Reinventing

performance evaluation
Reinventing
performanceevaluation

DiverseoSAS2016

Executivesummary
Business leaders increasingly recognize their
performance evaluation processes need some
fixes. Performance evaluation1 systems do not
serve their business goals anymore: theyfail to
producefullengagement,appropriatelyevaluate
achievements,helpdesigntherightdevelopment
plansorsetfaircompensation.Crucialintodays
largecorporations,theyfailtoleverageawider
rangeoftalents.
Yet, objectively and effectively evaluating
performance is more needed than ever. As a
result,thereiscurrentlyamajordebatemainlyin
the US about removing numbered performance
ratings and annual performance reviews.
Researchhasshownthateventhosewhoreceive
a good review will be negatively affected by
receivinganumericalassessment:peopledonot
like being labelled. In addition, a single annual
performancereviewismostoftenlessobjective
thanmorefrequentones,forexamplebyproject.
However, frequently touted replacement
solutions often carry significant, sometimes
unexpected,impact.Whilemanagersquiteoften
have apreference forthese solutions and their
appealingsimplicitysuchasaskingWouldyou
liketoworkwiththispersonagain?,howdo
youthenmake,fairly,thistime,thekeydecisions
which used to depend on the rating: bonuses,
trainings, promotions? In many countries, the
absence of ratings increases the legal risk of
askingalowperformertoleave.Moreover,such
new solutions do not always bring the
objectiveness needed to leverage a large and
diversetalentpool.Yettoday,morecompanies
need to integrate new skills and competencies
mainly related to IT and digital, to be better
equipped to make decisions under uncertainty
and develop their business in more remote
countries. Integrating such new competencies
anddifferenceswillthereforerequireatleastan
updateofthecurrent,traditionalprocesses.
Infact,veryfewcompaniesaregoodatobjectively
and effectively evaluating performance. Most
people fail to understand how their brain
processesinformationandmakesdecisions.This

is mainly because our brain operates like a

computer withtwo processors, asdescribed by
DanielKahneman,2002Nobellaureate.System2
is the rational processor, which we know and
control.System1isautomatic:everyonehasit;
no one is aware of it. System1 manages much
more information than System 2 and has more
influence that we think on our behaviors and
decisions. Most decision making processes are
designedtoaddressSystem2only;theydonot
effectively address the mistakes or biases
inducedbySystem1.Thisiscomplexbecausewe
cannotstopit,accessitorcontrolit.
Sometimes, our biases and decision making
processes even conspire to create unexpected
impacts. Such impacts can be surfaced only by
analyzing the process with a cognitive lens. For
example, when processes are too complex,
people get into cognitive overload and
unconsciouslyfocusonlyontheinformationthey
likeandthatconfirmstheirunconsciousbiases.
Whenprocessesaretoolight,theylackthefacts
neededtobeobjective.
In fact, while some key principles are helpful,
thereisnoonesizefitsallprocess,asitmustbe
adaptedtotheimplicitcultureandoperationsof
the company. Some of the current trends in
performance management such as removing
annualperformanceevaluationscomefromfast
paced technologies companies where annual
objectivesettingandperformanceevaluationsdo
not fit with fast changing business goals. While
encouragingmorefrequent,informalandformal
feedback is an excellent practice, completely
removing annual review is not necessarily
needed for more stable businesses. They still
help identifying quick learners. Best in class
processes are therefore significantly different
fromonecompanytothenext.
Insettingupanewprocess,somekeyprinciples
haveprovensuccessful:

Embedactual,businessrelatedperformance
driverstoallowmanagerstobeobjectiveand
best enhance the performance of their

1 Throughout the document, to avoid often unwieldy turns of phrase, we use evaluation to
describe performance evaluation.

