Introduc)on• P.J.Chambers• WebIXI(www.webixi.com)
– HarfordCountyWebsiteDesignCompany,1995– 2012HarfordAwardWinner– NamedOneoftheTop100SmallBusinessesintheNaNonbytheU.S.ChamberofCommerce
in2013– MembersoftheYear,2014–HarfordChamberofCommerce– StandardsforExcellenceLicensedConsultant,2015
• Focuson:– WebsiteDesign&Development(Responsive)– SearchEngineOpNmizaNon(SEO)– SocialMediaMarkeNng/OpNmizaNon– LocalSearchOpNmizaNon– ReputaNonManagement– CauseMarkeNngCampaigns– OutsourcedOnlineMarkeNngDirectors– OnlineMarkeNngAudits
GrowthHacking• ProcessofRapidExperimentaNonAcrossMarkeNngChannels&ProductDevelopment
• IdenNfytheMostEffecNve,EfficientWaystoGrowth
Crea)ngaGrowthHackingMindset
• IdenNfyingMicroImprovements• IdealSkillSets
Ac)onCoach’s5Ways
10%Increase10%Increase
NumberofLeads 4,000 4,400
X x x
ConversionRate 25% 27.5%
= = =
NumberofCustomers 1,000 1,210
X x x
NumberofTransacNons 2 2.2
X x x
Average$Sale $100 $110
= = =
Revenues $200,000 $292,820
X x x
Margins 25% 27.5%
= = =
Profits $50,000 $80,525.50
IdealSkillSets• Coder/Developer• Psychologist• Marketer/Branding• DataAnalyst
ScrumMee)ngs• AFocusedEffortfora30-dayPeriodTowardsFixedGoals
• ToDoListisConstantlyRe-prioriNzed• ScrumMaster’sRole– CoachesDevelopmentTeam– RemovesPossibleImpediments– WorkstoMakeSureTeamHasBestPossibleCircumstancesforSuccess
ScrumMee)ngRules
AgilityMarke)ng
AgileMarke)ngApproach• RelevantforStartUps&Nonprofits– ScarceResources–Can’tPutTime&MoneyIntoBigIdeasThatMayNotWork
– FailFast,FailCheap• ImplemenNngSmallerProjectsQuickly&Cheaply
• MakingDecisionsBasedonData,NotOpinions&Experiences
70-20-10Rule• MarkeNngEffortsShouldConsistOf– 70%PlannedMarkeNng– 20%AutomatedMarkeNng
– 10%ReacNonaryMarkeNng
TheGrowthHackingProcess• 5KeyPillarsofGrowthHacking
1. Product/MarketFit(CreateanMVP)2. UserandCustomerDataAnalysis3. ConversionRateOp)miza)on4. ViralGrowth5. RetenNonandScalableGrowth
CaseStudy:Instagram• OriginallyBrbn– CheckIns,FuturePlans,EarningPointsviaHangingwithFriends,PosNngPics
– Unsuccessful• UserAnalyNcstotheRescue
User&CustomerDataAnalysis• Goal:FindUserPanernsandTest/OpNmizeAcNviNesthatareLinkedtoGrowth
• 6MainAreasofUserTesNng– HeurisNcAnalysis– TechnicalAnalysis– MouseTrackingAnalysis– WebAnalyNcsAnalysis– QualitaNveSurveys– UsabilityTesNng
ResearchXLModel
FollowtheProcess• Succeedingwith– RunAsManyTestsAsPossible– WinAsManyTestsAsPossible– HaveasHighImpactPerSuccessfulTestAsPossible
• ResearchXLModelHelpswiththeDiscoveryofWhatManers.
• AListofTacNcsWon’tHelp–YouDon’tHaveGenericProblems,YouHaveSpecificProblems.
Heuris)cAnalysis• StructuredWebsiteReview– Relevancy– Clarity– Value– FricNon– DistracNon
• AvoidRandomComments–OnlyNamedCriteria
• GroupsWorkBest
TechnicalAnalysis• FixAllTechnicalIssuesFirst– BugsaretheMainConversionKillers–Always.
• YourSiteProbablyDoesn’tWorkPerfectlyAcrossAllBrowsers&Devices
• LowHangingFruit
CrossBrowser/DeviceTes)ng
SpeedAnalysis
GooglePageSpeedInsights
MouseTrackingAnalysis• UNlizing3rdPartySystems(LikeHotjar)• VisualizeWhatUsersAreDoingOnYourSite– ClickMaps– ScrollMaps– VideoRecordings
ClickMaps
ScrollMaps
CustomerRecordings
FormAnaly)cs• AnalyzeFormPerformance• DowntoIndividualFormFields• WhichFieldsCauseMostErrors?• WhichFieldsDoUsersHesitatetoFill?
(Milliseconds)• WhichFieldsDoUsersLeaveEmpty?
