Se­cret Scores Shape Cus­to­mer Ser­vice

L'Opinion - - The Wall Street Journal - Kha­dee­ja Saf­dar

Two people call cus­to­mer ser­vice at the same time to com­plain about the same thing. One waits a few se­conds be­fore a re­pre­sen­ta­tive gets on the line. The other stays on hold. Why the dif­fe­rence?

There’s a good chance it has so­me­thing to do with a ra­ting known as a cus­to­mer li­fe­time va­lue, or CLV. That se­cret num­ber is used by all man­ner of com­pa­nies to mea­sure the po­ten­tial fi­nan­cial va­lue of their cus­to­mers.

Your score can de­ter­mine the prices you pay, the pro­ducts and ads you see and the perks you re­ceive.

Cre­dit-card com­pa­nies use the sco­ring sys­tems to de­cide what to of­fer cus­to­mers who want to can­cel their cards. Wi­re­less car­riers route high-va­lue cal­lers im­me­dia­te­ly to their most skilled agents. At some air­lines, a high score in­creases the odds of a seat up­grade.

“There’s no free lunch,” says Su­nil Gup­ta, a mar­ke­ting pro­fes­sor at Har­vard Bu­si­ness School who has re­sear­ched mo­dels for cal­cu­la­ting li­fe­time va­lue. “The more pro­fi­table you are, the bet­ter ser­vice you will get.”

These days, com­pa­nies are re­sor­ting to all sorts of da­ta and scores to size up consu­mers and pre­dict their be­ha­vior. Re­tai­lers use risk scores to try to li­mit mer­chan­dise re­turns and prevent e-com­merce fraud. There are scores to mea­sure the li­ke­li­hood a per­son will be­come sick, can­cel a sub­scrip­tion or bad­mouth a com­pa­ny.

Eve­ryone with a bank ac­count, cell­phone or on­line shop­ping ha­bit has at least one CLV score, more li­ke­ly se­ve­ral. And most people have no ink­ling they even exist, let alone how they are used, what goes in­to them or how ac­cu­rate they are. Un­like cre­dit scores, CLVs aren’t avai­lable to consu­mers and aren’t mo­ni­to­red by any go­vern­ment agen­cy.

“There needs to be a pu­blic conver­sa­tion around the ac­cu­ra­cy of the scores being used,” says Pam Dixon, exe­cu­tive di­rec­tor of the World Pri­va­cy Fo­rum, a non­pro­fit di­gi­tal-pri­va­cy re­search group. “You can es­sen­tial­ly be ac­cu­sed of being cheap or a fraud­ster, and it may not even be true.”

To de­ter­mine how the scores are com­pi­led and how they are used, The Wall Street Jour­nal in­ter­vie­wed da­ta scien­tists who de­ve­lop the mo­dels and em­ployees of the soft­ware and ana­ly­tics firms that help com­pa­nies put them to use.

Most CLV score users contac­ted for this ar­ticle de­cli­ned to comment on how they score cus­to­mers, ci­ting com­pe­ti­tive rea­sons. Ma­ny say the scores make them more com­for­table of­fe­ring cost­ly ser­vices and pro­ducts in the short term be­cause they are confi­dent they will pick up more bu­si­ness in the long term. Some say they aim to in­crease each cus­to­mer’s li­fe­time va­lue by en­cou­ra­ging re­peat bu­si­ness.

In some res­pects, the scores are just a high-tech ver­sion of what shop­kee­pers have done for ge­ne­ra­tions—make judg­ments on a cus­to­mer’s va­lue ba­sed on how they look or be­have. As far back as 20 years ago, aca­de­mics were pu­bli­shing mo­dels to cal­cu­late the fu­ture va­lue of cus­to­mers.

Now there are hun­dreds of ana­ly­tics firms that cal­cu­late cus­to­mer li­fe­time va­lue, each with its own ap­proach. Some of them put a va­lue on shop­pers ba­sed sim­ply on what they spend, while others use hun­dreds of da­ta in­puts, ad­ding and de­duc­ting points for de­mo­gra­phic in­for­ma­tion such as ZIP Codes or be­ha­vio­ral de­tails such as the num­ber of re­turns they make or when they shop.

“Not all cus­to­mers de­serve a com­pa­ny’s best ef­forts,” says Pe­ter Fa­der, a mar­ke­ting pro­fes­sor at the Uni­ver­si­ty of Penn­syl­va­nia’s Whar­ton School who hel­ped po­pu­la­rize li­fe­time va­lue scores. His sco­ring me­thod is ba­sed on tran­saction his­to­ry, which he says is all com­pa­nies need to de­ter­mine how cus­to­mers will be­have in the fu­ture. This year, he sold the firm he co­foun­ded, Zo­diac Inc., which per­forms such ana­ly­sis, to Nike Inc.