DiverseoSAS2016

teams.Thissoundsasanobrainer,butmany
businesses have standard performance
evaluation criteria which do not fit the
companyorspecificjobs.
Sustain a growth mindset: Replace the
absolute judgement of a rating with a
dynamic assessment of the performance
trajectory; numerical ratings would be
replaced by constructive words. Provide
frequentformalandinformalfeedback.
Remain simple: complexity can generate
overload but also waste of time. In our
experience,manystagesanddocumentscan
oftenbeeliminatedwithnonegativeimpact.
Becalibrated:asmanagerstendtoperceive
performance of their direct reports
differentlydependingontheirage,gender
butalsotheirownstrengthsandweaknesses,
setting upfront clear expectations can help
enhanceobjectiveness.
Striveforhomogeneity:peopleinthesame
jobsshouldbeevaluatedhomogenously(i.e.
same frequency, criteria.). It also allows
bestcapturingeveryonescorestrengths.
Beculturallyneutral:differentcultureshave
different ways of perceiving and expressing
the same perception of performance. In a
global company, the process should ensure
thatpeoplefromdifferentcountrieswillbe
fairlyevaluatedsoastoincludeasmuchas
possibleawiderrangeoftalents.
Reduceindividualinferencesandbiaseswith
specificdecisionmakingtechniquessuchas
the 7 Steps: weighting the performance
criteria,hidingthenameofindividualwhen
assessingtheresults.
Fitintotheorganizationstructure:Ongoing
feedbackshouldbenurtured,butcorporate
needs should drive frequency of formal
evaluations,decisionmakerstoinvolve.
Improve continuously: key changes in the
labormarketorincorporatestrategynew
skillsneededshoulddirectlybeembedded
intotheprocess.
Keeplinemanagersownership:Theyarethe
ones who know their team members
performancebestandthereforeshouldbuy
into the new process, which should be
adapted to their understanding, needs and
constraints.

Companies considering a new approach should

structuretheworktoachieveeffectiveresults:
1. Decidewhyyouneedtochangeandwhatyou
needtochange:
Identifywhatyouexpectfromyourprocess
and the key interrelationships with other
businessprocessespeopledevelopment,
bonuses,
Performaquantitativeanalysisofcurrent
key elements such as performance
distributions, differences across social
groups. How many people receive each
rating, do women tend to systematically
receivelowerratings?...
Understand how managers implicitly
assess performance and what changes
theybuyinto.Managersusuallyintuitively
knowwhatmakesagoodperformanceand
have developed habits. It is crucial to
understandwhatdrivestheirbehavior,and
whatneedsandcanbechanged.Besides,
assessingtheirabilitytoprovidedaytoday
feedbackcanprovideaclearperformance
enhancermovingforward.
Surfaceandquantifythebiasesthatneed
to be mitigated. Some biases are
unexpected. Knowing them will allow
addressingthem.
Assess effectiveness, and especially time
spent.Searchforsimplifications.
2. Change only what needs to change, simplify
and adjust to your context. There is no one
sizefitsallprocess.Involvemanagersinpilots,
ensure the new process enhances
objectivenessandleveragesmanagerstime.
3. Effectivelyrollitout:
Involve line managers to set jobrelated
performance criteria and formalize who
shouldbeinvolvedateachstage,
Communicate why you changed the
processandwhatitwillbring,
Trainmanagers,evenifthenewprocessis
simple. Show how to provide ongoing,
formalandinformalfeedback,befairand
engage their teams. The simpler the
process, the more training is needed as
managerswillhavefewerguidelines.

****************

DiverseoSAS2016


Reinventingperformance
management
Evaluation systems often fail to
engage employees and deliver
improved performance, despite
considerableinvestmentintime
andmoney
Evaluatingperformanceisoneofthecoreman
agerial tasks in todays organizations; it helps
driveemployeeengagementandperformance.
Butbusinessesdevotesurprisinglylittletimeto
thinking through what actually constitutes a
good performance evaluation system2. This is
notbecausetheyarehappywiththeirexisting
systems: business leaders increasingly recog
nize that their systems do not properly serve
theirbusinessgoals:theydonotfullyproduce
engagement, appropriately evaluate achieve
ments,helpdesigntherightdevelopmentplans
or set fair compensation. Crucially in todays
large organizations, they fail at leveraging a
wider diversity of talents: in our experience,
performanceratingdependsmoreonsuchfac
tors as professional background, gender, and
ethnicity,ratherthanactualperformance.
In the majority of our recent client projects,
most employees felt that their performance
was indeed assessed subjectively and did not
provide a clear view of what was expected of
them.Thisledtoreducedengagement.Employ
eeswerenottheonlydissatisfiedstakeholders
intheprocess:managersfelttheywastedcon
siderable time and money for unsatisfying re
sults.CEB,aconsultancy,estimatedthatonav
erage, a business with 10,000 employees
spendsabout$35millionayearassessingper
formanceinaprocessthatnotonlyaddslittle
valuetotheirteamsperformance,butresults
inmanydifficultevaluationconversations.

