a new form of competition based on the extensive use of analytics, data, and fact-based decision making. The analytics quantitative or statistical models to analyze business problemsmay be applied to a variety of business problems, including cus- tomer management, supply chains, and financial performance. The research assessed 32 firms with regard to their orientation to analytics; about one-third were classified as fully engaged in analytically oriented strate- gies. Both demand and supply factors for analytical competition are described. Of the two, demand factors are the more difficult to create. The presence of one or more committed senior executives is a primary driver of analytical competition. On What Basis Do Companies Compete Today? I n vi r tual l y every i ndustry, many former strategi c al ternati ves are no l onger vi abl e or l i kel y to be successful . Today, there are few regul ated monopol i es, or compa- ni es wi th uni que geographi cal access. Propri etary technol ogi es are rapi dl y copi ed by compet i t or s, and br eakt hr ough i nno- vat i on i n ei t her pr oduct s or ser vi ces i s r ar e. Most of t he compet i t i ve st r at e- gi es or gani zat i ons ar e empl oyi ng t oday i nvol ve opt i mi zat i on of key busi ness processes. Instead of serving all customers, t hey want t o ser ve opt i mal cust omer s t hose wi t h t he hi ghest l evel of pr of - i t abi l i t y and l i f et i me val ue. I nst ead of r ecei vi ng goods and ser vi ces whenever t hey happen t o ar r i ve, t hey at t empt t o opt i mi ze suppl y chai ns t o mi ni mi ze di sr upt i ons and i n- pr ocess i nvent or y. I nst ead of l ooki ng backwar d at t hei r busi ness per f or mance and maki ng ex post f act o adj ust ment s, t hey seek t o under st and how opt i mum nonf i nanci al per f or mance dr i ves opt i mum f i nanci al per f or mance, and t o make accur at e f or ecast s of f ut ur e per f or mance so t hey can r eact i n advance of si t uat i ons. I nst ead of t hr owi ng money at busi ness pr obl ems, t hey seek t o opt i mi ze t hei r use of capi t al . But st r at egi es i nvol vi ng opt i mi zat i on r equi r e somet hi ng di f f er ent t han t hose based on t aki ng busi ness as i t comes. Above al l , t hey r equi r e ext ensi ve dat a on t he st at e of t he busi ness envi r on- ment and t he company s pl ace wi t hi n i t , and ext ensi ve anal ysi s of t he dat a t o model t hat envi r onment , pr edi ct t he consequences of al t er nat i ve act i ons, and gui de execut i ve deci si on maki ng. Mor eover, t hey r equi r e anal yst s and deci si on maker s who bot h under st and t he val ue of anal yt i cs and know how best t o appl y t hese f or dr i vi ng enhanced per f or m- ance. Compani es t hat st r i ve t o opt i mi ze t hei r busi ness per - f or mance usi ng t hi s dat a- i nt ensi ve appr oach ar e compet - i ng on anal yt i cs and anal yt i cal capabi l i ti es. Many compani es are pursui ng opti mi zati on- WORKI NG KNOWLEDGE RESEARCH REPORT Competing on Analytics THOMAS H. DAVENPORT, DON COHEN, AND AL JACOBSON MAY 2005 About This Research Thi s research repor t i s based on anal ysi s of 32 organi zati ons from a vari ety of i ndustri es (Fi gure 1) that are successful both i n terms of thei r overal l per formance and i n thei r use of busi ness anal yti cs. The research was under taken to i nvesti gate and document how and why these compani es not onl y use sophi sti cated anal yti cs, but al so make them the basi s of thei r competi ti ve strategi es, and adopt or move toward an enterpri se- l evel approach to busi ness i ntel l i gence. Tel ephone or i n-person i ntervi ews were conducted wi th ei ther I T or busi ness execu- ti ves at 30 fi rms; three fi rms were anal yzed sol el y on the basi s of secondary research. I NDUSTRI ES REPRESENTED NUMBER Fi nanci al Servi ces 10 Consumer Products and Retai l 6 Travel , Transport, and Entertai nment 5 Pharmaceuti cal and Chemi cal 4 Informati on Technol ogy and Communi cati ons 3 Heal th Care 2 Engi neeri ng 1 Government 1 Figure 1: Industries of Companies Surveyed COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 2 based strategi es, but most have fai l ed to devel op the anal yti cal capabi l i ti es neces- sary to make them succeed. The i dea of competi ng on anal yti cs i s not enti rel y new. A few organi zati onsmost wi thi n fi nanci al servi ces and par ti cul arl y i n financial