When the Target Becomes the Job is an essay on perverse incentives in organizations for executive and leadership teams who set targets and measures for others. It shows how targets come to reward harmful behavior and how to check a target for the behavior it will reward before it is set.
A target tied to pay teaches people what to do, and it can teach thousands of people to do the wrong thing.
In his 2010 letter to shareholders, Wells Fargo's chief executive, John Stumpf, explained why the bank wanted every customer to hold eight of its products: eight rhymed with great. The bank set sales goals for its branch employees to drive that number up, and by the bank's own analysis employees opened about 1.5 million deposit accounts that customers may not have authorized. By September 2016 the bank had fired at least 5,300 employees for improper sales practices, and that month Stumpf told the Wall Street Journal that "there was no incentive to do bad things."
The chief executive explained the number with a rhyme, the people paid to achieve it were fired for how they achieved it, and he denied that the incentive had encouraged what they did. Perverse incentives in organizations often follow this pattern: the people who met the target are blamed for how they met it, while the target and its rewards stay in place.
An incentive becomes perverse when it encourages behavior that defeats the purpose it was meant to serve. Every perverse incentive comes from three design decisions that leaders make: what to measure, what target to set on that measure, and what reward or penalty to attach to the target. The trouble often starts with the first of the three, because what an organization chooses to count tells people what to do. "Tell me how you will measure me, and I will tell you how I will behave," Eliyahu Goldratt wrote. Leaders make all three decisions, and they own what the three produce together.
Many scorecards count activity and report results too late to act on
A measure is an observation that reduces uncertainty, and it matters when it can change a decision. In our practice, a measure also matters when it changes behavior, and it proves its utility when that change moves the organization toward the performance it wants.
Many scorecards and dashboards measure busyness, utilization or output, such as the number of features shipped or the story points a team completes each sprint, and they track lagging indicators with no leading ones. Measures of busyness, utilization and output reinforce the logic of the existing operating model: a scorecard that treats full utilization as proof of performance tells people to keep everyone busy, whatever the work is worth to a customer. Lagging indicators record outcomes already produced. A leading indicator gives leaders time to act when it tracks a condition that shapes the outcome. Each choice of what to measure, and of which indicators to track, is a design decision about what the organization will pay attention to. When a measure has no decision attached to it, observation becomes reporting, and reports pile up while decisions go unmade.
Leading indicators help, and they too drift from the condition they stand for once a reward attaches to them. A count of products per customer can serve as a leading indicator of deeper customer relationships until people raise it without deepening any relationship. In its 2014 and 2015 annual reports Wells Fargo described its cross-sell measure as counting "products used by customers," and the Securities and Exchange Commission found in 2020 that the count included accounts customers seldom or never used and accounts opened without their consent. In the 1970s the economist Charles Goodhart observed that a statistical pattern breaks down once authorities lean on it to steer policy, and the social scientist Donald Campbell observed that an indicator used for decisions comes under pressures that distort both the indicator and the work it monitors. A reward attached to the indicator adds to those pressures.
Watermelon reporting, green on the outside and red on the inside, has two common sources. Gamed measures produce it when people raise the number without improving the work. Old measures produce it when a scorecard kept from an earlier strategy goes on rewarding what that strategy needed after the business has changed course.
A reward can be earned without the result it was meant to buy, and a fine can turn an obligation into a price
A target sets the value people are expected to achieve on a measure. A reward tied to a single count makes the count the goal. It pays the same for work that produces the result as for work that produces only the number. When producing only the number costs less, some of the people paid on it optimize for it, and the purpose behind the target drops out of their reckoning. The cobra effect gives the pattern its name. In a story, possibly apocryphal, from colonial Delhi, officials worried about venomous snakes offered a bounty for every dead cobra. The officials treated a dead cobra as proof that a wild snake had been removed from the city. Residents found a cheaper way to produce dead cobras than hunting them: they bred cobras and killed them. The count of dead cobras rose, and so did the number of live cobras in the breeders' pens.
When officials discovered the breeding, they canceled the bounty. The breeders were left holding snakes nobody would pay for, and they released them. In the story, the released snakes joined the wild ones, and the city ended with more cobras than it had before the policy began. The officials paid for dead snakes while hoping for fewer live ones, the pattern Steven Kerr called the folly of rewarding A while hoping for B.