Areformalratingsuseful?
Inpartasaresultofthisfrustration,thereiscur
rently a debate around performance ratings
and standard annual evaluations. In the US
mainly, some organizations are moving away
fromratings.Deloitteforexamplehasreplaced
ratings with four simple questions, such as
Given what I know of this persons perfor
mance,Iwouldalwayswanthimorheronmy
team;orGivenwhatIknowofthispersons
performance and if it were my own money,
would I award this person the highest pay in
creaseandbonus?.
Itistemptingtoditcholdprocessesinfavorof
such a light system. It saves managers signifi
canttimeandenhancestheirshorttermsatis
faction.However,itoftendoescomewithunin
tended consequences. Certainly, ratings are
criticizedforbeingunfairandoverlydependent
onthepersondoingtherating.Butisthealter
native more objective? Without a rating, how
doesonemakefairly,thistimethekeydeci
sions onbonuses, trainings,promotions?How
do managers objectively assess the quality of
the how and what someone achieved? In
case of litigation, how do you justify you de
cided to terminate an employee due to poor
performance?Importantly,someresearchalso
showsthatclearandtransparentperformance
feedbackincreasesengagement. Whereis the
rightbalance?
Inadditiontomovingawayfromratings,some
companies are also letting managers decide
whenandatwhatfrequencytoevaluateperfor
mance.AtGap,forexample,managersareen
couragedtodiscussperformanceeverymonth,
but this is not mandatory. In some organiza
tions,suchfreedomhasledmanagerstoevalu
ate the performance of people they like the
mostoftenwhite,youngmalesonceamonth
whileethnicminorities,women,elderemploy

Throughout the document, to avoid often unwieldy turns of phrase, we use evaluation to
describe performance evaluation.
2

DiverseoSAS2016

ees,areevaluatedonlytwiceayear.Andob
viously, no matter what system is finally se
lected, the job of objectively assessing some
onesperformancestillneedstobedone:itis
necessarytoidentifythebestperformersthat
willreceivehigherbonusesandpayincreases.

Yet objectively and effectively

evaluatingperformanceismore
neededthanever
Beyondthedebateaboutnumericalratingsand
annualreviews,itisclearthatclassicevaluation
frameworks no longer can account for the in
creasingdiversityandcomplexityofthetalent
pool.Theyshouldevolvetoensurethatmanag
erscanmakemoreobjectivedecisions.
Large organizations need to constantly
adapttheiroperationstotheirenvironment,
which often requires integrating new com
petenciesandtechnologies.Intodayseraof
digitaltransformation,hiringandintegrating
highlyneededdatascientists,specializedIT
developers, etc., is often a daunting task
evenforthemostattractiveemployers.De
mandforsuchhighlycompetentpeoplecur
rently exceeds their availability. The best
onesmightnotevencomefromtheorgani
zationshomecountrynorbewillingtorelo
catewherethecompanyneedsthem.They
oftenhavedifferentmindsetsandworkhab
its than current employees, including the
managerswhowillbeinchargeofmotivat
ing, developing, evaluating and retaining
them. Providing both employees and man
agerswithanadaptedappraisalprocessfos
tering objectiveness to bridge the cultural
gaps and potential misunderstandings is a
core requirement to success. Job satisfac
tionanddeliveryforsuchcrucialemployees
isessentialtothecompanyssuccess.
In a more uncertain world, strategic and
technological options are more difficult to
select, and often costlier to explore. Re
searchshowsthattheabilitytointegratedif
ferentviewpointssignificantlyimprovesthe
quality of decision making under such cir
cumstances. So does nurturing creativity
5

andagility.Mostcurrentevaluationsystems
aretypicallygearedtowardsselectinglook
alike and thinkalike individuals. They also
fail to value intellectual agility and seldom
rewardpeoplebringingnewperspectives.
Most systems are calibrated and tested in
countries where the current core business
lies.Andyet,significantgrowthpotentialis
oftenlocatedincountriesthatareculturally
quitedifferent.Asaresult,evaluationscan
beopentointerpretation,ifnotinoperative,
ingrowthareas.

Fewcompaniesaregoodateval
uating because people fail to
take into account how their
brainprocessesinformationand
makesdecisions
Hiringandretainingmorediversepeople,cre
atinganinclusiveandfairworkingenvironment,
assessingeachpersonsperformanceinatruly
factbased manner is therefore more needed
than ever. Yet few organizations achieve this.
Cognitive biases hamper managers ability to
truly select the best, enhance their perfor
manceonadailybasisbutalsoobjectivelyas
sesstheirperformanceandensurethebestwill
bepromoted.Thisisbecausetheyoverlookthe
workings of the most important tool that em
ployeesusetogetresults:theirbrain.

