way that made the fraudulent behavior visible to those with the power to stop it. Or — a more troubling reading — the picture was visible enough that those in power could have seen it if they had chosen to look.

Consequences and Accountability

The consequences of the Wells Fargo scandal were distributed unequally:

For branch employees: Over 5,300 employees were terminated — many of whom were lower-level workers who had been trapped in an impossible system. Most received no severance or meaningful support. Some had difficulty finding subsequent employment because they appeared on industry regulatory databases as terminated for compliance violations.

For executives: CEO John Stumpf was forced to retire but left with approximately $130 million in accumulated compensation. He was subsequently fined $17.5 million by the OCC and banned from the banking industry. Other executives faced fines and industry bans. No criminal charges were brought.

For the institution: Wells Fargo paid $3 billion in fines and settlements, and agreed to a Federal Reserve asset cap that significantly limited the bank's growth for years. The bank's reputation suffered lasting damage.

For customers: Victims were entitled to compensation through the settlement, though amounts were often small relative to the harm done to credit scores, financial relationships, and trust in banking institutions.

The accountability distribution reflects a fundamental feature of the performance surveillance system: metrics flowed upward (employees' performance was visible to management), but accountability for the surveillance system's design flowed imperfectly downward (executives who designed the incentive system bore less consequence than employees who were trapped in it).

Discussion Questions

1. Wells Fargo's performance measurement system created conditions in which fraud was the rational response to the incentive structure. Does this mean individual employees bear no moral responsibility? How do we think about individual culpability within systems that constrain choices?

2. Customer complaint data about unauthorized accounts existed in Wells Fargo's systems for years before the scandal. At what point does ignorance become culpable? What obligations did executives have to look beyond the metrics they were regularly receiving?

3. The 5,300 employees who were terminated were primarily lower-level workers; the executives who designed the system received comparatively minor consequences. How should accountability be structured in cases where a performance management system produces systemic wrongdoing?

4. The cross-sell ratio was initially a reasonable measure of customer engagement. At what point did it stop being a good measure? What signals should have prompted leadership to question whether the metric was still measuring what they thought it was measuring?

5. Apply the concept of Goodhart's Law to another organization you are familiar with — a school, a government agency, a sports team, a healthcare provider. Identify a metric that might be subject to similar gaming, and describe what the gaming might look like.

Further Context

• CFPB Consent Order against Wells Fargo, September 2016
• U.S. Senate Banking Committee hearing, "An Examination of Wells Fargo's Unauthorized Accounts and the Regulatory Response," September 20, 2016
• Stacy Cowley, "Wells Fargo Review Finds 1.4 Million More Potentially Fake Accounts," New York Times, August 2017
• Caitlin McCabe and Rachel Louise Ensign, "Wells Fargo Fires More than 100 Workers for Allegedly Defrauding Pandemic-Relief Program," Wall Street Journal, 2020 (showing that the cultural problems were not resolved by the scandal)