investment and trading businesses have competed on thi s basi s for decades. The tradi ng of stocks, bonds, currenci es, and commodi ti es has l ong been dri ven by anal yti cs. What i s new i s the spreadi ng of anal yti cal competi ti on to a vari ety of other i ndustri esfrom consumer fi nance to retai l i ng to travel and enter tai nment to consumer goodsand wi thi n compa- ni es f r om i ndi vi dual busi ness uni t s t o an enterpri se-wi de perspecti ve. Even the most tradi ti onal l y i ntui ti ve i ndustri es are movi ng i n thi s di recti onprofessi onal spor ts teams, for exampl e. Two of Boston s sports teams have had some envi abl e success of l ate, at l east i n part because of thei r anal yti cal capabi l i ti es. The New Engl and Patri ots footbal l team, for exampl e, won i ts thi rd Super Bowl i n four years. The team uses data and anal yt- i cal model s extensi vel y, both on and off the fi el d. In-depth anal yti cs hel p the team sel ect i ts pl ayers, stay bel ow the sal ary cap, deci de whether to punt or go for i t on fourth down, and try for one poi nt or two after a touchdown. Both i ts coaches and pl ayers are renowned for thei r exten- si ve study of game fi l ms and stati sti cs, and Head Coach Bi l l Bel i chi ck peruses arti cl es by academi c economi sts on stati sti cal probabi l i ti es of footbal l outcomes. Off the fi el d, the team uses detai l ed anal yti cs to assess and i mprove the total fan experi - ence. At every home game, for exampl e, 20 to 25 peopl e have speci fi c assi gn- ments to make quanti tati ve measurements of the stadi um food, parki ng, personnel , bathroom cl eanl i ness, and other factors. External vendors of servi ces are moni tored for contract renewal and have i ncenti ves to i mprove thei r performance. The Boston Red Sox basebal l team i s al so a conver t to anal yti cs (fol l owi ng, i n many ways, the l ead of the pi oneeri ng but l ess wel l -fi nanced Oakl and A s). The abi l i ty to extract knowl edge from data presum- abl y hel ped the Sox wi n the Worl d Seri es i n 2004 for the fi rst ti me i n 86 years. Boston has begun to sel ect pl ayers l ess on tradi ti onal cri teri a such as batti ng aver- age, but rather on newer, more subtl e factors such as on-base percentage. Bi l l James, consi dered the godfather of base- bal l stati sti cs or sabermetri cs, was hi red by the Red Sox as an advi ser. The Sox al so have become more anal yti cal off the fi el d. Li ke the Patri ots, they map and moni tor key aspects of the fan experi - encefrom the deci si on to go to a game, to the routes taken by fans to the game, to the effecti veness of the cl eani ng crew. The team s management has maxi mi zed revenues from Fenway Park, the smal l est basebal l park i n the maj or l eagues, by cal - cul ati ng ti cket pri ce el asti ci ti es, establ i sh- i ng an onl i ne market for season ti cket resal es, and model i ng revenue i ncreases from addi ng seats i n unused l ocati ons (i ncl udi ng on top of the Green Monster, the toweri ng l eft fi el d wal l ). Anal yti cal competi ti on i s not onl y taki ng root i n U. S. spor ts. Some soccer teams i n Europe al so have begun to empl oy si mi l ar techni ques. AC Mi l an, one of the l eadi ng teams i n Europe, uses predi cti ve model s to prevent pl ayer i nj uri es by anal yzi ng physi ol ogi cal , or thopedi c, and mechani cal data from a vari ety of sources. Bol ton, a fast-ri si ng Engl i sh soccer team, i s known for i ts manager s use of extensi ve data to eval uate pl ayer per formance and team strategi es. Anal yti cal cul tures and processes are appeari ng not onl y i n professi onal spor ts teams, but i n any busi ness that can har- ness extensi ve data, compl ex stati sti cal processi ng, and fact-based deci si on mak- i ng. Anal yti cs i s becomi ng a pri mary basi s for competi ti on for these fi rms. They use anal yti cal tool s to change the way they compete or to per form substanti al l y better i n the exi sti ng busi ness model . The gam- i ng fi rm Harrah s, for exampl e, has chosen to compete on anal yti cs for customer l oy- al ty and servi ce, rather than on bui l di ng the mega-casi nos i n whi ch i ts competi tors have i nvested. I ts CEO, Gary Loveman, has commented, We use database mar- keti ng and deci si on-sci ence-based anal yti - cal tool s to wi den the gap between us and casi no operators who base thei r cus- tomer i ncenti ves more