UNJUST DESERTS? - Poppendieck

DESERTS?UNJUST

www.stickyminds.com JULY/AUGUST 2004 BETTER SOFTWARE 33

Manage People & Projects

THE TEAM HAD DONE AN INCREDIBLE JOB, AND THEY KNEW IT. ITERATION

by iteration they had built a new software product, and, when the deadlinecame, everything that had to be operational was working flawlessly. At anafternoon celebration, the division vice president thanked everyone who hadcontributed to the effort, and the team members congratulated each other asthey relived some of the more harrowing moments of the last six months.

JOH

N K

LE

BE

R

A job well done deserves aproper reward, right? Butin collaborative efforts such asagile development, can you singleout individual efforts withoutkilling the team’s morale andproductivity? When it comes timeto divvy up the pie, canjustice ever be served? Or willyour team members be left with

by Mary Poppendieck

The MorningAfterThe next day, the team’s Scrum Masterwas catching up on long-ignored emailwhen Dave, the development manager,called. “Say, Sue,” he said, “great jobyour team did! I’ve been waiting for theproduct launch before I bothered youwith this, but the appraisal deadline isnext week. I need your evaluation ofeach team member. And if you could, I’dlike you to rank the team from who con-tributed the most down to who con-tributed the least.”

Sue could almost hear the air escapingas her world deflated. “I can’t do that,”she said. “Everyone pitched in 100 per-cent. We could not have done it other-wise. In fact, collaboration is at the coreof our Agile process.”

“But Sue,” Dave said, “there musthave been a most valuable player, a run-ner-up, and so on.”

“No, not really,” Sue replied. “Butwhat I can do is evaluate everyone’s con-tribution to the effort.”

Sue filled out an appraisal input formfor each team member. She rated every-one’s performance but found that shehad to check the “far exceeded expecta-tions” box for each team member. Afterall, getting out the product on time was aspectacular feat, one that far exceededeveryone’s expectations.

TheAftershocksTwo days later, Sue got a call from Janicein human resources. “Sue,” she said,“great job your team did! And thanks forfilling out those appraisal input forms.But really, you can’t give everyone a toprating. Your average rating should be‘meets expectations.’ You can only haveone or two people who ‘far exceeded ex-pectations.’ Oh, and by the way, sinceyou didn’t rank the team members, wouldyou please plan on coming to our rankingmeeting next week? We are going to needyour input on that. After all, at this com-pany we pay for performance, and weneed to evaluate everyone carefully sothat our fairness cannot be questioned.”

Sue felt like a flat tire. In the past,when she had a particularly difficultproblem, she had always consulted theteam, and they had always come up withcreative solutions; so she decided to con-sult them once again. She thought shemight convince them to elect an MVP ortwo, to help her put some variation intothe evaluations.

The next morning, the entire team lis-tened as Sue explained her problem. Suewas disappointed and surprised when

after hearing herdilemma, insteadof jumping in tohelp solve theproblem, the teammembers deflatedjust as quickly as

she had. The best they could do was in-sist that everyone had given 200 percenteffort, that they had all helped each oth-er, and that they had thought that everysingle person had done a truly outstand-ing job. They were not interested in elect-ing a most valuable player, but they werewilling to choose a least valuable player:the unnamed manager who was askingSue to choose among them.

Now Sue really had a problem. Shehad no idea how to respond to Dave andJanice, and her plan to involve the teamhad only succeeded in making them angryand suspicious. Tomorrow, they wouldhave to start working together on the nextrelease. How could something that wassupposed to boost performance do such athorough job of crushing the team’s spirit?

Sue is not the only one who has hadtrouble with merit pay evaluation andranking systems. One of the greatestthought leaders of the twentieth century,W. Edwards Deming, wrote that immea-surable damage is created by ranking sys-tems, merit raises, and incentive pay. (Seethis issue’s StickyNotes for more details.)Deming believed that every business is asystem, and the performance of individu-als is largely the result of the way the sys-tem operates. In his view, the system caus-es 80 percent of the problems in a

business, and the system is management’sresponsibility. He wrote that using exhor-tations and incentives to get individualsto solve management problems simplydoesn’t work. Deming opposed rankingbecause it destroys pride in workman-ship, and he opposed merit raises becausethey address the symptoms, rather thanthe causes, of problems.

