68 CHAPTER 3 MARKETINGRESEARCH:ANAIDTODECISIONMAKING
Butevenif a managerwouldlikemoreirrformationinordertomakea decision,itis
notalwayswiseforhertoconducttheresearchthatwouldberequired.Onereasonis that
thetimeinvolvedmaybetoogreat.Anothermorecompellingreasonis thatthecostofthe
r~searchmayexceeditscontribution.Inprinciple,itiseasytounderstandhowsucha cost
testmightbeapplied.Ifthecostofconductingtheresearchislessthanitscontributionto
theimprovementofthedecision,theresearchshouldbecarriedout.Ifitscostisgreater,
it shouldnotbeconducted.Theapplicationofthisptincipleinactualpracticeissomewhat
morecomplex.Finally,goodresearchshouldhelpintegralemarketingwiththeotherareas
ofthebusiness.
INTEGRATED MARKETING •
RESEARCHBRINGSITTOGETHER
It'sthebaneofmodernbusiness:toomanydata,notenough
information.Computersareeverywhere,accumulatinggiga-
bytesgalore.Yetit onlyseemstogethardertofindthefor-
estforthetrees-toextractsignificancefromtheblizzardof
numbers,facts,andstats.Buthelpis ontheway,intheform
ofa newclassofsoftwaretechnologyknownbroadlyasdata-
mining.Firstdevelopedtohelpscientistsmakesenseofexper-
imentaldata,this softwarehasenough smartsto "see"
meaningfulpatternsandrelationships-toseepatternsthat
mightotherwisetaketensofman-yearstofind.That'sa huge
leapbeyondconventionalcomputerdatabases, whicharepow-
erfulbutunimaginative. Theymustbetoldpreciselywhatto
lookfor.Data-miningtoolscansiftthroughimmensecol-
lectionsofcustomer,marketing,production,andfinancialdata,
and, usingstatisticalandartificialintelligencetechniques,iden-
tifywhat'sworthnotingandwhat's not.
Thepayoffscanbehuge,asMCICommunicationsislearn-
ing. Likeotherphonecompanies,MCIwantstokeepitsbest
customers. Onewayis toidentifyearlythosewhomightbe
consideringjumpingtoa rival.Ifit candothat, theCartier
cantrytokeepthecustomerwithoffersofspecialratesand
services,forexample.
Howtofindthecustomersyouwanttokeepfromamong
themillions?MCl'sanswerhasbeentocomb marketingdata
on 140 millionhouseholds,eachevaluatedonasmanyas
10,000attributes-characteristicssuchasincome,lifestyle,
anddetails aboutpastcallinghabits. Butwhichsetofthose
attributesisthemostimportanttomonitor,andwithinwhat
rangeofvalues?A rapidlydecliningmonthlybillmayseem
likea deadgive-away,butis therea subtlerpatternininter-
nationalcallingtobelookingfor,too?Orinthenumberof
callsmadetoMCl'scustomer-servicelines?
Tofindout, MCIregularlyfiresupitsIBMSPI2super-
computer-its"datawarehouse"-whichidentifiesthemost
tellingvariablestokeepaneyeon.Sofar,theSPI2hascom-
pileda setof 22 detailed-andhighlysecret-statisticalpro-
filesbasedonrepeatedcrunchingofhistoricalfacts.Noneof
themcouldhavebeendevelopedwithoutdata-miningpro-
grams,saysLanceBoxer,MCT'sChiefInformationOfficer.
Data-mininginitselfis a relativelytinymarket:salesof
suchprogramswillgrowto maybe$750millionby2001.But
thetechnologyiscrucialingettinga bigpayofffromwhat
informationtechnologyexecutivesthinkwillbeanimmensely
importantgrowthareaincomingyears:datawarehousing.
Therearetheenormouscollectionsofdata-sometimestril-
lionsofbytes-compiledbymassmarketers,retailers,orserv-
icecompaniesastheymonitortransactionsfrommillionsof
customers. Datawarehouses,runningonultrafastcomputers
withspecializedsoftware,arethebasisonwhichcompanies
hopeto operatein realtime-instantlyadjustingproductmix,
inventorylevels,cashreserves,marketingprograms,orother
factorstochangingbusinessconditions.
Source:JohnW.Verity,"CoaxingMeaningoutofRawData,"
BusinessWeek,February3, 1997,pp.134-138;"ResearchersInte-
grateInternetToolsinTheirWork," R&DMagazine,June2000,
vol.24,No.6,p.E13; "SmarterKids.ComChoosesQuadstons-
TheSmartestCustomerDataMiningSolution,"BusinessWeek,
July31,2000.