Thetwosystembrain
Recent research in mind sciences has estab
lishedthatourbrainoperateslikeacomputer
withtwoprocessors,whichNobellaureateDan
ielKahnemandescribesasSystem1andSystem
2.
System 2 is the rational system, the one we
knowandcontrol.
System1,ontheotherhand,workslargelyau
tomatically,withouttheindividualbeingaware.
ItprocessesmuchmoreinformationthanSys
tem2:about200,000timesmoreinformation,
accordingtoresearch.Alotofithastodowith
reflexiveactionssuchasbreathing,gettingthe
DiverseoSAS2016

hearttobeat,etc.Butitalsoinfluencesdecision
makinginveryconcreteways.
NomatterwhatSystem2,therationalsystem,
perceives and thinks it decides, System 1, the
automaticone,hasreactedfirstandinfluenced
thedecision:forexample,whenwefirstmeet
someoneorassesstheirreactions,System1as
sociatestheindividualwithattributes,injusta
fewmilliseconds,withoutusevenbeingaware.
Suchunconsciousreactionswillimpacthowwe
perceivesomeonesperformanceandhowwe
rateit.Theautomaticworkingsofthemindcan
notbeoverridden.
System1hasasignificantimpact,yetmosteval
uationprocesses,evenbestinclass,arede
signedasifonlyourrationalSystem2existed.
As a result, they show limited effectiveness in
reducing the widespread unwanted impact of
System1whichissystematicallyassociatingat
tributes often irrelevant, sometimes damag
ingtoindividuals.
Effectively reducing the impact of System 1 is
complex,inpartbecausewecannotstopit:we
havetolivewithit.Weconstantlymakemental
shortcuts,interprettheinformationwereceive
byintegratingourownautomaticassumptions.
But recent progress in understanding these
mental mechanisms allows addressing these
flaws and making sustainable and deep pro
gress.

When biases and decision mak

ingprocessesconspiretocreate
unexpectedimpacts
Decision making processes, including perfor
manceevaluationprocesses,canfurtherexac
erbatetheunintendedinfluenceofouruncon
sciousbrain.
Someprocessesaretoocomplexandgenerate
what is known as cognitive overload, leading
people to revert to intuition. Othersare too
lightanddonotfacilitatefactbasedanalysisof
performance. Some simply trigger biases right
fromthestart:startingaperformanceevalua

tioncalibrationmeetingbyprovidingtheprevi
ousperiodsratingscreatesananchoringbias,
whichisdifficulttoovercome.
Small, invisible cues can sometimes also have
significant consequences depending on the
companysculture.Wehaveseenthatthenum
berofwordsinselfappraisalscanimpacthow
managersperceiveperformance,inwayswhich
aredependentontheculture:atsomeclientor
ganizations,alongselfappraisalunconsciously
leads evaluators to believe the individual is a
highperformerthatachievedalot,whileinoth
ers, it will indicate that he/she is a low per
formerneedingtojustifyhim/herself.
Thusanevaluationprocesswhichworksforone
company could be completely unsuited to an
other.Eachcompanyshouldstartbyidentifying
key business performance drivers and how
these should be integrated in the evaluation
process. But crucially they should also assess
thenhowcultureandexistingprocessesleadto
biases in how people process information in
their minds and ultimately in their decision
making.Onlybyintegratingtheseelementscan
a company create an evaluation process that
enablesobjectivedecisionmaking.

Thereisnoonesizefitsallpro
cess. It must be adapted to the
implicit culture and the opera
tionsofthecompany
Objectiveness and effectiveness of an evalua
tionprocessaremainlyderivedfromthecom
binationof:
the corporate culture, and how it fosters
opennessandfairness;
theprocessitselfwithitsstructure,criteria,
onlinetools,documentsandspecificrespon
sibilitiesassignedtodecisionmakers.
People often equate formality and complete
ness with objectivity. The more complete and
formal the process, goes the argument, the
moreobjectiveandscientifictheresult.Thisis
not always true. Highly structured decision
DiverseoSAS2016