on i ntui ti on than evi dence. Amazon. com uses extensi ve anal yti cs to predi ct what products wi l l be successful , and to wri ng every bi t of effi - ci ency out of i ts suppl y chai n. Amazon CEO Jeff Bezos notes, For every l eader i n the company, not j ust for me, there are deci si ons that can be made by anal y- si s. These are the best ki nds of deci si ons. They re fact-based deci si ons. At the mutual fund company Dreyfus, anal ysi s of customer i nformati on defi ned segmenta- ti on that hel ped reduce fund attri ti on from 22 to 7 percent a year. These compani es, and a vari ety of others, are cl earl y com- peti ng on anal yti cs. Analytical cultures and processes are appearing not only in professional sports teams, but in any business that can harness extensive data, complex statistical processing, and fact-based decision making. Attributes of Analytically Oriented Companies Vi r tual l y every maj or company uses some form of stati sti cal or mathemati cal anal y- si s, but some take anal yti cs much fur ther than others. I n our research on the topi c, we have i denti fi ed several key attri butes of fi rms that compete on anal yti cs: One or more seni or executi ves who are strongl y advocati ng anal yti cs and fact-based deci si on maki ng; Wi despread use of not j ust descri pti ve stati sti cs, but predi cti ve model i ng and compl ex opti mi zati on techni ques; Substanti al use of anal yti cs approaches across mul ti pl e busi ness functi ons or processes; Movement toward an enterpri se-l evel approach to managi ng anal yti cal tool s, data, and organi zati onal ski l l s and capabi l i ti es. Each of these attri butes i s descri bed bri efl y bel ow: One or More Senior Executives as Advocates The adopti on of a broad anal yti cal approach to busi ness requi res changes i n cul ture, process, behavi or, and ski l l s for mul ti pl e empl oyees. Such changes don t happen by acci dent; they must be l ed by seni or executi ves wi th a passi on for ana- l yti cs and fact-based deci si on maki ng. I deal l y, the pri mary advocate shoul d be the CEO, and i ndeed we found several chi ef executi ves who were dri vi ng the shi ft to anal yti cs at thei r fi rms. These i ncl uded Gary Loveman, CEO of Harrah s; Jeff Bezos, the founder and CEO of Amazon; Ri ch Fai rbank, the founder and CEO of Capi tal One; and Barry Beracha, formerl y CEO of Sara Lee Bakery Group. Each of these executi ves has stated both i nternal l y and publ i cl y that thei r compa- ni es are engaged i n some form of anal yti - cal competi ti on. For exampl e, Fai rbank commented: Its all about collecting information on 200 million people youd never meet, and on the basis of that infor- mation, making a series of very critical long-term decisions about lending them money and hoping they would pay you back. Fai rbank summari zes thi s approach as I nformati on-Based Strategy. Beracha, before he reti red as CEO of Sara Lee Bakery, si mpl y kept a si gn on hi s desk sayi ng, I n God we trust; al l others bri ng data. Wi thout the push from the top, i t s rare to fi nd a fi rm maki ng the cul tural changes necessary to become an anal yti cal com- peti tor. We found some fi rms, for exampl e, i n whi ch si ngl e functi onal or busi ness uni t l eaders were tryi ng to engi neer an anal yt- i cal l y ori ented shi ft i n thei r fi rms, but weren t abl e to suffi ci entl y change the cul ture by themsel ves. Thi s doesn t mean, of course, that such an executi ve coul dn t l ead such a change under other ci rcum- stances, and we di d fi nd organi zati ons i n whi ch l ower-l evel advocates were maki ng progress on changi ng the cul ture. Any cross-functi onal or cross-depar tmental change, and cer tai nl y any enterpri se-wi de effor t, cl earl y requi res the suppor t and attenti on of executi ves seni or enough to di rect and coordi nate effor ts i n those separate uni ts. Widespread Use of Predictive Modeling and Optimization Techniques Every fi rm can cal cul ate si mpl e descri pti ve stati sti cs about aspects of i ts busi ness (the average revenue per empl oyee, or average order si ze), but the most aggres- si ve anal yti cal competi tors go wel l beyond basi c stati sti cs. They are usi ng predi cti ve model i ng, for exampl e, to i denti fy not onl y the most profi tabl e customers, but those wi th the most profi t potenti al , or those most l i kel y to stop bei ng customers. They are