An incentive can also make a problem worse by replacing a reason people already had. In 1998 the economists Uri Gneezy and Aldo Rustichini studied ten private day-care centers in Haifa, Israel, where some parents collected their children late. Six of the centers began fining parents ten shekels for arriving ten or more minutes late. Over the next few weeks, late arrivals at those centers rose to almost twice their earlier level. The parents gamed no count. The researchers concluded that parents had come to treat the fine as the price of extra time, and that the price had replaced the obligation they felt toward the teachers who waited with their children. When the centers removed the fine twelve weeks later, late arrivals stayed at the higher level. Removing the price did not restore the obligation.
The cobra breeders optimized for a count that could be produced without the result. The parents stopped acting on a reason they already had, because a payment took its place. Both failures are design problems, and they belong to the people who design the incentive. Officials designed the bounty, the day-care centers designed the fine, and Wells Fargo's leaders designed the sales goals that produced accounts customers never asked for.
A target that cannot be met legitimately leaves each person a choice between their job and their integrity
When leaders set a target higher than people can achieve by doing the work as intended, each person measured against it faces a private question: should I do something I know is wrong in order to meet the target, or should I do the right thing, potentially miss the target and put my job in jeopardy? Across thousands of branches and desks, the choices people make add up to the organization's behavior.
At Wells Fargo, those choices added up to what the Justice Department called the opening of millions of accounts without customer authorization. The bank built its retail business on cross-selling to existing customers, and all three design decisions were in place: the measure was products per customer, the target was the bank-wide goal of eight that its 2011 annual report repeated, broken down into branch and individual sales goals, and the reward was pay. The Community Bank's leaders set some of those goals knowing that many regions would miss them. The bank's own name for such goals was "50/50 plans," reflecting "an expectation that only half the regions would be able to meet them," according to the board's own investigation. The Consumer Financial Protection Bureau found employees "spurred by sales targets and compensation incentives," and its director said they opened unauthorized accounts "to hit sales targets and receive bonuses." The bureau also reported that employees applied for about 565,000 credit card accounts that customers may not have authorized, alongside the 1.5 million deposit accounts.
In February 2020 the bank agreed to pay $3 billion in a settlement with federal prosecutors and the Securities and Exchange Commission. The settlement record says the Community Bank division's onerous sales goals, and the management pressure that came with them, led thousands of its employees to engage in unlawful conduct. From 2002 to 2016, employees forged signatures, activated debit cards customers had not requested and transferred funds so that unauthorized accounts looked funded. Inside the bank these practices were called "gaming."
Numerical quotas set for others can put the same choice to people every day, and W. Edwards Deming argued that they should be abolished. Most organizations keep targets anyway, and many of their leaders believe a target is what motivates people, even a target that is poorly designed. A sounder reason to keep them is that a target makes a commitment visible and lets people coordinate their work around it, which reduces their uncertainty about what others will do even when the number itself is wrong. A wrong number still coordinates people, and it coordinates them around the wrong commitment. Organizations have other ways to make commitments visible, such as explicit promises between teams, limits on the work each team takes on at once, and service levels agreed with customers. Targets are the tool most organizations inherit, so they are the tool people pick up first. Our first answer to Deming's objection is to ask, before a target is set, whether people can meet it honestly with the work as it runs today.
Leaders who keep a target after learning it produces misconduct are accountable for the misconduct
Forging a customer's signature is personal misconduct, and the people who did it are responsible for it. In Kurt Lewin's equation, behavior is a function of both the person and the environment. The bank's sales policy shaped the environment in which every employee measured against it faced that choice. The leaders who set the targets and the rewards designed that environment, so they are accountable for the dilemma it posed and for what thousands of people did inside it.
Their accountability comes from designing the sales policy, and at Wells Fargo it also comes from awareness of what that policy was producing. The division's senior leaders knew about the gaming as early as 2002, and knew it was increasing because of the goals and the pressure. In 2004 an internal investigator called it a "growing plague." In 2005 another said it was "spiraling out of control." In a 2004 email to Stumpf, Carrie Tolstedt, who later led the Community Bank, wrote: "if you incent a team of bankers on top of that around sales per day alone you are asking for trouble." The board's investigation found that she "did not follow her own advice." Those leaders refused to change the sales model. According to the settlement record, they treated the damage to sales quality and integrity as "the cost of doing business," and they cast the problem as individual misconduct when the sales model itself was driving it.
The firings reached branch managers and their managers, and the sales goals the employees were fired for chasing stayed in place until October 2016. In April 2017 the bank's board took back about $75 million in pay from Stumpf and Tolstedt. In January 2020 the Office of the Comptroller of the Currency charged five former senior executives, and Stumpf agreed to a $17.5 million penalty and a ban from banking.
