5
Loyalty Builders Data Dictionary of Calculated Metrics on Each Customer Predictive Customer Intelligence for More Effective Marketing Document Version 1.2 © 2015 Loyalty Builders, Inc. Confidential. All rights reserved. www.loyaltybuilders.com Simple, automated, cloud-based predictive analytics services enabling marketers to effectively target customers based on loyalty, value, risk, or product propensity, and deliver the specific 1:1 product recommendations to each customer that will lift revenue.

Loyalty Builders Predictive Metrics Dictionary 1.3F

Embed Size (px)

Citation preview

Page 1: Loyalty Builders Predictive Metrics Dictionary 1.3F

LoyaltyBuildersDataDictionaryofCalculatedMetricsonEachCustomerPredictiveCustomerIntelligenceforMoreEffectiveMarketing

DocumentVersion1.2©2015LoyaltyBuilders,Inc.Confidential.Allrightsreserved.www.loyaltybuilders.com

Simple,automated,cloud-basedpredictiveanalyticsservicesenablingmarketerstoeffectivelytargetcustomersbasedonloyalty,value,risk,orproductpropensity,anddeliverthespecific1:1productrecommendationstoeachcustomerthatwillliftrevenue.

Page 2: Loyalty Builders Predictive Metrics Dictionary 1.3F

LoyaltyBuildersMarketingLiftServiceTechnicalOverview�Page2©2016LoyaltyBuilders,Inc.Allrightsreserved.

www.loyaltybuilders.com

DataDictionaryofCalculatedMetricsPredictiveLoyaltyMetricsforEachCustomer:Metric Type Range Explanation ExampleUsesLoyaltySegment*

Text–7values

Loyalist,Nurturer,Underperformer,Fader,WinBack,1&2Buyer,Inactive

AsegmentationofanentirecustomerpopulationbasedoncustomerValueRankandRiskScore.Eachcustomerisassigned1of7designations.

Seedescriptionbelow.*LoyaltySegmentisacombinationofLoyaltyScoreandRiskScore.

LoyaltyScore

Numeric(0.0000)

0-100(eachcustomerforceranked)

Aproprietarymeasureofpotentiallifetimevalueofeachcustomerbasedontimingofpurchases,spendingtrends,itemspurchased,populationpatterns,andseveralotherfactors.

LoyaltyScoreisusedtotargetmarketingbasedoneachcustomer’srelativelong-termvaluewithinthecustomerbase.

RiskScore Numeric(0.00000)

0to1 Theprobabilitythatacustomerwillnotmakeapurchasebeforegoing“inactive.”Theactiveperiodisdeterminedbyanalysisofinter-orderfrequenciesamongallcustomers.

RiskScoreisusedtowin-backfadingcustomersortargetloyalcustomers.

LikelyBuyerScore

Numeric(0.00000)

0to1,foreachof10periods

Theprobabilitythatacustomerwillmakeapurchasewithineachof10timehorizons,whichareconfigurable(e.g.7,14,21,28,35,42,49,56,63,and90daysfromanalysis).

Providesguidanceonhowmucheffort,cost,ormargintoinvestinacustomer.Example:ifacustomerhasa60%probabilityofbuying,doesadiscountmakesense?Ifnegligibleprobability,shouldtheygetanexpensivemailer?

ExpectedValue

Currency($0.00)

$0tounlimited Theexpectedamountofrevenuefromthecustomerwithineachof10timehorizons,whichareconfigurable(e.g.7,14,21,28,35,42,49,56,63,and90daysfromanalysis).

Indicateswhetherexpensivemarketingprograms,suchasprintmailingscatalogs,orsalescalls,arejustifiable(basedonexpectedvalueoffuturepurchases).

*Furtherdescriptionprovidedbelow.PredictiveProductRecommendationsforEachCustomer:Metric Type Range Explanation ExampleUsesRe-SellProbability

Percent(float)

0to1,foreachof10periods

Theprobabilitythatagivencustomerwillpurchaseeachitemtheyhavepurchasedinthepastwithin10futuretimehorizons,configurableindays(e.g.7,14,21,28,35,42,49,56,63,and90daysfromanalysis).

Recommendationsthatencouragecustomerstore-purchaseproducts.Particularlyusefultoofferhighprobabilityre-sellofferstocustomerswithdeterioratingloyalty.

Cross-SellProbability

Percent(float)

0to1,foreachof30,60,and90days

Theprobabilitythatagivencustomerwillpurchaseeachitemtheyhaveneverpurchasedbeforein3futuretimehorizons,configurableindays(e.g.30,60,and90daysfromanalysis).