It’s a bit difficult to take Deming atface value on this; after all, companieshave been using merit pay systems for

decades, and their use is increasing.Moreover, Deming was mainly involvedin manufacturing, so possibly his think-ing does not apply directly to knowledgework like software development. Still,someone as wise as Deming is not to beignored; so let’s take a deeper look intoemployee evaluation and reward systemsand explore what causes them to becomedysfunctional.

DYSFUNCTION #1:

CompetitionAs Sue’s team instinctively realized, rank-ing people for merit raises pits individualemployees against each other and strong-ly discourages collaboration, a corner-stone of Agile practices. Even when therankings are not made public, the factthat they happen does not remain a se-cret. Sometimes ranking systems are usedas a basis for dismissing the lowest per-formers, making the practice even morethreatening. When team members are incompetition with each other for theirlivelihood, teamwork quickly evaporates.

Competition between teams, ratherthan individuals, may seem like a goodidea, but it can be equally damaging.Once, I worked in a division in whichthere were two separate teams developingsoftware products that were targeting sim-ilar markets. The members of the teamthat attracted the larger market sharewere likely to have more secure jobs andenhanced career opportunities. So, eachteam expanded the capability of its prod-uct to attract a broader market. The teams

34 BETTER SOFTWARE JULY/AUGUST 2004 www.stickyminds.com


SHE HAD TO CHECK THE “FAR EXCEEDS EXPECTATIONS”

BOX FOR EACH TEAM MEMBER. AFTER ALL, GETTING OUT

THE PRODUCT ON TIME WAS A SPECTACULAR FEAT.

page 35full-page adTech Excel

www.stickyminds.com JULY/AUGUST 2004 BETTER SOFTWARE 37


plained how the whole idea of rankingmade no sense for a team effort, especial-ly in an Agile environment. She explainedhow she had asked for advice from theteam and ended up with an angry andsuspicious team.

“You should never have talked to theteam about this,” said Janice.

“Hold on a minute,” Wayne jumpedin. “I thought our goal in this company isto be fair. How can we keep our evalua-tion policies secret and expect people toconsider them fair? It doesn’t matter if

we think they are fair; it matters if em-ployees think they are fair. If we think wecan keep what we are doing a secret,we’re kidding ourselves. We need to betransparent about how we operate; wecan’t make decisions behind closed doorsand then try to tell people, ‘Don’t worry,we’re being fair.’”

Sue was amazed at how fast the na-ture of the discussion changed afterWayne jumped to her defense. Apparent-ly, she wasn’t the only one who thoughtthis ranking business was a bad idea.Everyone agreed that Sue’s team haddone an excellent job, and the new prod-uct was key to their business. No one hadthought that it could be done, and indeedthe team as a whole had far exceededeveryone’s expectations. It became ap-parent that there wasn’t a person in theroom who was willing to sort out whohad contributed more or less to the ef-fort, so Sue’s top evaluation for everyteam member was accepted. More im-portantly, the group was concerned thata de-motivated team was a serious prob-lem. Eventually, the vice president agreedto go to the next meeting of the team anddiscuss the company’s evaluation poli-cies. Sue was sure that this would go along way to revitalize the team spirit.

Now the management team membershad a problem of their own. They knewthat they had to live within a merit paysystem, but they suspected they needed torethink the way it was implemented. Sincechanges like that don’t happen overnight,they formed a committee to look into var-

ious evaluation and pay systems.The committee started by agreeing

that evaluation systems should not beused to surprise employees with unex-pected feedback about their performance.Performance feedback loops must be farshorter than annu-al, or even quarter-ly, evaluation cy-cles. Appraisals aregood times to re-view and updatedevelopment plans

for an employee, but if this is the onlytime the employees find out how they aredoing, a lot more needs fixing than theappraisal system.