The da­ta that goes in­to a score can come from tran­saction re­cords, web­site in­ter­ac­tions, cus­to­mer-ser­vice con­ver­sa­tions, so­cial-me­dia pro­files and third-par­ty bro­kers such as Acxiom LLC and Al­liance Da­ta Sys­tems Corp.’s Ep­si­lon, which sell in­for­ma­tion on such things as the num­ber of be­drooms in a house and the type of cre­dit card so­meone car­ries. Each piece of da­ta is weigh­ted ba­sed on past pat­terns and per­cei­ved le­vel of pre­dic­ta­bi­li­ty.

Ma­ri­tal sta­tus is of­ten fac­to­red in, with some com­pa­nies as­su­ming that singles are bet­ter cus­to­mers, and others, the op­po­site. Age al­so is a com­mon in­put, po­ten­tial­ly pe­na­li­zing ol­der people be­cause of their shor­ter pro­jec­ted li­fes­pans.

Some com­pa­nies de­duct points from shop­pers who ex­hi­bit cost­ly be­ha­viors. Banks so­me­times take in­to ac­count the calls people make to cus­to­mer­ser­vice agents or the num­ber of times they vi­sit branches. On­line re­tai­lers track shop­pers who buy things on­ly when they are dee­ply dis­coun­ted. People ex­pec­ted to cost more than they spend can have a ne­ga­tive score.

Com­pu­ter sys­tems so­me­times tag cus­to­mers as high­va­lue or low-va­lue. Mar­ke­ting staf­fers or ser­vice agents gauge in­ter­ac­tions ba­sed on the sta­tus. Ven­dors such as Ze­ta Glo­bal and Sa­les­force Inc. can au­to­ma­ti­cal­ly of­fer dis­counts and other in­cen­tives ba­sed on the scores.

At wi­re­less car­riers such as Ve­ri­zon Com­mu­ni­ca­tions Inc. and Sprint Corp., li­fe­time va­lue can de­ter­mine mar­ke­ting of­fers and other perks. At some car­riers, high-va­lue cus­to­mers who are at risk of swit­ching to ano­ther car­rier are prio­ri­ti­zed and get rou­ted to a top-ra­ted call cen­ter.

Ve­ri­zon and Sprint de­cli­ned to pro­vide spe­ci­fics about how they as­sess cus­to­mer va­lue. “The pre­do­mi­nant way we route calls is ba­sed on the rea­son for the call,” says a Sprint spo­kes­wo­man. She says cus­to­mer li­fe­time va­lue is “one of ma­ny ways we guide cus­to­mer in­ter­ac­tions.”

Ze­ta Glo­bal, whose clients in­clude wi­re­less car­riers, ge­ne­rates scores using da­ta points such as the num­ber of times a cus­to­mer has dia­led a call cen­ter and whe­ther that per­son has brow­sed a com­pe­ti­tor’s web­site or sear­ched cer­tain key­words in the past few days. The firm says it has a da­ta­base of more than 700 mil­lion people, with an ave­rage of over 2,500 pieces of da­ta per per­son.

When a per­son’s “churn” score, which pre­dicts his or her chances of swit­ching to ano­ther car­rier, ex­ceeds a cer­tain thre­shold, Ze­ta’s sys­tem flags that cus­to­mer to a cus­to­mer-ser­vice agent. The hi­gher the cus­to­mer’s li­fe­time va­lue, the more li­ke­ly that Ze­ta will re­com­mend re­spon­ding to the cus­to­mer more qui­ck­ly and of­fe­ring free phones and other perks, says Da­vid Stein­berg, Ze­ta’s chief exe­cu­tive. “Most of this comes down to how you’re mar­ke­ted to and how you’re trea­ted,” he says.

Ap­pa­rel re­tai­lers of­ten com­pare a shop­per’s li­fe­time va­lue with the cost of mar­ke­ting to that per­son be­fore de­ci­ding whe­ther to woo him or her and how much mo­ney to spend doing so.

“What CLV does is al­lows us to see beyond the day-to-day to en­sure we’re fo­cu­sed on the qua­li­ty of the new cus­to­mers we’re ac­qui­ring, not just the quan­ti­ty,” says Ed Boyle, se­nior di­rec­tor of per­for­mance mar­ke­ting at Bo­no­bos, an ap­pa­rel re­tai­ler ac­qui­red by Wal­mart Inc.

