new product forecasting using structured analogies...judgment is always going to be a big part of...
TRANSCRIPT
WHITE PAPER
New Product Forecasting Using Structured Analogies
SAS White Paper
Table of Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
The Challenges of New Product Forecasting . . . . . . . . . . . . . . . . . . . . 1
The Structured Analogy Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
An Example of Forecasting New Product Sales . . . . . . . . . . . . . . . . . . 3
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Content for this white paper was provided by Michael Gilliland, Product Marketing Manager for
SAS Forecasting. It is based on a patent-pending method of new product forecasting developed by
Michael Leonard, Tom Dickey, Sam Guseman and Michele Trovero.
1
New Product Forecasting Using Structured Analogies
IntroductionNew product forecasting (NPF) is a recurrent challenge for consumer goods manufacturers and retailers . There are many kinds of NPF situations these organizations can encounter:
• Entirelynewtypesofproducts.
• Newmarketsforexistingproducts(suchasexpandingaregionalbrandnationwideorglobally).
• Refinementsofexistingproducts(suchas“newandimproved”versionsorpackaging changes) .
TherearemanyNPFapproachesavailabletotry.Someofthecommononesinclude:
• Executive opinion–topmanagementprovidestheforecast.
• Sales roll-up–abottom-uppollofthesalesforce.
• Delphi method – a structured formal process for anonymously gathering forecasts andbuildingaconsensus.
• Prediction market – anonymous wagering used to gather group opinion .
• Analogy–expectinganewproducttobehavelikesimilarproductsfromthepast.
Allofthesemethodsusejudgmenttosomeextentandtherearegoodreasonswhy.Judgmentcompensatesforthelackofhistoricalinformationbecausewearedealingwith new products with no historical sales . Judgment also compensates for lack of futureinformation–itmaybetoodifficultorcostlytoconductmarketresearchtests,orto quantify such things as the direction of fads and fashion trends .
The Challenges of New Product ForecastingWhileuseofjudgmentisnecessaryinnewproductforecasting,ithasitsdisadvantagesaswell.Judgmentisfrequentlybiasedwithover-optimism,orallowingrecenteventstohaveunwarrantedimpact.Judgmentisalsocloudedbypersonalorpoliticalagendas,where the forecast is used to represent the forecasters’ wants or needs rather than what theyhonestlybelievewillhappen.
Asanexample,ifasalesrepknowshisforecastisgoingtobeusedtosetthesalesquota,thereisanaturaltendencytoforecastlow–tosetalowquotathatiseasiertobeat.Ifaproductmanagerwantstointroduceanewproduct,thereisanaturaltendency to forecast high – at least high enough to meet any hurdles for getting the new productapprovedfordevelopment.(Haveyoueverheardofanyoneforecastinganewproduct to fail in the marketplace?)
The Structured Analogy ApproachTheuseofanalogiesisacommonNPFpractice.Weseethisintherealestatemarket,whereanagentwillpreparealistof“comps”–similarhousesintheareathatareonthemarketorhaverecentlysold–andusethistosuggestasellingprice.
2
SAS White Paper
SAShasanewpatent-pendingapproachtoNPFthatcombinestheuseofanalogieswithstructuredjudgment.This“structuredanalogy”approach:
• Includesguidedstatisticalanalysisthatincorporateshumanjudgment.
• Attemptstoremovejudgmentalbiasbyprovidingahistoricalcontextforeachdecision .
• Attemptstovalidateandtestthedecisions.
• Driveschoiceofanalogybyastatisticalprocess.
Thestructuredanalogyapproachrequirestwotypesofdata:productattributes(forpriorandnewproducts)andhistoricalsales(forpriorproducts).Productattributescanincludemanythings,suchas:
• Producttype(toy,music,clothing,shirts,etc.).
• Seasonofintroduction(summeritem,winteritem,etc.).
• Financial(ownprice,competitorprice,etc.).
• Targetmarketdemographic(gender,age,income,ethnicity,etc.).
• Physicalcharacteristics(style,color,size,etc.).
• Andmanyothers.
Historicalinformationonpastnewproductintroductionsisalsoneeded.Herearescaledthumbnailsofthefirsteightweeksofsalesfor100newDVDreleases:
Figure 1: Scaled thumbnails of the first eight weeks of sales for 100 new DVD releases.
3
New Product Forecasting Using Structured Analogies
Thestructuredanalogyprocessfornewproductforecastinghassixmainsteps:
• Query step:Findasetofcandidateproductsthathavesimilarattributestothenew product .
• Filter step:Manuallyremoveinappropriateoroutlierproductsfromthesetofcandidate products .
• Cluster step:Clusterthecandidateproductsaccordingtotheirsalespattern,andmanuallyselectthemostappropriateclustertoserveasthesurrogateproducts.
• Model step: Select the most appropriate statistical model for the cluster of surrogateproducts,andextractthestatisticalmodelfeatures.