Leaders can find the cheaper routes to a target before it goes live
Along with whether people can meet the target honestly, we put two more questions to a leadership team before a target is announced. What choice will this measure inform, and who makes that choice? What is the easiest way to achieve the target without producing the benefit it was meant to deliver? If that route costs less than producing the benefit, someone among the people paid for the number will find it, and the organization will pay without getting the benefit.
Each target also needs a paired measure that tracks the harm the target could cause. A count of tickets closed can be reported next to a count of repeat calls about the same problem within a week, and accounts opened next to accounts customers still use after ninety days and complaints about products nobody requested. Wells Fargo had a measure of this kind: its Community Bank tracked the share of new accounts that customers funded, which it called the Rolling Funding Rate, and the rate fell from about 90 percent in 2005 to below 80 percent in 2012 while the sales goals kept rising. A paired measure can be fooled too, because continued use cannot prove that a customer agreed to an account, and the transfers that made unauthorized accounts look funded could also lift a funding measure. A paired measure still shows leaders where harm may be happening, and someone then has to look. Paired measures stay out of anyone's pay, because a measure that decides pay comes under the pressures Goodhart and Campbell described.
A target that sets an amount and says nothing about method leaves the method to whoever is under the most pressure. Leaders also own the decision rights around a target. Before launch, name who watches the paired measure, who hears first when people at the front line say the target cannot be met honestly, and who may change or pause the target and its rewards, because a warning changes little until someone able to act receives it. The person who may pause a target should not be paid on it. At Wells Fargo the leaders who received the warnings were paid by the same sales system, and they kept the goals.
In our experience, once a paired measure, or any measure meant to help a team learn, starts to decide individual pay, people report it less honestly. We keep learning measures with the team that uses them and never let them decide pay, and we review each target as we would any design, by watching what people do under it.
When a target becomes the job, people do the job the target describes, and the work it was meant to improve can go undone. Keeping the target in the service of that work is the leader's job: choosing what to count, setting a target the work can meet, attaching rewards that leave people's reasons intact, and changing any of the three once it starts to reward harm.
The workshop Strategy Design and Its Management Systems takes a leadership team through choosing the measures and targets its strategy depends on, and through finding what each one will reward. Request a briefing.
Questions leaders ask about perverse incentives and performance targets
What causes perverse incentives in organizations?
Perverse incentives in organizations start with what leaders choose to measure and grow when rewards attach to it. Measures of busyness, utilization or output, and scorecards with lagging indicators and no leading ones, reinforce the existing operating model. Once pay or job security depends on a target, some people produce the number without the result, a payment can replace a reason people already had, and a target beyond honest reach forces a choice between the job and integrity.
How can leaders check a target before they set it?
Ask whether people can meet the target honestly with the work as it runs today, what choice the measure will inform and who makes it, and what the easiest way is to achieve the target without producing its intended benefit. Then pair the target with a measure that tracks the harm it could cause, and name who may pause the target and its rewards.
What is the difference between vanity metrics and useful metrics?
A vanity metric is a number that can only go up, such as total registered users or cumulative downloads. Because it never shows a change that calls for a response, it cannot tell a team whether its last action worked, and so it cannot guide a change in behavior. Useful metrics are actionable, showing a team which of its actions changed the number, accessible to the people whose work they describe, and auditable back to customers and transactions.
References (21)
- W. Edwards Deming, Out of the Crisis, MIT Center for Advanced Engineering Study, 1986 (first published in 1982 as Quality, Productivity, and Competitive Position). Quotas and numerical goals set for others tend to produce the opposite of what was sought, and a goal the system cannot deliver goes unmet.
- Independent Directors of the Board of Wells Fargo & Company, Sales Practices Investigation Report, 10 April 2017. https://s3.documentcloud.org/documents/3549312/Board-Report.pdf
- United States Securities and Exchange Commission, order instituting cease-and-desist proceedings against Wells Fargo & Company, 21 February 2020. https://www.sec.gov/Archives/edgar/data/72971/000007297120000214/exhibit994secorder.htm
- Steven Kerr, "On the Folly of Rewarding A, While Hoping for B," Academy of Management Journal, vol. 18, no. 4, 1975, pages 769 to 783. Organizations that reward one behavior while hoping for another.
- Eliyahu M. Goldratt, The Haystack Syndrome: Sifting Information Out of the Data Ocean, North River Press, 1990, page 26. People behave according to how they are measured.