In a re­search pa­per last year, ASOS, an on­line re­tai­ler, said it scores shop­pers on over 100 da­ta in­puts, in­clu­ding a cus­to­mer’s age and lo­ca­tion. Since ASOS doesn’t re­coup de­li­ve­ry costs for re­tur­ned items, “cus­to­mers can ea­si­ly have ne­ga­tive li­fe­time va­lue,” the pa­per said. The com­pa­ny de­cli­ned to comment on the pa­per.

Brad Birn­baum, chief exe­cu­tive of cus­to­mer-ser­vice plat­form Kus­to­mer Inc., says some of his e-com­merce clients use scores, in­clu­ding CLV, to re­spond to email in­qui­ries. “If you’ve got an an­gry shop­per with a high li­fe­time va­lue, you might want to bump up the prio­ri­ty,” he says.

Shop­pers with hi­gher scores, ho­we­ver, won’t ne­ces­sa­ri­ly get the best deals all the time, says Jer­ry Jao, chief exe­cu­tive of Re­ten­tion Science, which has wor­ked for com­pa­nies such as Tar­get Corp. and Proc­ter & Gamble Co. Re­tai­lers so­me­times wi­th­hold dis­counts to high-va­lue cus­to­mers un­til they are at risk of lo­sing them. “Why waste a 25% of­fer when the per­son is going to buy any­way?” Mr. Jao says.

At au­to dea­ler­ships, a high score can mean ac­cess to loa­ner cars, pre­fe­ren­tial ser­vice slots and spe­cial events, says Scot Ei­sen­fel­der, chief exe­cu­tive of Af­fi­ni­tiv Inc., which uses li­fe­time va­lue to create mar­ke­ting cam­pai­gns for dea­ler­ships. The sco­ring helps dea­ler­ships weed out cost­ly cus­to­mers. “This is what you call grin­ders—people who vi­sit 16 stores to get the ab­so­lute lo­west price,” he ex­plains.

Mr. Ei­sen­fel­der says his firm de­ve­lops scores by crun­ching da­ta on things such as pre­vious car pur­chases, whe­ther a hou­se­hold has a tee­na­ger, where else a per­son has shop­ped and ZIP Codes, which can be used as a proxy for in­come. So­meone who has a Nei­man Mar­cus cre­dit card is going to be more va­luable for a car dea­ler­ship than so­meone with a cre­dit card from a dis­count chain, he says.

At air­lines, CLV scores in­cor­po­rate frequent-flier in­for­ma­tion and other da­ta. A high score can in­crease a per­son’s chances of get­ting seat up­grades or bet­ter ser­vice, says Laks Sri­ni­va­san, co-chief ope­ra­ting of­fi­cer of Ope­ra So­lu­tions LLC, which works with air­lines, re­tai­lers, banks and other com­pa­nies.

The firm’s scores can draw from more than 5,000 da­ta “si­gnals” per cus­to­mer, Mr. Sri­ni­va­san says, trans­la­ting them in­to re­com­men­da­tions for flight at­ten­dants, gate agents and other per­son­nel. The com­pa­ny tracks, for example, the num­ber of times a per­son calls to com­plain over the prior 90 days, which can af­fect the CLV.

An air­line can com­pare how of­ten a shop­per com­plains with his or her li­fe­time va­lue and cus­to­mer ex­pe­rience score, which mea­sures in­con­ve­niences such as num­ber of times in the middle seat, flight de­lays and lost bags.

“A high-va­lue cus­to­mer who had a real ser­vice dis­rup­tion and ne­ver calls to com­plain should be com­pen­sa­ted more qui­ck­ly than so­meone who is com­plai­ning and cos­ting time and mo­ney,” Mr. Sri­ni­va­san says.

To cal­cu­late li­fe­time va­lue, cre­dit-card com­pa­nies ana­lyze spen­ding be­ha­vior, pay­ment his­to­ry and cre­dit scores, among other things. “Banks know what you buy, and where and when you buy it,” says Ar­pan Das­gup­ta, head of fi­nan­cial ser­vices and te­le­com prac­tices at Frac­tal Ana­ly­tics, which helps com­pa­nies ana­lyze cus­to­mer da­ta. “It’s po­wer­ful da­ta that can be use­ful for CLV.”

The score can de­ter­mine which cus­to­mers re­ceive cre­dit­card of­fers and other in­cen­tives. When cus­to­mers call to can­cel at a card com­pa­ny such as Ame­ri­can Ex­press Co., their re­la­tion­ship with the is­suer and past spen­ding be­ha­vior are some of the cri­te­ria used to de­ter­mine what perks will be of­fe­red to re­tain them.


Some re­tai­lers de­duct points from shop­pers who ex­hi­bit cost­ly be­ha­viors, such as buying things on­ly when they are dee­ply dis­coun­ted.

Newspapers in French

Newspapers from France

© PressReader. All rights reserved.