97% of users churn silently — here's why - Mixpanel

Finally, users churn when product teams incorrectly assume that they know what makes users stay. Without the granular data that analytics ...   Ifyouencounterissues,pleasedisableyouradblocker. Thankyouforcontactingus!We’llbeintouchshortly. Therewasanerrorwithsubmittingtheform. ProductFoundations 97%ofuserschurnsilently—here’swhy Lastedited:Aug22,2022 ChristopherGillespie Writer@Mixpanel Whoareyourapp’susers?They’repeoplejustlikeyouandme.Theyhavebusylives,friends,dailycommutes,andwhenconfrontedwithchallenges,theyprefertoavoidconfrontation.It’sonlynatural.AccordingtoresearchbythinkJar,that’swhy25outof26customerschurnsilentlywithoutcomplaining. Asaproductmanager,thisisawfulnews.Witheachofthesedepartingusers,youloseatreasuretroveofunvoicedfeedback.Whydon’ttheyspeakup?Becausecomplainingcarriesacost.Theywouldhavetoreachouttoyou,waitforaresponse,andhopethatyoursupportteamisunlikemostsupportteamsthey’reusedto.Theirotheroption—tosearchforalternatives—isbothfasterandeasier. Forproductteams,fightingthissilentuserchurnwithoutanyinsightislikefillingaleakybucket.Howcanyoupatchyoursup?Productanalytics. Reportingonchurn Tounderstandchurn,productmanagersneedawaytoreportonuserbehavior.Theycaneitherbuildoutasystemthemselvestowarehouseandanalyzetheirdataortheycaninvestinproductanalytics.Eitherway,thegoalisthesame:determinewhatusersarethinkingandfeelingateachstageoftheirjourney,inordertodrivetherightbehaviors.Justas,accordingtoTheVerge,Facebookcanpredictthatcoupleswillbreakupbeforethecouplesthemselvesknow,youtoocanknowifyourusersareplanningonbreakingupwithyousothatyoucantrytoworkitoutfirst. Andofcourse,you’llneedtogetdetails.Simplyknowingthatabreakupisimminentdoesn’ttellyouhowtopreventit.Andjustknowingwhichdoorsomeoneexitedthroughdoesn’thelpeither.Youranalyticssystemneedstoallowyoutogodeepandtrackindividualuserjourneysandaskspecificquestionsrelatedtoyourcorebusinessgoal,bethatpurchases,usage,ormoreusers. Questionsproductleaderscommonlyaskoftheiranalytics: Whateventsarehighlycorrelatedwithmybusinessgoal? Whateventsarehighlycorrelatedwithusersnotreachingmybusinessgoal? Aremysubjectivehypothesessupportedordisprovedbytheobjectivedata? Next,you’llneedawaytoactonwhatyouknow.Thiscanmeanautomatingin-appnotifications,email,pushnotifications,personalization,SMSmessages,A/Btests,andmore.Whateveryouuseforanalytics,itneedstohookintoorprovidethismessagingsystem. Withtheabilitytoseeandinstantlyreact,youcanstarttouncoverinsightsandplugleakyholes. Herewecoverthefourtypicaltimesuserschurn,andwhatyoucandoabouteach: 1.Userschurniftheydon’tcompleteactivation Usersareindangerofchurningiftheydon’t‘getit’rightaway.Howcanyouspotifthisishappening?Youneedtheproductanalyticstoseewhatthey’vedone,stepbystep.Sometimesit’sobvious,andthedrop-offsoccurataloginpage,orafterusersadditemstotheirshoppingcart.Othertimes,it’shardertotell.Forexample,whatifusersdropoffafterreadingafewarticles?Ordisappearinthemiddleoffillingouttheirprofile?Orwhatifthere’snoclearpattern?Didtheseusersleavebecausetheydidn’tfindwhattheywerelookingfor,ordidtheyleavebecausetheyfoundit? There’snothingtodoherebutbeginaskingquestionsandtrackingtheanswers.Askingwhyuserschurningeneralisn’tpossible,butaskingquestionsaboutwhatleadstochurnsis.Forexample,ifyouA/Btestasimpleronboardingprocess,doesitcausethemtostick?Ifyouaddawalk-through,doesitimprovetimespentin-app?Ifusersleave,cantheybebroughtback?Doestheuser’sacquisitionsourcematter,anddosomesourcesproducemorevaluableusersthanothers?Allofthesequestionsdeserveconsideration. OnesportsgamingcompanyandMixpanelcustomerincreasedtheirconversionratesbyansweringthesequestions.Beforeimplementinganalyticstheycouldseethatuserswereconvertingatalowrate,butdidn’tknowwhy.Withanalyticsandtheabilitytoseeuserjourneys,theynoticedthatthegreatestdropoffoccurredjustafteruserscreatedtheirsportslineupsbutbeforetheysubmittedacashbet.Theanalyticsalsoshowedthatuserswhoreachedthispointbutdidn’tprogresswerehighlyunlikelytoreturn.Thepaywall,itappeared,wascausingthechurn.Thesolution?Theteamthereautomatedpushnotificationsofferingadiscountwheneveruserscreatedlineupsbutthendidn’tsubmitbets.Theirconversionspromptlytickedup. 