Recommendationsthatencouragecustomerstobuymore,differentproducts.Particularlyusefulforincreasingvalueofmoreloyalcustomersbywideningthebasketofproductstheypurchase.

Page 3: Loyalty Builders Predictive Metrics Dictionary 1.3F

LoyaltyBuildersMarketingLiftServiceTechnicalOverview�Page3©2016LoyaltyBuilders,Inc.Allrightsreserved.

www.loyaltybuilders.com

Cross-SellRankScore

Numeric(0.00)

0to10,foreachof30,60,and90days

Thepotentialinterestagivencustomerhasinpurchasingeachitemneverpurchasedbeforein3futuretimehorizons,configurableindays(e.g.30,60,and90daysfromanalysis)asarankedscore.

ThismaybemoreusefulthanCross-SellProbabilityinsituationswherethebestproductoffersareneeded,irrespectiveofthecustomer’sloyaltyorcurrentlikelihoodtobuy.

AdditionalAnalyticsforEachCustomer:Metric Type Range Explanation ExampleUsesLoyaltyRank

Numeric(0.0000)

0-100(eachcustomerforceranked)

AforcedrankingoftheLoyaltyScoretoindicateeachcustomer’sloyaltyrelativetooneanother.

ValueRankisusedtotargetmarketingspendforarangeofcustomersbasedontheirrelativevaluewithinthecustomerbase.

PurchaseDelay

Numeric-Days(0)

0toN

Numberofexpectedpurchasesmissedbyacustomer,basedontheaverageinter-ordertimebetweenpurchases.Thedelayperiodisuniquetoeachcustomer.Anumberof2indicates2missedexpectedpurchasesbasedontheaverageestablishedpattern.

Usedasanearlywarningtoaddresscustomerswhoarestartingtofalloffexpectedpurchasingpattern.

CategoryScore

Numeric(0)

0toN(N=NumberofProductCategories)

ThenumberofProductCategories(productsasgroupedforanalysis)thecustomerhaspurchasedfromhistorically(e.g.,iftheanalysiswasrunattheSKUlevel,CategoryScorereferstonumberofSKU’spurchased).

Usedwithothermetricstodeterminehowextensivelytomakecross-selloffers.

Retention Numeric-Days(0)

0tounlimited Thetime,indays,sincethefirstpurchase.

Usedtotailorspecificmessages,basedoncustomerseniority.

Recency Numeric-Days(0)

0toN(N=Inactivedesignation)

Thetime,indays,sincethelastpurchase.

Usedtotailorspecificmessages,basedonrecentcustomeractivity(orlackthereof),orbettertimeforadditionalcommunications.

NumberofOrders

Numeric(0)

0tounlimited Totalnumberofordersbyacustomer(eachorderhavingauniquetransactiondate).

Usedwithothermetricstodeterminetypesandfrequencyofoffers.

AmountSpent

Currency($0.00)

0tounlimited Threeseparatemetrics,includingthetotalamountspentbyeachcustomerinthelastperiod(e.g.month),last12months,andhistoricallyoverentiredataset.

Usedwithothermetricstotailorspecificmessages.

ProductAffinitiesTable:TheProductAffinitiesTablecontainsproductsalsopurchasedbycustomerswhena“PredictorProduct”hadbeenpurchased,rankedbylikelihoodtomaketheadditionalproductpurchase.Thefilecontainspair-wiseproductcombinations.ThefirstProductIDisthe“PredictorProduct.”ThesecondProductIDpairedwithitisthe“CompanionProduct,”whichisidentifiedwithitsliftvalueintermsoflikelihoodtobepurchasedbybuyersofthePredictorProduct.EverypairofProductCategoryID’sisevaluatedandallpairswithliftvaluesaboveauser-definedthresholdareincludedinthefile.

Page 4: Loyalty Builders Predictive Metrics Dictionary 1.3F

LoyaltyBuildersMarketingLiftServiceTechnicalOverview�Page4©2016LoyaltyBuilders,Inc.Allrightsreserved.

www.loyaltybuilders.com

*LoyaltySegmentsAtwodimensionalsegmentationiscreatedforthecustomerpopulationbasedontwometrics:predictedLoyaltyScoreandRiskScore.LoyaltySegmentscanalsobethoughtofas“lifecyclestages.”One-andtwo-timebuyersarenotincludedinthemainloyaltysegmentation,butarescoredforallmetrics.ReportsareproducedindicatingcustomersineachLoyaltySegment,rolledupfromtheirindividualscores,andidentifyingcustomertrendsandmovementsbetweensegments.Thisisonewayoffindingcustomerstotargetbasedonloyaltyandlifecyclestage.