- Charles A. E. Goodhart, "Problems of Monetary Management: The U.K. Experience," paper to the Reserve Bank of Australia Conference in Monetary Economics, Sydney, July 1975, published in Papers in Monetary Economics, vol. 1, Reserve Bank of Australia, 1976. A pattern in the data breaks down once authorities lean on it to steer policy.
- Donald T. Campbell, "Assessing the Impact of Planned Social Change," Occasional Paper Series, paper 8, Public Affairs Center, Dartmouth College, December 1976. Reprinted in Evaluation and Program Planning, vol. 2, no. 1, 1979, pages 67 to 90. Indicators used for decisions come under pressure that distorts them and the work they monitor.
- Douglas W. Hubbard, How to Measure Anything: Finding the Value of "Intangibles" in Business, John Wiley and Sons, 2007. Measurement as a reduction of uncertainty based on observation, and what makes a measurement matter: whether it can change a decision.
- Eric Ries, The Lean Startup: How Today's Entrepreneurs Use Continuous Innovation to Create Radically Successful Businesses, Crown Business, 2011, chapter 7, "Measure." Vanity metrics, and actionable, accessible and auditable metrics.
- Uri Gneezy and Aldo Rustichini, "A Fine Is a Price," Journal of Legal Studies, vol. 29, no. 1, 2000, pages 1 to 17. Day-care centers in Haifa where a fine for late pickup increased lateness, which stayed high after the fine was removed. https://ideas.repec.org/a/ucp/jlstud/v29y2000i1p1-17.html
- Horst Siebert, Der Kobra-Effekt: Wie man Irrwege der Wirtschaftspolitik vermeidet, Deutsche Verlags-Anstalt, 2001. The term cobra effect.
- Kurt Lewin, Principles of Topological Psychology, translated by Fritz Heider and Grace M. Heider, McGraw-Hill, 1936. Behavior as a function of the person and the environment. https://archive.org/details/principlesoftopo00lewi
- Andrew S. Grove, High Output Management, Random House, 1983. Each indicator matched with a second one that watches for its side effect.
- United States Department of Justice, "Wells Fargo Agrees to Pay $3 Billion to Resolve Criminal and Civil Investigations into Sales Practices Involving the Opening of Millions of Accounts without Customer Authorization," press release, 21 February 2020. https://www.justice.gov/archives/opa/pr/wells-fargo-agrees-pay-3-billion-resolve-criminal-and-civil-investigations-sales-practices
- Office of the Comptroller of the Currency, "OCC Issues Notice of Charges Against Five Former Senior Wells Fargo Bank Executives, Announces Settlement With Others," news release 2020-6, 23 January 2020. https://occ.gov/news-issuances/news-releases/2020/nr-occ-2020-6.html
- Consumer Financial Protection Bureau, "Consumer Financial Protection Bureau Fines Wells Fargo $100 Million for Widespread Illegal Practice of Secretly Opening Unauthorized Accounts," press release, 8 September 2016. https://www.consumerfinance.gov/about-us/newsroom/consumer-financial-protection-bureau-fines-wells-fargo-100-million-widespread-illegal-practice-secretly-opening-unauthorized-accounts/
- Richard Cordray, "Written Testimony of Richard Cordray, Director, CFPB, Before the Senate Committee on Banking, Housing, and Urban Affairs," Consumer Financial Protection Bureau, 20 September 2016. https://www.consumerfinance.gov/archive/newsroom/written-testimony-richard-cordray-director-cfpb-senate-committee-banking-housing-and-urban-affairs/
- United States Senate, Committee on Banking, Housing, and Urban Affairs, An Examination of Wells Fargo's Unauthorized Accounts and the Regulatory Response, hearing, 20 September 2016. Senators Shelby and Warren quote Stumpf's 2010 letter to shareholders on the goal of eight. https://www.govinfo.gov/content/pkg/CHRG-114shrg23001/pdf/CHRG-114shrg23001.pdf
- Fortune, report of 14 September 2016, reporting Stumpf's interview with the Wall Street Journal. https://fortune.com/2016/09/14/wells-fargo-ceo-vast-majority-employees/
- Wells Fargo & Company, 2011 Annual Report to Stockholders, exhibit 13 to Form 10-K, 2012. https://www.sec.gov/Archives/edgar/data/72971/000119312512084528/d280360dex13.htm
- Associated Press, "Wells Fargo claws back $75 million from top execs in sales scandal," 10 April 2017, via Minnesota Public Radio. https://www.mprnews.org/story/2017/04/10/wells-fargo-board-faults-aggressive-culture-sales-scandal