makingprocessesaresometimessocomplexto
navigatethattheygeneratecognitiveoverload.
Imposingexternallydesignedcompetencegrids
andprocesses,whichdonotcorrespondtohow
people actually evaluate performance, only
compoundstheissue:itoftenleadsmanagers
tofirstintuitivelydecidewhatperformanceas
sessmentsomeonedeservesandthenbacken
gineerallthecontentsoftheevaluationtojus
tifythedesiredresult.Inotherwords,theycan
unintentionally increase biases. They are also
extremelycostlyasmanagersneedtoallocate
toomuchtimetoadministertheprocess.
The best processes are tailored to the com
panyscultureandbusinessneeds.Someorgan
izations can achieve highly objective evalua
tionswitharelativelylightprocessthanksto
anopenculture.Othersneedamorestructured
processwithmoreguidanceprovidedtoman
agerstogetthesameobjectivity.Itistherefore
necessary,tofosterobjectivenessandenhance
performance, that the evaluation process re
flects the organizations specific performance
drivers, culture and best fit into the organiza
tionstructure.
Thisisallthemoreimportantforverysuccessful
companieswithastrongculture.Theirprocess
need to capture the core DNA of companys
performancedrivers,whileallowingtheappro
priate degree of differentiation to account for
functionsandbusinessunitsidiosyncraticsuc
cessfactors.Itshouldthenbestallowdecision
makers to assess those performance drivers
fairly across different groups of people. While
there is no standard system, there are best
practice principles in system design, with
proven effectiveness to generate better out
comes
Reflect actual performance, as it is, not as the
grid tells you. As mentioned earlier, the best
performanceevaluationprocessesarebasedon
performance criteria which truly reflect actual
performance.Manyorganizationsareimposing
external, standard competencies grids. These
oftendonotfitwithhowmanagersimplicitlyas
sess performance. They intuitively know what
makesadifference,andareequippedtoassess
7

performanceaccordingtothejobsactualper
formance drivers: on this job, what really
makes a difference? To provide a useful and
unbiased evaluation, they should also be en
couragedtobecurious,openandobservewhat
peopledo.
Create a growth mindset. The latest neurosci
enceresearchindicatesthattheveryfactofgiv
ing a rating creates disengagement, even for
thoseemployeeswhoreceiveapositiverating.
Numericalperformanceratingstendtoreduce
peoples future performance: with a rating,
peoplefeellabelled,whichisgenerallyfrustrat
ing.Whenpeopleareratedaverageorbelow,
this unconsciously anchors in their mind this
lowerabilitytoachieve,whichunderminesfu
ture performance. Research also shows that
peopleneedclearandtransparentfeedbackto
beengaged.Therefore,givingpeopleadynamic
assessmentoftheirperformanceandshowing
them how to grow enhances their future per
formance. In other words, organizations need
tocreateagrowthmindset,inwhichfeedback
movesfromyouhavenotmasteredthiscom
petencyandthereforeyoudeservea2ratingto
youshouldbettermasterthiscompetency,and
here is how you can achieve this in the next 6
months. Such a dynamic assessment still as
sesses how the employee performs, but re
ducesanchoringbiasinmanagement.Asenior
leader at a bank recently told us he gave the
lowestratingtoabankeronhisteamtwoyears
ago.Sincethen,thisbankerhassignificantlyim
provedhisperformance.Yet,heiscurrentlystill
not considered for promotion while he de
serves it, and his manager supports him. All
otherseniorexecutiveshavelabelledhimasa
poorperformerandcannotchangetheirmind
sets.
Anotherkeyelementtocreateagrowthmind
set is to provide frequent, ongoing feedback,
whichshouldbebothformalandinformal.Such
frequentfeedbackcanbeembeddeddirectlyin
the performance evaluation process, but spe
cialattentionneedstobedrawntomakesure
that such ongoing, frequent feedback is con
structiveanddoesnotundulyincreasestress.

DiverseoSAS2016

Simple.Toimprovequalityofperformanceas
sessment, organizations have quite often cre
ated sophisticated and complex performance
evaluation systems. In fact, too much infor
mationi.e.manycompetencecriteria,longand
open supporting documents and guidelines,
crowded performance evaluation IT screens,
not only are timeconsuming but also actually
reduce objectiveness. For most people, the
brain cannot manage more than 6 evaluation
criteria at a time. Therefore, when decision
makersneedtonavigateaframeworkwith10
competenciesandmore,theywillpicktheones
that support the outcome that they uncon
sciouslyprefer.Womenwillbemorediscussed
for their collaborative skills and men for their
leadership.Womenwillthenbemoredirected
tocoordinationandsupportingroleswhilemen
willbeencouragedtotakeleadership,lineman
agementroles.Inotherwords,toomuchdata
killsdecisionmakingquality.
Calibrated.System1leadsustoperceivediffer
entlytheperformanceoftwopeoplewhode
liverthesameresults.Dependingonwhothey
are and where they come from, we will inter
pretdifferentlythequalityoftheirresults:ah
yes,buthefacedsuchatoughmarketenviron
ment.Managersshouldsetandcommuni
cate upfront clear expectations for the job,
openlyandtransparently update those expec
tationsasneeded.Theyshouldthenworkwith
therightlevelofcoldandhomogeneousfacts
to assess how their expectations were
achieved.Suchfactscanserveasabasefordis
cussion for talent reviews where several deci
sionmakerscanchallengeeachother,thereby
fosteringcalibratedevaluationsandpromoting
fairness.
Homogeneous. Employees in a similar position
shouldbehomogeneouslyevaluatedi.e.same
criteria,samefrequencyoffeedbackandevalu
ation, including across different geographies.
Mostofthevariationonratingsdependsonthe
raters and not on the performance of the
ratees.Specifictoolsthereforeshouldmitigate
thebiasintheperformanceperceptionofthe
evaluator, even in companies that have re
movedperformanceratings.Suchhomogeneity
8