With this disclaimer in mind, thecommittee developed some guidelines fordealing with various forms of differentialpay systems.

GUIDELINE #1:

Make SurethePromotionSystem IsUnassailableIn most organizations, significant salarygains come from promotions that movepeople to a higher salary grade, not frommerit increases. Where promotions arenot available, as is the case for manyteachers, merit pay systems have a ten-dency to become contentious, becausemerit increases are the only way to makemore money. When promotions areavailable, employees tend to ignore themerit pay system and focus on the pro-motion system. Of course, this system ofpromotions tends to encourage people tomove into management as they run outof promotional opportunities in technicalareas. Companies address this problemwith “dual ladders” that offer manage-ment-level pay scales to technical gurus.

The foundation of any promotion sys-tem is a series of job grades, each with asalary range in line with industry stan-dards and regional averages. People mustbe placed correctly in a grade so thattheir skills and responsibilities match thejob requirements of their level. Initialplacements and promotion decisionsshould be made carefully and reviewedby a management team.

Usually, job grades are embedded intitles, and promotions make the new jobgrade public through a new title. A per-son’s job grade is generally consideredpublic information. If employees are fair-ly placed in their job grades and promot-ed only when they are clearly performingat a new job grade, then salary differ-ences based on job grade are generallyperceived to be fair. Thus, a team canhave both senior and junior people, gen-eralists and highly skilled specialists, allmaking different amounts of money. Aslong as the system of determining jobgrades and promotions is transparentand perceived to be fair, this kind of dif-ferential pay is rarely a problem.

The management team at Sue’s compa-ny decided to focus on a promotionprocess that did not use either a rankingor a quota system. Instead, clear promo-tion criteria will be established for eachlevel; when someone has met the criteria,that person will be eligible for promotion.A management committee will revieweach promotion proposal and gain a con-sensus that the promotion criteria havebeen met. This will be similar to existingcommittees that review promotions to fillopen supervisor or management positions.

GUIDELINE #2:

De-emphasizethe MeritPay SystemWhen the primary tool for significantsalary increases is promotion, then it’simportant to focus as much attention aspossible on making sure the promotionsystem is fair. When it comes to the eval-uation system that drives merit pay, it’sbest not to try too hard to sort people



ended up competing fiercely with eachother for the same customer base as wellas for division resources. In the end, bothproducts failed. A single product wouldhave had a much better chance at success.

DYSFUNCTION #2:

ThePerception ofUnfairnessThere is no greater de-motivator than areward system that is perceived to be un-fair. It doesn’t matter if the system is fairor not. If there is a perception of unfair-ness, then those who think that they havebeen treated unfairly will rapidly losetheir motivation.

People perceive unfairness when theymiss out on rewards they think theyshould have shared. What if the vicepresident had given Sue a big reward butnot rewarded the team? Even if Sue hadacknowledged the hard work of her teammembers, they would probably have feltthat she was profiting at their expense.You can be sure that Sue would have hada difficult time generating enthusiasm forwork on the next release, even if the eval-uation issues had not surfaced.

Here’s another scenario: What wouldhave happened if Sue’s team had beenasked out to dinner with the VP and eachmember had been given a good-sizedbonus? The next day the operations peo-ple who worked late nights and week-ends to help get the product out on timewould have found out and felt cheated.The developers who took over mainte-nance tasks so their colleagues couldwork full time on the product also wouldhave felt slighted. Other teams mighthave felt that they could have been equal-ly successful, except that they got as-signed to the wrong product.

DYSFUNCTION #3:

ThePerception ofImpossibilitySue’s team met its deadline by followingthe Scrum practice of releasing a high-

quality product containing only the high-est-priority functionality. But let’s try a dif-ferent scenario: Let’s assume that the teamwas given a non-negotiable list of featuresthat had to be done by a non-negotiabledeadline, and let’s further speculate thatthe team was 100 percent positive that thedeadline was impossible. (Remember thisis hypothetical; surely this would neverhappen in real life.) Finally, let’s pretendthat the team members were promised abig bonus if they met the deadline.