2.Userschurnwhenretentioneffortsfail Userscanchurnwhentheproductdoesn’tshowthemenoughvalueearlyon.Facebookknewthisandfamouslydiscoveredthattheiruserswerelesslikelytostayiftheymadefewerthansevenfriendsintendays.Thisso-calledmagicnumberbecameoneofmanythingsFacebookoptimizedfor.Thisledthemtocreatesuchfeaturesasfriendrecommendations,aprompttoimportcontacts,andtheabilitytoinvitenewfriends. Yourownapp’sgoalsandKPIswillbeunique,butproductanalyticsshouldallowyoutoidentifythekeycharacteristicsofbothlongandshort-termusers.Typically,thedifferencebetweenthemiswhethertheyfoundenoughvalueinyourapp. Valuecomesinmanyforms.Forsomeusers,it’sutility:iforderingtaxisthroughtheirphoneismoreconvenient,they’llkeepdoingit.Forothers,it’sanexperience,suchasavideothatmakesthemlaugh,orjustfeelingwell-informed. Howdoyoudetectvalue?Again,productanalytics.Lookatwhereusersspendtheirtime.It’sourmostpreciousresourceintoday’sattentioneconomyandinmostcasesagreatbellwetherforvalue: Wheredousersspendmostoftheirtime? Wheredotheyspendmore/lesstimethanexpected? What’sthefirstplacetheynavigatetowhentheyopentheapp? Dotheyspendtimeherebychoice,oroutoffrustration? 3.Userschurnwhentheybecomefrustrated Userschurnwhentheygetfrustrated,andmostfrustrationscanbeavoidedwithbetterdesign.Toimproveyourdesign,watchyouruserscloselyandseewheretheygetstuck.They’lloftenshowyouthesolutionthattheywant.ThisishowtheUniversityofCaliforniaBerkeleyisrumoredtohavedesigneditsquad:insteadofplacingpaths,theyplantedalawnandwatchedwherestudentswalked.Whereverareasofgrassweretrampledandnaturaltrailsemerged,theyplacedcementwalkways. Talkofthis‘desirepath’methodiscommonamongdesignersbutrareinpractice.Manyproductteamsbecomeattachedtotheirpre-determineduserpathsandthenaredisappointedwhenactualuserbehaviorsdiverge.But,asexperienceteachesus,thesurestwaytobecomedisappointedwithusers(andtodisappointthem)istohaveinflexibleexpectations. “Neverblametheuser,”saysMattAronoff,co-founderoftheappdesignfirmLogicalAnimal.“Thereareonlytworeasonsusersgetlost:eithertheappdoessomethingyoudidn’texpect,liketheuser’snetworkconnectiongoesaway,ortheuserdoesn’tdosomethingthatyouexpect.Bothareyourfault.” It’seasyforproductteamstoblameuserswhentheylacksinsights,andevenmistakeuserfrustrationforsuccess.IrzanaGolding,adatascientistatCisco,providestheexampleofachatbotteamcelebratingafterseeingthenumberofmessagesspikewhilefailingtorealizethatthosemessageswerecomingfromirateuserswhodidn’thavetheirquestionsansweredsufficiently. Howcanproductteamsdetectuserfrustrationsandeasetheirpain?Delveintodeeperproductinsightstofind: Repeatedactions Excessive‘undo’actions Refreshing Abnormallyfrequentvisitstothemenu Longclick-stringsbeforeusersarriveatvalue Drop-offsmid-action 4.Userschurnwhentheproductteamconfusescorrelationwithcausation Finally,userschurnwhenproductteamsincorrectlyassumethattheyknowwhatmakesusersstay.Withoutthegranulardatathatanalyticsprovide,productteamsonlyhaveintuitiontoworkwith,andwhiletheymaybeexpertguessers,we’reallhuman.Everyoneissubjecttologicalfallacies.Sometimestheyconfusecorrelationwithcausationandassumethatjustbecausetwoeventsoccurtogether,theyarerelated. Mostofusdothisallthetime,fromwearingourteam’sjerseyongamedaytobuildingtowerswithoutthethirteenthflooror,say,assumingthatmoretimein-appleadstomorepurchases.Withtheentirecompanyoftenrelyingonthebusiness-criticaldecisionstheproductteammakes,it’simportantnottobebuildingneedlesssignaltowersandrunways.  Howcanproductteamsavoidthecausationtrap?Bymakingsurethattheyhavebothqualitativeandquantitativedatatobacktheirassumptions.Thequalitativedatacomesfromtheirexperienceanduserinterviews,andthequantitativedatacomesfromproductanalyticswheretheycanaskthequestion,‘Arethesefactorscausalorsimplycorrelated?’andgetafirmanswer. Mostuserschurnsilently,butwithadequateproductanalytics,theirstorieswon’tgountold.Productteamsneeddeep,granular,user-levelinsightstodistillnarrativesfromthedata.Whentheyachievethat,theycanfinallyaddresstheirchurnandfreetheiruserstogrowandflourishwithintheapp. customerdata Relatedarticles ProductFoundations 5thingsmarketerswishtheirPMsknewaboutworkingtogether Overthecourseofmycareer,I’vesatindozensofconferencerooms,Zoommeetings,retreatcenters,andteamoutings… AshleyHockney MarketingLeader,Ex-BufferandCodecademy ProductAnalytics Top6customerretentionmetrics Whenaproductdoesn'thavegoodcustomerretention,it'slikeworkingwithaleakybucket. Withoutexistingcustomerscomingbackto… TristanJung Editor@Mixpanel Innovators BorisJabesontheproblemsyoucansolvebyputtingyourdataintoaction BorisJabes’passionforsolvingproblemscanbetracedbackover30years.Whilemost80skidsfellintocomputers… DanielBean ManagingEditor@Mixpanel GetthelatestfromMixpanel Thisfieldisrequired.