intheprocessdoesthereforenotimplythatall
individuals are expected to have the same
strengths. It ensures that everyones core
strengths, no matter how different, are fairly
captured.
Culturallyneutral.Differentcultureshavediffer
ent perceptions of performance and different
waysofexpressingit.Inourglobalprojects,Chi
nese people tend to have a more modest ex
pressionoftheirachievementsthanAmericans,
and certainly tend to be less assertive when
communicatingtheirperformanceperceptions.
Especially for companies that develop them
selvesinternationally,culturalneutralityshould
beagoal.
For example, at one global industrial leader
where leadership is a key performance crite
rion, executive committee members implicitly
equateditwithextraversionandassertiveness.
ThusChinesemanagerstendedtoreceivelower
performanceratingsthantheirWesterncoun
terparts,evenmoresowhentheyreportedto
Westernexecutives.Weworkedwiththatclient
tocreateafactbaseddefinitionwhichallowed
allleaders,nomattertheirculture,tobestun
derstandwhatwasexpectedofthemandben
efit from a fair evaluation of their leadership
skills:aleaderissomeonewhosystematically
improves the market position of a business
he/she will be responsible for while ensuring
large scale followership as measured by 360
feedbacks. Such a definition might not fit all
companies, but our clients senior executives
andtheirdirectreportsallacceptedit,nomat
terwheretheywerebasedorcamefrom.Asa
result,Chinesehighperformersreceivedpraise
fortheirleadershipandweremoreengaged.
Reduce individual mental inferences with tech
niquessuchasthe7steps:somedecisionmak
ing techniques can reduce individual evalua
torsunconsciousbiases.Forexample,directing
managerstoexplicitlywritedownperformance
criteria, weight them according to their im
portance for the job, assess the quality of the
resultsachievedforeachcriterion,whilehiding

DiverseoSAS2016

thenameoftheevaluatedpersontheanony
mizationwesuggestistargeted,enhancesde
cisionmakingquality.
Best fit into the organization structure. When
movingtoanewsystem,businessesshouldtake
thetimetocustomizeittotheiroperationsand
constraints. Many companies right now are
thinking of emulating Deloitte and Accenture,
companieswhicharediscontinuingannualtheir
evaluation conversations and process. In our
opinion,botharerighttodoso.Whenanalyzing
the effectiveness of the performance evalua
tion process of one of the Big 4 accounting
firms, we found that the evaluations given at
theendofeachprojectweresignificantlymore
objectivethanannualevaluations.Thesewere
basedonasynthesisofthevariousprojectper
formanceassessments,andthenrevieweddur
ingevaluationmeetingsbypeoplewhohadno
directexperiencewiththeindividualdiscussed.
The end result was often disconnected from
project managers, teams and clients feed
backs.ButwhatmakessenseforDeloitteorAc
centuremightnotmakesenseforothercompa
nies.Soitisbesttofirstdeterminewhatprocess
andfrequencywouldbestfityourneeds,before
thinkingofcopyingneworwidelyhypedsys
tems.Legalconstraintsalsoplayarole:insome
countries, formal annual performance discus
sionsareanobligation.
EmbedcontinuousimprovementsBestinclass
companies usually update the performance
process as needed, mainly by adapting to
changesinthelabormarketorcorporatestrat
egy. Such ongoing updates should ideally be
limited in scope and/or depth to ensure that
managers will best know how to use the up
datedprocessandtrustit.Otherwise,ifthesys
tem evolves too often, managers would per
ceiveitasinconsistent,unstable,andlesseffi
cientthantheinformalonestheycoulddevelop
bythemselves.
Owned by line managers, supported by HR. Di
rectmanagersknowbesttheirteammembers
and peers performance. Their ownership and
trustinthesystemwillactuallydriveitsoverall
impact.Bestinclassprocessesthereforefirstfit
9

with managers skills and operational con

straintsandcapturetheirbuyin.
Generally speaking, a cumbersome process
with many documents and significant online
worktendstoreduceoverallmanagerialbuyin
andwillingnesstogetpositivelyinvolved.Itre
sultsinheightenedtemptationtodefendones
candidate and reinforces decision making bi
ases: managers become emotionally involved
instead of conducting an open and factbased
analysis.Bestinclasscorporationshavethere
foreautomatedfactgatheringanddatareport
ingasmuchaspossiblewithuserfriendlysys
tems.Thisfreesupmanagerstofocusonvalue
added tasks such as discussing actual perfor
mance.