There are two things that could hap-pen in this scenario. Financial incentivesare powerful motivators, so there is achance that the team might have found away to do the impossible. However, themore likely case is that the promise of abonus that was impossible to achievewould make the team cynical, and theteam would be even less motivated tomeet the deadline than before the incen-tive was offered. When people find man-agement exhorting them to do what isclearly impossible rather than helping tomake the task possible, they are likely tobe insulted by the offer of a reward andgive up without even trying.

DYSFUNCTION #4:

Sub-OptimizationI recently heard of a business owner whooffered testers five dollars for every de-fect they could find in a product about togo into beta release. She thought thiswould encourage the testers to workharder, but the result was quite different.The good working relationship betweendevelopers and testers deteriorated astesters lost their incentive to help devel-opers quickly find and fix defects beforethey propagated into multiple problems.After all, the more problems the testersfound, the more money they made.

When we optimize a part of a chain,we invariably sub-optimize overall per-formance. One of the most obvious ex-amples of sub-optimization is the separa-tion of software development fromsupport and maintenance. If developersare rewarded for meeting a schedule evenif they deliver brittle code without auto-mated test suites or an installationprocess, then support and maintenance

of the system will cost far more than wassaved during development.

DYSFUNCTION #5:

DestroyingIntrinsicMotivationThere are two approaches to giving chil-dren allowances. Theory A says that chil-dren should earn their allowances; moneyis exchanged for work. Theory B saysthat children should contribute to thehousehold without being paid, so al-lowances are not considered exchange forwork. I know one father who was raisedwith Theory B but switched to Theory Afor his children. He put a price on eachjob and paid the children weekly for thejobs they had done. This worked for awhile, but then the kids discovered thatthey could choose among the jobs andavoid doing the ones they disliked. Whenthe children were old enough to earn theirown paychecks, they stopped doinghousehold chores altogether, and the fa-ther found himself mowing the lawnalongside his neighbors’ teenage children.Were he to do it again, this father says hewould not tie allowance to work.

In the same way, once employees getused to receiving financial rewards formeeting goals, they begin to work for therewards, not the intrinsic motivation thatcomes from doing a good job and help-ing their company be successful. Manystudies have shown that extrinsic re-wards like grades and pay will, overtime, destroy the intrinsic reward thatcomes from the work itself.

One WeekLaterSue was nervous as she entered the roomfor the ranking meeting. She had talkedover her problem with her boss, Wayne,and, although he didn’t have any easy so-lutions, he suggested that she present herproblem to the management team. Short-ly after the meeting started, Janice askedSue how she would rank her team mem-bers. Sue took a deep breath, got a smileof encouragement from Wayne, and ex-

WHEN THE PRIMARY TOOL FOR SIGNIFICANT SALARY

INCREASES IS PROMOTION, THEN IT’S IMPORTANT TO

FOCUS AS MUCH ATTENTION AS POSSIBLE ON MAKING



out. Studies show that, when informa-tion sharing and coordination are neces-sary, organizations that reduce pay dif-ferences between the highest- and thelowest-paid employees tend to performbetter over time.

Use evaluations mainly to keep em-ployees at an appropriate level in theirsalary grade. Evaluations might flagthose who are ready for promotion andthose who need attention, but thatshould trigger a separate promotion orcorrective action process. About fourevaluation grades are sufficient, and acompetent supervisor with good evalua-tion criteria and input from appropriatesources can make fair evaluations thataccomplish these purposes.

Even when annual raises are looselycoupled to merit, evaluations will alwaysbe a big deal for employees, so attentionshould be paid to making them fair andbalanced. Over the last decade, balancedscorecards have become popular formanagement evaluations—at least in the-ory. Balanced scorecards ensure that themultiple aspects of a manager’s job all re-ceive attention. A simple version of a bal-anced scorecard also might be used formerit pay evaluations, to emphasize thefact that people must perform well inmany dimensions to be effective. A su-pervisor might develop a scorecard witheach employee that takes into accountteam results, new competencies, leader-ship, and so on. It is important that em-ployees perceive that the input to thescorecard is valid and fairly covers themultiple aspects of their job. It is impor-tant to keep things simple because toomuch complexity will unduly inflate theattention paid to a pay system that worksbetter when it is understated. Finally,scorecards should not be used to feed aranking system.