Qualita)veSurveys• TwoWaystoSurveyWebTraffic– ExitSurveys– OnPageSurveys
• ComponentsofaSuccessfulSurveySoqware– ConfigureWhichSpecificPagesOfferSurveys– SerngYourOwnQuesNons– DeterminetheCriteriaforWhentoShowSurvey
CustomerSurveys• SendanEmailSurveytoRecentClients– OnlyThoseWithNoPreviousRelaNonship– PotenNaltoSkewResults
• QualityofQuesNonsDetermineQualityofResults– AvoidYes/NoQuesNons– AvoidMulNpleChoice
NetPromoterScores(NPS)• NPSHelpsGaugetheLoyaltyofanOrganizaNon’sCustomer
RelaNonships• AlternaNvetoTradiNonalCustomerSaNsfacNon• CorrelatedtoRevenueGrowth
CustomerFeedback
ChurnRates• ChurnisaWayofExpressingRecurringCustomerAnriNon
• 100CustomersattheBeginningoftheMonth• DuringtheMonth5ofThemCancel• 5%CustomerChurnRate(5/100)
ChurnRate• SameScenario…• Earning$10,000/monthfromthose100Customers
• TheFiveYouLostWerePayingMoreThanAverage
• ThoseFiveWorth$1,000worthofRevenue• RevenueChurnRateis10%($1,000/$10,000)
HowDoesChurnRateAffectUs?• ChurnActsasaSpeedLimitonRevenueGrowth
• TheEffectGrowsastheCompanyGetsLarger– AsaCompanyGrows,theCustomerBaseBecomesLargeEnoughThatAnyKindofChurnAgainstTheBaseBecomesaLargeNumber
– LossofRevenueRequiresMore&MoreRevenueComingFromNewCustomersJusttoReplacetheChurn
– GrowthSlowsSubstanNally
ChurnImpactonRevenue• MostCompaniesUnderesNmatetheProblemChurnwillEventuallyBecomeAsTheyGrow– FailtoInvesttheTimetoMeasureandManageChurnBeforeItBecomesaProblem
ChurnImpactonRevenue
ChurnRateonRevenue
HowDoWeReduceChurn?• WeReduceChurninMulNpleWays:– IncreaseCustomerInteracNon– IncreaseCustomerSaNsfacNon
• TheGoalistoIncreaseCustomerLoyalty• UNlizetheDataYouAreCollecNng
UserTes)ng• ObservingActualPeopleUsing/InteracNngwithYourWebsite– CommenNngOutLoud
• ThreeTypesofTaskstoTest– ASpecificTask– ABroadTask– FunnelCompleNon
• Passthe“MotherInLaw”Test?• DoNotAskForOpinions.ObserveWhatTheyDoandSay
Crea)ngtheMasterList• CreateaMasterListofIssuesWeIdenNfied• TheFiveBuckets– Test– Instrument– Hypothesize– JustDoIt– InvesNgate
• Score&Rank– ConsiderEaseofImplementaNonandOpportunityScore
– FollowtheMoney
ConversionRateOp)miza)on
TheObamaCampaign
TheObamaCampaign
TheObamaCampaign• FourBunons&SixMediaTypes– ThreeImages,ThreeVideos
• SuccessMetric:SignUpRate• MulNvariateTest(AllCombinaNonsRunningattheSameTime)
• StaffBelieved“Sam’sVideo”WouldWin• Winner– SignUpRateof11.6%AgainstOriginal8.26%(40.6%)– Equatedto2.8MillionEmailAddresses– AddiNonal$60MinDonaNons
CommonFactors• ThereisNoBestColor– BlueMayBePeople’sFavorite–Doesn’tMeanMoreConversions
• ThereisNoBestFont– EasytoRead,Clear,Concise– No“Cool”Fonts
• TheCommonTesNngFactorsAreGenerallyCommonSense&GoodDesign
TheObvious
BestPrac)ces• BaseTestsoffHypotheses
• BaseHypothesesoffResearchXL
• GutReacNonsUsuallyResultinWastedTime&Money
• AlwaysBeTesNng
A/BSplitTes)ngMistakes• WhileManyServicesMakeA/BSetupEasy,ManyOrganizaNonsMakeCommonMistakes– WastedTime&Money
CallingtheResultsEarly• StaNsNcalSignificance–95%MINIMUM.• SampleSizesByDefaultAreTooSmall– OpNmizelydefaultsto100visitorspervariaNon,25conversionstodeclareawinner
• CalculateYourSampleSizeBEFOREStarNng– TakeIntoConsideraNonHowManyTesNngVariables
TestsNotRunForFullWeeks• 7DayRule• ExternalFactors(SeasonalFluctuaNons)
OtherCommonMistakes• A/BTesNngDonewithNoTraffic• TestsAreNotBasedonaHypothesis• TestDataNotSenttoGoogleAnalyNcs• TesNngSomethingStupid• GivingUpAqerFailure• FalsePosiNves• IgnoringSmallGains
ValidityThreats• InstrumentaNonEffect• HistoryEffect• SelecNonEffect• BrokenCodeEffect
Recap• ThePowerofMicroImprovements• User&CustomerDataAnalysis– ResearchXLModel– ThePowerofChurnRates
• ConversionOpNmizaNon– CommonTesNngFactors– CommonA/BTesNngMistakes
• AnyQuesNons?