Howtogetthere?
Companiesconsideringanewprocessshould
structure their approach to get beyond
lengthy discussionand achieve effectivere
sults:
First,haveaclearideaofwhythecompany
needstochangeandwhatneedstobefixed

Clearlyidentifywhatyouexpectfromyour
performance evaluation process, along
with the key interrelationships with other
business processes development, bonus
setting, Over time, business processes
growincomplexitywhilesomefeaturesare
not needed. For example, research has
shownthatactualperformancetendstobe
splitonapowerbellcurve.Veryfewpeople
have either a poor or an excellent perfor
mance. The small number of outstanding
performersdeliversaveryhighpercentage
of the total performance. Therefore, one
objective of the performance evaluation
system can be to identify and retain the
outstanding performers while driving oth
ers to become outstanding performers.
Witha5ratingsscale,averyhighpercent
ageofthepopulationwillbeawardedrat
ings of 2 to 4. Therefore, investing signifi
cant time to differentiate people across

DiverseoSAS2016

10

those three ratings and distributing com

pensationacrossthesecanbeawasteofre
sources.
Performaquantitativeanalysistoidentifyis
sues nonobjective processes usually im
pacthowdifferentsocialgroupsincluding
people with different professional back
grounds and skillsets are developed and
promotedinsidetheorganization.Bylever
aginginternaldataitispossibletofindout
whether different genders, ethnic groups,
degrees etc.. are indeed evaluated differ
ently.
Analyzehowburdensometheprocessis.This
involves getting core performance evalua
tionKPIs:overalltimespentontheprocess,
percentageofpeoplewhohaveacomplete
evaluation, potential differences in ratings
comingfromtheraterhimself/herself,how
understandwhattheyneedtoimprove
Identifyhowmanagersreallyassessperfor
mance. Managers intuitively know what
makes a high performance. They also fre
quentlydiscusstheirteammembersperfor
mancewiththeirbossesandpeers.Real,im
portantdecisionsareoftenimplicitlymade
moreinsuchdiscussionsthanwithinformal
processes. In fact, when formal decision
makingprocessesareexternallyimposedto
managers,theyrarelyapplythem.Addition
ally,implicitcriteriaareoftenquitedifferent
fromtheexplicitandformalones.Suchim
plicitcriteriaareoftenkeybusinessperfor
mancedriverswhichshouldbecapturedby
the process. Uncovering the informal pro
cessesandunderstandinghowdecisionsare
actually made will allow setting a process
that fits with managers needs while giving
themthetoolstoenhancetheobjectivityof
the decisions they make. Such an updated
processwillbemoreeffective,morenatural,
andbetteraccepted.
Identify biases in decision making. This is
not straightforward as biases are uncon
scious,andsometimesuniquetotheorgan
ization. Checking biases off a preestab
lishedlistisnotenough.Consciouslytrying
tocounterthemwithoutverifyingthatthey

existisjustasdangerous.Atoneofourcli
ents, people were so convinced that they
hadaproHarvardbiasthatmanagersac
tually systematically downgraded Harvard
MBAgraduates.TheHarvardMBAbiasrat
ingdowngradedoesnotappearonusual
listsofcognitivebiases.Touncoverspecific
biases, we interview decision makers; we
usequestionsthatarecarefullydesignedto
beneutral,tounderstandhowtheyauto
matically think, we review individual per
formance evaluation files, and look
whetherbiasescanbetrackedinthedata.
Wealsolookathowdecisionmakersauto
matically decipher information on docu
ments and identify what needs to be
changed to allow them to enhance the
qualityoftheirdecisions.

Second,focusspecificallyonwhatneedsto
bechanged,simplifyandadjusttoyouror
ganization. Pilot key elements before large
scalerollout.
Overthelastfewweeks,severalofourclients
haveaskedus:shouldIdiscontinueratingsand
annualperformancereviews?Perhaps,wean
swer,buttherealquestionshouldbe:howcan
Ibestdesignanobjective,growthorientedand
effectiveperformanceevaluationprocess?
Thiswillcertainlyrequireembeddingcorebusi
nessperformancedriversinperformanceeval
uation, and enhancing factbased, deductive
reasoning, reducing inferences and uncon
sciousbiasesforeachandeverymanager.Itwill
certainlyimplycloser,probablymorefrequent
andlighterconversationswithdirectreports.
Insucharedesign,specificattentionneedsto
bebroughttotheimpactofthecontemplated
processonunconsciousbiasessomethingthat
hyped processes often lack. Importantly,
thoseelementsofthecurrentprocessthatac
tually are effective and foster objectiveness,
shouldbekeptasis.
Special attention also needs to be brought to
theselectionortotheevolutionoftheITsys
tem, as a poor choice can annihilate the pro
jectsimpact.
DiverseoSAS2016