• Forecast step:Usetheextractedstatisticalmodelfeaturestoforecastthenewproduct .
• Override step: Make manual adjustments to the statistical model’s forecast .
An Example of Forecasting New Product SalesLet’sworkthroughanexampleofforecastingsalesofanewDVDmovierelease.TheQuerystepbeginswithselectingapubliclyavailabledatasetonhistoricalDVDsales,andthenspecifyingtheattributesofpriorDVDsthatmatchthenewDVD.Inthiscase,thenewDVDisanR-ratedhorrormovie,sowedecidetospecifytwoattributes:“Horror”fortheGenre,and“R”fortheMPAArating.
Figure 2: Specifying attributes for candidate products.
4
SAS White Paper
Notethatweareusingjudgmenttodeterminewhichattributesaremostrelevant–thesystemisnotgoingtotellusthis.However,thesystemdoesautomatealltheworkofextractingtheR-ratedhorrormoviesfromthedatasetofallDVDs,andtheseformourpool of candidate products .
TheoutputfromtheQuerystepisaprofileoverlayingtheinitialsalesofallthecandidateproducts,alongwithalistofallthecandidates.
Figure 3: Overlay of candidate product sales over their first 20 weeks.
JudgmentagaincomesintoplayintheFilterstep,aswedecidethatDawn of the Dead isanoutlier,anduncheckittoremoveitfromthecandidatepool.
Figure 4: Remaining candidate products (after filtering).
5
New Product Forecasting Using Structured Analogies
TheQuerystepletusexplorecandidateDVDswithsimilarattributes,andtheFilterstepallowedustomakeajudgmentoninappropriatecandidatestobefilteredout.Afterthefilteringisapplied,theClusterstepclusterstheremainingcandidatesaccordingtosimilarityoftheirsalespatterns.Judgmentagaincomesintoplay,astheuserselectsasurrogateclusterbasedontheanticipatedsalespatternforthenewproduct.InthiscasetheR-ratedhorrormoviesfellintothreeclusters,andthefirstwaschosen.Wearenow ready for the Model step .
Figure 5: User-selected cluster of surrogate products.
TheModelstepgeneratesastatisticalmodelthatfitsthesurrogatecluster.Theuserhasaccesstoavarietyofmodelstoutilize,includingdiffusion,mixed,smoothing,Bayesianandothers.Afterthemodelisselected,theForecaststepgeneratesaforecastforthenewproduct,whichisshownbelowinblueoverlaidonthesurrogateproducthistories.
Figure 6: Forecast model predictions overlayed on surrogate products.
6
SAS White Paper
JudgmenthasbeenusedthroughoutthestructuredanalogyprocessandisusedoncemoreintheOverridestepwheremanualadjustmentscanbemadetothestatisticalmodel forecast .
Figure 7: Users can override model predictions.
Afteranymanualoverridesaremade,thenewDVDforecastscanbeexportedtodownstream planning systems .
SummaryThestructuredanalogyapproachcanbeusefulinmany,butcertainlynotall,newproductforecastingsituations.Itattemptstoimproveonhumanjudgmentalonebyautomatingthehistoricaldatahandlingandincorporatingstatisticalanalysis.Thesoftwaremakesitpossibletoquicklyextractcandidateproductsbasedontheuser-specifiedattributecriteria.Italigns,scalesandclustersthehistoricalpatternsautomatically,providinganeasytounderstandvisualizationofpastnewproductbehavior.Thisvisualizationhelpstheforecasterrealizetherisks,uncertaintiesandvariabilityinnewproductbehavior,sothattheorganizationcanmaketheappropriatedecisionsbasedontheseuncertainties.
Judgmentisalwaysgoingtobeabigpartofnewproductforecasting.AcomputerwillneverbeabletotelluswhetherlimegreenorDay-Gloorangeisgoingtobethehotfashioncolornextseason.Butjudgmentneedsassistancetokeepitontrackandasobjectiveaspossible.Theroleofstructuredanalogysoftwareistoautomatethedataextractionandprocessingwork,providevisualizationofthehistoricalcontextforjudgmentsandmaketheNPFprocessasefficientandobjectiveaspossible.
SAS Institute Inc. World Headquarters +1 919 677 8000To contact your local SAS office, please visit: sas.com/offices
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and product names are trademarks of their respective companies. Copyright © 2012, SAS Institute Inc. All rights reserved. 104009_S94634_1112
About SASSASistheleaderinbusinessanalyticssoftwareandservices,andthelargestindependentvendorinthebusinessintelligencemarket.Throughinnovativesolutions,SAShelpscustomersatmorethan60,000sitesimproveperformanceanddelivervaluebymakingbetterdecisionsfaster.Since1976SAShasbeengivingcustomersaroundtheworldTHEPOWERTOKNOW® .