Applying Data Science and Analytics in Marketing
2 likes758 views
1) The document discusses applying data science in marketing to create data-driven marketing strategies through techniques like marketing mix modeling, predictive analytics, A/B testing, and attribution modeling. 2) Key benefits of these techniques include optimizing marketing budgets, identifying influential factors, improving customer experience, and tracking success metrics. 3) Building a data-driven marketing organization requires promoting data literacy, investing in technologies, embracing new technologies, and looking at data as an integral part of the business.
![TypesofAttributionModel
•LastInteractionAttributionModel–100%salescreditisattributedtothelasttouchpoint
•FirstInteractionAttributionModel–100%creditisgiventothefirsttouchpoint
•LinearAttributionModel–Givesequalcredittoeachtouchpointthroughoutthe
conversionpath.
•TimeDecay–Givestouchpointsclosesttotheconversionmostofthecredit.
Whichattributionmodelisbest?
“Anyattributionmodelisgoingtobemessy,”Mechanicsaid.“Findonethatmakessomedegreeofsenseandstickwithit.Whetherit’sfirsttouch,last
touchorblended,thereallyimportantthingisgettingeverybody[onateam]tobuyintoitandthenstickwiththatovertime.”--ChrisMechanicCEO
andco-founderofdigitalagencyWebmechanix](https://image.slidesharecdn.com/tobespeaker-201104004419/85/Applying-Data-Science-and-Analytics-in-Marketing-20-320.jpg)
Applying Data Science and Analytics in Marketing
- 1. By Tochukwu (Tobe) Matthias Data & Analytics Manager at Molina Healthcare & Founder of Data Techcon
- 2. AboutSpeaker Founder • Data Techcon, LLC Professional Expertise • Data & Analytics Manager, Molina Healthcare • Data Subject Matter Expert, Comptia Leadership • Chapter Lead Women in Data, Los Angeles Education • Msc Computer & Information Systems • Bsc. Computer Science & Statistics • Digital Analytics Certification - MIT
- 3. Agenda • Introduction to Data Science in Marketing • Marketing Mix Model • Data Collection Methods • Aggregating Data from Disparate Sources to build a Holistic Analytics Solution • Measurement & Metrics • Predict Analytics • A/B Testing • Attribution Model • Building a Data driven Marketing Structure in an Organization
- 4. DataScienceinMarketing • Applying Data Science in marketing involves leveraging data and advanced analytics in building analytics solution and statistical model that provides valuable insights that helps improve the effectiveness of marketing tactics for a more optimized data-driven strategy that maximizes profit and increases ROI. • One core goal of applying data science in marketing is to create Data-driven marketing strategies which will help drive informed and actionable decisions geared towards optimizing marketing tactics for the highest possible return on investment (ROI).
- 5. WhyImplementData-drivenMarketing ”Data-driven organizations are 23X likely to acquire customers, 6x as likely to retain & 19x likely to be profitable”. - McKinsey Global Institute “Marketers who want to stand the test of time must start harnessing data to drive real-time marketing strategies and influential consumer connections.” – Patrick Salyer “Tracking marketing is a cultural thing. Either tracking matters or it doesn’t. You’re in one camp or the other. Either you’re analytical and data-driven, or you go by what you think works. People who go by gut are wrong.” - Stuart McDonald
- 6. MarketingMixModel • MarketingMixModeling(MMM)istheapplicationofstatisticalanalysistomarketinghistoric dataacrossmediamixchannelslike;Paidsearch,PaidSocial,Display,TV,Radio,Onlineto quantifytheeffectivenessofmarketingcampaignsandoptimizemarketinginvestments. • Basedonanalysisthefourfactorsinfluencingmarketingare;Product,PricePlaceand Promotion. • Marketingmixcanbeimprovedbytesting,collectingdata,measuring,analyzing,monitoring andreporting.