Last,ensuringbuyinandsupportfromdecision
makersthroughouttheredesignphaseisakey
success factor. Piloting some key elements to
checkhowtheyareperceivedandwhetherthey
fitthecontextishelpful.Settingtherightpro
ject team and process, involving influencers
along the way to make sure they take owner
shipfortheupdatedprocessareothersuccess
pillars.

Third,effectivelyrolloutthenewprocess.
Depending on the redesign, the implementa
tionandrolloutoftheadaptedprocesswillmo
bilizedifferentresources.Infact,therolloutof
the updated performance process always re
quiressomepracticalcompromises.Inourex
perience, the previous old habits tend to
creepbackandreduceimpact.Bestinclassim
plementationgenerallyincludesthreepillars:

Invitelinemanagerstocontributebysetting
factbasedperformancecriteriaforspecific
jobswithintheirteams.WorkshopswithHR
andlinemanagers,targetingkeystructural
features, can also help identify the deci
sions makers that need to be involved at
each stage. Workshops can also be con
ducted with decision makers to calibrate
criteriawithrelevantandcompellingfacts.
Communicate the need for a new process
andwhatitwillbring.Useinternalsocialnet
works. Mobilize opinion leaders to acceler
atechange.

Trainmanagerstomakethebestuseofthe
newprocess,evenwhenitisverysimple.Ex
plainwhychangewasimportant.Showman
agershowtobefair,objectiveinassessing
performanceandreducedecisionmakingbi
ases.Beyondtheformalperformanceevalu
ation system, encourage managers to pro
vide ongoing, descriptive and daily feed
back. Get them to exercise how to deliver
growthorientedfeedbackandenhanceday
to day performance; formal weekly check
and priorities setting for team members
haveshownpositiveimpact.Dailyniceand
positive conversations too. These seem
basic but are often neglected and do en
hance quality of work environment and
overallperformance.

Toknowmoreabouthowtobestenhancequal
ityofperformance:
NathalieMalige,CEODiverseoSAS
Nathalie.malige@diverseo.com

MartinSchoeller,COO,DiverseoSAS
Martin.schoeller@diverseo.com

Biography
Reportsofthedeathofperformancereviewsareexaggerated;Themeasureofaman;TheEconomist,Schumpeter;Feb20th2016|Fromtheprintedition
MicrosoftAbolishesEmployeeEvaluationSystem,TheNewYorkTimes,ByNickWingfied,November13,2013
WhyGEhadtokillitsannualperformancereviewsaftermorethanthreedecades,Quartz,MaxNisen August13,2015
ThePushAgainstPerformanceReviews,JULY24,2015,byVauhini Vara,TheNewYorker
PerformanceManagementCanBeFixed:AnOntheJobExperiential LearningApproach forComplex Behavior Change,ElaineD.Pulakos andRoseMuellerHanson,PDRIACEB
Company,Arlington,Virginia;SharonArad,Cargill,Minneapolis,Minnesota;Neta Moye,PDRIACEBCompany,Arlington,Virginia;IndustrialandOrganizationalPsychology,8(1),
pp5176March2015.
InformalandFormalPerformanceManagement:BothAreNeeded.IndustrialandOrganizational Psychology;RobertL.Cardy,2015.
Does raterpersonality matter?Ametaanalysis ofraterBig Fiveperformanceratingrelationships;MichaelB.Harari;Cort W.RudolphandAndrewJ.Laginess;Journalof
OccupationalandOrganizationalPsychology,2015
ForcedDistributionRatingSystems:WhenDoesRankandYankLeadtoAdverseImpact?GaryW.Giumetti; AmberN.Schroeder;FredS.Switzer,III;JournalofApplied Psychology,
2015
Thebestandtherest:revisitingthenormofnormalityofindividualperformance;ErnestOBoyleJr,DepartmentofManagement,CollegeofBusinessandEconomics,Longwood
University;HermanAguinis,DepartmentofManagementandEntrepreneurship,KelleySchoolofBusiness,IndianaUniversity;Personnelpsychology2012
TheLeadershipDecoder;Diverseo;October2014

11

DiverseoSAS2016

1. >

12

DiverseoSAS2016