GUIDELINE #3:

Tie ProfitSharingto EconomicDriversNucor Steel decided to get into the steel-making business in 1968, and thirty

years later it was the biggest steel compa-ny in the United States. When Nucorstarted up, Bethlehem Steel considered ita mere gnat, but thirty-five years later,Bethlehem Steel was not only bankrupt

but sold off forassets. So, NucorSteel is one verysuccessful compa-ny that has done alot of things rightin a tough indus-

try. Quite surprisingly, Nucor has adecades-old tradition of paying for per-formance. How does the company avoidthe dysfunctions of rewards?

Nucor Steel started with the realiza-tion that profit per ton of finished steelwas its key economic driver, and basedits profit-sharing plan on the contribu-tion a team makes to improve this num-ber. So, for example, a team that suc-cessfully develops a new steel-makingprocess or starts up a new plant onschedule will not see an increase in payuntil the process or plant has improvedthe company’s profit per ton of steel.Thus, Nucor avoids sub-optimizationby tying its differential pay system asclosely to the economic driver of itsbusiness as possible.

GUIDELINE # 4:

RewardBasedon Span ofInfluence,Not Span ofControlConventional wisdom says that peopleshould be evaluated based on results thatare under their control. However, evalu-ating individual results, rather than group

results, creates competition rather thancollaboration among the team members.In order to encourage collaboration, Nu-cor makes sure that its profit-sharing for-mula rewards relatively large teams, notjust the individuals or small groups whohave direct responsibility for an area. Ac-cording to their policies and principles, ifa software program creates a significantprofit increase, everyone, including thosewho brought the idea into the company,the developers and testers, the operations

and support people, and the end-users,should share in any reward. This rewardsystem fits particularly well within an Ag-ile environment, which naturally takesthe same broad approach of involvingeveryone (end-users, testers, support peo-ple, etc.) in the development process.

Nucor Steel works hard to create alearning environment where expertsmove from one plant to another, machineoperators play a significant role in select-ing and deploying new technology, andtacit knowledge spreads rapidly through-out the company. Its reward system en-courages knowledge sharing by reward-ing people for influencing the success ofareas they do not control.

How, exactly, can rewards be basedon span of influence rather than span ofcontrol? I recommend a technique called“Measure UP.” No matter how hard youtry to evaluate knowledge work or howgood a scorecard you create, somethingwill go unmeasured. Over time, the un-measured area will be de-emphasized,and problems will arise. We have a ten-dency to add more measurements to thescorecard to draw attention to the ne-glected areas.

However, it is a lot easier to catcheverything that falls between the cracksby reducing the number of measurementsand raising them to a higher level. For in-stance, instead of measuring software de-velopment with cost, schedule, andearned value, try creating a P&L or ROI

CONVENTIONAL WISDOM SAYS THAT PEOPLE SHOULD BE

EVALUATED BASED ON RESULTS THAT ARE UNDER THEIR

CONTROL. HOWEVER, THIS KIND OF EVALUATION CREATES

COMPETITION RATHER THAN COLLABORATION.

(Continued on page 47)

page 39full-page ad

SD Best Practices

sions on behalf of them all? Or where theassigned customer representative lacksthe knowledge, time, or inclination to dothe job? A colleague once told me, “One[customer representative] sat in ourmidst and still managed to avoid us all.”