- 7. MarketingMixChannels Digital Marketing Channels • Paid Search – Google & Bing • Paid Social – Facebook, IG • Affiliate – Media Alpha, Datalot • Streaming Audio – Hulu, Pandora • Streaming Video – Pandora, Youtube • Display - Verizon • Native – • Videos - Youtube Traditional Marketing Channels • TV • Radio • OOH • Direct Mail • Print
- 8. 4P’sofMarketingMix • Product–Thisisthefoundationofevery marketing.Buildinganoutstandingproductand userexperienceisessentialtoacompany’success. • Price–Itisimportanttonotethatthepriceof anycompany’sproductmustbeprofitable. • Place–Thisisthespacewhereconsumerengage withtheproducts.Itcanbeonlineoroffline. • Promotion–Thisisthemethodofadvertisement topromotetheproduct Imagesource:investinganswers
- 9. Benefitsof MarketingMixModel • Toidentifywhichmarketingchanneldrivesthehighestreturnsmaximumreturn(ROI) • Toforecastfuturemarketingbudgetacrossmediamixchannels. • IdentifyingKeyinfluencingfactorsthatdrivesengagementandgrowthsuchaschannel,medium,geo,price, competition,andmacro-economic factors) • Tohelpoptimizemarketingspendacrossallchannelsbasedonperformance • ToTrackandMonitorsuccessmetricsbasedonmarketingobjectives
- 10. DataRequiredforMarketingMixModel • Advertisementorcampaigndata • Demographicandbehavioraldata • Productdata • Websitedataforuserbehavior • Salesdata • Geolocationdata
- 11. DisparateDataSourcesinMarketing BUSINESS INTELLEGENCE TOOLS • Tableau • PowerBI • Qlik • Domo • Sisense • Looker • GoogleDataStudio
- 13. MeasuringSuccessinMarketing–KPIMetrics • Awareness–Reach,Impressions • Interest/Engagment–Clicks,Visits,CTR • Conversion–Leads,Purchases • Retention–CLTV,ChurnRate
- 14. Applicationof DataScienceinMarketing Optimize Marketing Budget Lead Scoring Predictive Model Advanced Audience Targeting Improve Customer Experience Retargeting Testing Segmentation Forecasting Attribution Modeling
- 15. PredictiveAnalyticsinDigitalMarketing Figure 1: This chart shows how predictive analytics provides greater business value by giving insight into the likelihood of future events. (Source: OLSPS Analytics n.d.) •Analysis •Prediction •Monitoring •Reporting
- 17. Casestudy–BuildaPredictiveModeltoPredictROAS DefinethebusinessObjective • ThebusinessgoalistoleveragehistoricdatafromgoogleandfbadstobuildapredictivemodeltoforecastROAS. Identifythedatasourcesandextractdatasets • Extract,Transformandloaddatafromvarioussources- Googleads,facebookads,Googleanalytics,python ExploratoryDataAnalysis • Prepareandcleansethedata-applydescriptivestatistics,removeduplicates,treatmissingvalue,convertcategoricalto numericvariable, removeunwantedcolumns,renamecolumns FeatureEngineering • UsingdomainexpertisetoCreatingnewfeaturesfromthedatasets likecalculatingCTR,CPL,ROAS
- 18. CaseStudy–BuildPredictiveModeltoPredictROAS ModelBuilding • Definethepredictor andinputfactors.ROASwillbetheindependentvariablehere. • Splitdataintotrainandtest(80%train,20%test) TraintheModel • Usethetrainingdatasettotrain,fitandtuneyourmodel usingcrossvalidation TesttheModel • Thetestdataisusedtotestthemodelaftertraining ValidatetheModel • Selectwinning model–youwillhave1bestmodelforeachalgorithm • Chooseperformancemetricsandevaluate.ForregressionwerecommendMeanSquaredError(MSE),MeanAbsoluteError(MAE). Lowervaluesarebetter.ForClassificationwerecommendArearUnderROCCurve(highervaluesarebetter)
- 21. A/BTesting • A/Btesting(alsoknownassplittesting)isaprocessconductinganexperimentontwovariationsof elementsonthewebpageandrandomlyassigningdifferentversionstotwodifferentsamplesizes. • WebsiteelementstoconductconductA/btesting • Logo,CTAButtons,Headlines,Images,Videos,Ads • Whya/bTesting
- 23. A/bTestingCaseStudy–CheckoutPageForm Experiment goal: Increase CTR on the billing and shipping page We have to run experiments to improve our goal
- 24. CaseStudy–A/bTestingCheckoutPageForm • Hypothesis–AddingaPayPalcheckoutoptionormakingtheformshortercanhelpimprove theclickthroughrate. • SuccessMetrics–DAU,Clicks,CTR • ExperimentProcess • Randomlyassign50%ofthesamplesizeoruserstocontrolgroupthentheother50%toexperiment group • Decideontheexperimentduration • Comparetheengagement metrics;CTRtoseeiftheyarestatisticallysignificant • Result–Sawa22%increaseinCTRtothenextfunnelwhichincreasedconversionrate
- 25. BuildingaData-DrivenMarketingStructure • PromoteDataLiteracy • Buildanalyticsintodigitaltransformation • Investintherighttechnologies • Embracenewtechnologies • Don’tlookatdataasaseparatepartofthebusiness