Belowarethedescriptionsofthesegments:Loyalists–Highvalue,low-riskcustomers.Thesearethebestofthebestcustomerswhobuymostfrequently,mostbroadly,andspendthemost.Thesecustomersareusuallythebesttargetsforcross-sellitems,astheyhavebrandloyaltyandhavenumeroustransactions,whichleadtostrongerscoringrecommendations.Also,manyLoyalistswillmakeapurchasewhethertheyaremarketedtoornot,solargediscountsarenotnecessarytomovethesecustomers.AstrategythatsomeofourSubscribersuseistoofferadiscountonlytoLoyalistsiftheyincludeacross-sellitemintheirpurchase,thusincreasingtheirtotalbasketsize.Nurturers–Lowrisk,midtolow-valuecustomers.Thisgroup,andUnderperformers,areusuallythemostinterestinggroupsofcustomers.Nurturersarebuyingconsistentlyandperiodically,sincetheyhavealowRiskScore.Buttheyarenothighvaluecustomers,inthatthedollarstheyspendarelowortheymaynotbuyverybroadlyacrosscategories,orboth.Eithertheyarebuyingatcapacity,inwhichcasetheywillstaywheretheyare,ortheyarebuyingotherproductsfromcompetitors.Alittleinvestigationandsometargetedprogramscouldincreasethedollarsperorderforthesecustomersandincreasetheirvalue.Underperformers–Highvalue,midtohigh-riskcustomers.Thesearecustomersthatusedtobebettercustomers,butrecentlyhavefallenoffpattern,eitherbybuyinglessperorder,orbuyinglessfrequently,orboth.Therearemanyvaluablecustomersinthissegment,andreversingtheirbehaviorisessentialtoincreasingrevenue.Fortunately,thesecustomerstypicallyhavean

©2014 E

psilon Data M

anagement, LLC

. Private &

Confidential 13

Customer Migration Heat Map – December 2015 [Color – 6 Months Ago]

Highlights • Movement to the

top/left is the goal • Large group of Loyalty

that are drifting to Underperformers

• Huge opportunity in the Underperformer Segment

Page 5: Loyalty Builders Predictive Metrics Dictionary 1.3F

LoyaltyBuildersMarketingLiftServiceTechnicalOverview�Page5©2016LoyaltyBuilders,Inc.Allrightsreserved.

www.loyaltybuilders.com

establishedpurchasinghistory,somakingrelevantproductoffersiseasytodo.However,thesecustomersmayneedsomeadditionalincentivetogetbackontrack,sodeeperdiscountsarerecommendedforthesecustomers.Faders–Lowvalue,midtohigh-riskcustomers.Thesearecustomersthatneverreallytookofforestablishedasignificantbuyingpattern.Theymakesporadicpurchasesoflowdollaramounts,andtheytypicallybuyverynarrowlyacrosscategories.Othermetrics,suchasExpectedValue,canbeusedtodeterminewhichcustomersinthisgroupareworththecostandefforttosavevs.whichonesarenotprofitableinthelongrun.Deepdiscountsandrichoffersareusuallyrequiredtomovethesecustomers,butmaynotbecosteffective.WinBacks–High-riskcustomers.Thesearecustomersthathavenotmadeapurchaseinalongtimeandhaveahighlikelihoodofdefecting.Thesecustomerscanalmostbeconsideredasacquisitionsbecauseoftheeffortinvolvedinregainingtheirbusiness.OrderingthesecustomersbyValueRankcangiveastartingpointforacampaign.Theadvantageofthesecustomersoveracquisitionsisthattheyhavesomepurchasinghistory,andrelevantproductoffersarepossible.1&2xBuyers–Thesearecustomersthathavemadeatmost2purchases,whereapurchaseisdefinedasauniquetransactiondatewithpositiverevenue.Thesecustomersmaynotyethaveaclearloyaltysegment,andmaybeseparatelytargetedfromothercustomersinre-sellorcross-sellcampaigns.InactiveCustomers–Ananalysisofthetransactiondatacanbeusedtodeterminethemostappropriatecut-offpoint,intermsofnumberofdayssinceapurchase,todesignateacustomeras“Inactive”(cut-offpointcanbeoverriddenbytheSubscriberifspecialcircumstancesexist.)Inactivecustomersaregenerallynotworthpursuing,althoughsomecustomerswhoformallywerehighvalueatsomepointbeforegoinginactivemightbeworthsometargetedeffort.

LoyaltyBuilders,Inc.+1.603.610.8800

[email protected]