Stop Debating. StartDelivering.The goal of software process improve-ment is to reduce the cost of building andmaintaining software. Agile is simply an-other approach to process improvement.The issue isn’t Agile versus non-Agile, orAgile versus plan-driven, or people versusprocess, or iterative versus sequential, orlow ceremony versus high ceremony. Theissue is business success. Let’s stop postur-ing (“Yeah? Well, my dad’s more agilethan your dad”) and focus on the real ob-jective of how best to deliver business val-ue to our project stakeholders. {end}

Karl Wiegers, Ph.D., is principal consul-tant with Process Impact in Portland,Oregon, where hespecializes in require-ments engineering,peer reviews, processimprovement, and project management.Karl is the author of Software Require-ments, 2nd Edition, Peer Reviews in Soft-ware, and Creating a Software Engineer-ing Culture. You can reach him at www.processimpact.com.

BETTER SOFTWARE 47

The Last WordManage People & Projects

for the project and helping the team usethese tools to drive tradeoff decisions.

GUIDELINE #5:

Find BetterMotivatorsthan MoneyWhile monetary rewards can be a power-ful driver of behavior, the motivation theyprovide is not sustainable. Once peoplehave an adequate income, motivationcomes from things such as achievement,growth, control over one’s work, recogni-tion, advancement, and a friendly work-ing environment. No matter how goodyour evaluation and reward system maybe, don’t expect it to do much to drivestellar performance over the long term.

In the book Hidden Value, CharlesO’Reilly and Jeffrey Pfeffer present sever-al case studies of companies that obtainsuperb performance from ordinary peo-ple. (See StickyNotes for the complete

reference.) These companies have people-centered values that are aligned with ac-tions at all levels. They invest in people,share information broadly, rely on teams,and emphasize leadership rather thanmanagement. Finally, they do not usemoney as a primary motivator; they em-phasize the intrinsic rewards of fun,growth, teamwork, challenge, and ac-complishment.

Treat monetary rewards like explo-sives because they will have a powerfulimpact whether you intend it or not. Souse them lightly and with caution. Theycan get you into trouble much faster thanthey can solve your problems. Once yougo down the path of monetary rewards,you may never be able to go back, evenwhen they cease to be effective, as theyinevitably will. Make sure that people arefairly and adequately compensated, and

then move on to more effective ways toimprove performance.

Six MonthsLaterSue’s team is having another celebration.The team members had been surprisedwhen the VP came to their team meetingsix months earlier. But they quickly recov-ered and told her that they each wantedto be the best, they wanted to work withthe best, and they did not appreciate theimplication that some of them were betterthan others. When the VP left, the teamcheered Sue for sticking up for them andthen got down to work with renewed en-thusiasm. Now, two releases later, thecustomers were showing their apprecia-tion with their pocketbooks.

There haven’t been any dramatic payincreases and only the occasional, well-de-served promotion; however, the companyhas expanded its training budget, andteam members have found themselvesmentoring other teams. Sue is ratherproud of them all as she fills out the newlyrevised appraisal input forms, which havemore team-friendly evaluation criteria.This time, Sue is confident that her judg-ment will not be questioned. {end}

Mary Poppendieck, a Cutter Consortiumconsultant, is a seasoned leader in bothoperations and new product develop-ment. With more than 25 years of IT ex-perience, she has led teams implementinglean solutions, ranging from enterprisesupply chain management to digital me-dia, and built one of 3M’s first Just-in-Time lean production systems. Mary iscurrently the president of PoppendieckLLC, located in Minnesota. Her bookLean Software Development: An AgileToolkit, which brings lean productiontechniques to software development,won the Software Development Produc-tivity Award in 2004.

StickyNotes

For more on the following topics go towww.stickyminds.com/bettersoftware

■ References and further reading

(Continued from page 38)UNJUST DESERTS

TREAT MONETARY REWARDS

LIKE EXPLOSIVES BECAUSE

THEY WILL HAVE A POWER-

FUL IMPACT WHETHER YOU

Tell Us WhatYou ThinkTell us what you like—and don’tlike—about this or any otherissue of Better Software. Emailyour comments to [email protected] or faxthem to us at 904-278-4380.Comments may be edited forspace and readability andpublished in a future issue ofBetter Software.

(Continued from page 48)JACK BE AGILE...BUT NOT TOO HASTY

UNJUST DESERTS? - Poppendieck

Documents

Transcript of UNJUST DESERTS? - Poppendieck