combi ni ng and pool i ng both i nternal and external data i n such a way as to gai n a more comprehensi ve pi cture and under- standi ng of thei r customers than was ever thought possi bl e. They are opti mi zi ng thei r suppl y chai ns, so they can determi ne the i mpact of an unexpected constrai nt, si mul ate al ternati ve suppl y chai ns, and route shi pments around probl ems. They are establ i shi ng pri ces i n real ti me so as to provi de the hi ghest yi el d possi bl e from a customer transacti on. I n fi nanci al per- formance anal ysi s, they create compl ex model s of how thei r operati onal and cost measures rel ate to thei r fi nanci al per form- ance. No matter what the busi ness func- ti on, i t s possi bl e to i mprove per formance through the use of sophi sti cated anal yti cal techni ques. Compani es that are recogni zed l eaders i n usi ng anal yti cal techni ques for per form- ance i mprovements al so are usi ng sophi s- ti cated experi mental desi gns to measure the overal l i mpact or l i ft of i nterventi on strategi es and usi ng these resul ts to con- ti nuousl y i mprove subsequent anal yses. Capi tal One, for exampl e, conducts more than 30, 000 experi ments a year wi th di fferent credi t card i nterest rates, i ncen- ti ves, di rect mai l packagi ng, and other parameters to maxi mi ze both the l i kel i - hood that a potenti al customer wi l l si gn up for a credi t card, and that they wi l l pay Capi tal One back. Progressi ve I nsurance employs similar experiments. The company defi nes narrow groups of customers (or cel l s)for exampl e, motorcycl e ri ders ol der than 30 wi th no previ ous acci dents, a col l ege educati on, and a credi t score hi gher than a cer tai n l evel . For each cel l , the company per forms regressi on anal ysi s to i denti fy the factors that most cl osel y correl ate wi th i ts l oss experi ence. They set pri ces for each cel l they bel i eve wi l l enabl e them to earn a profi t across a por tfol i o of customer groups. They use si mul ati on software to COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 3 test the fi nanci al i mpl i cati ons of these hypotheses. Through thi s anal yti cal approach, Progressi ve can profi tabl y i nsure customers i n tradi ti onal l y hi gh-ri sk categori es, such as motorcycl i sts. Use of Analytics across Multiple Functions and Business Units One of the hal l marks of an anal yti cal l y ori ented fi rm i s the use of sophi sti cated anal yti cs not j ust i n one busi ness functi on or process, but across mul ti pl e aspects of the busi ness. Successful anal yti cal competi tors have real i zed the power of these tool s and approaches, and are adopti ng them across thei r busi nesses. UPS, for exampl e, has tradi ti onal l y focused on anal yti cs for operati ons and l ogi sti cs. More recentl y, i t has devel oped a strong anal yti cal focus on customers, assessi ng the l i kel i hood of customer attri - ti on, or i denti fyi ng sources of probl ems for customers. Several fi rms, descri bed bel ow, are even extendi ng thei r anal yti cal ori entati ons to di rect use by customers. As we will argue later, however, there is a balance to be maintained in terms of broadening the focus on analytics, and employing them to address a specific busi- ness domain. Executives at several analytical competitors warned against losing a clear business purpose for analytics. Harrahs, for example, has targeted much of its analysis on increasing customer loyalty, although it has extended it into such related areas as pricing and promotions as well. Analytical competitors can broaden their focus beyond a narrow function, but they are careful not to become too diffuse in their analytical targeting so that they continue to support their primary strategies. An Enterprise-Level Management Approach Business intelligence applications often have been managed at the departmental level, with analytically oriented business functions selecting their own tools, managing their own data warehouses, and training their own people. However, if analytics are to be a companys basis for competition, and i f they are to be broadl y adopted across the fi rm, i t makes more sense to manage them at an enterpri se l evel . Thi s ensures that there i s a cri ti cal mass of ski l l s, that cri ti cal data and other resources are protected, and that data from mul ti pl e busi ness functi ons can be correl ated. The enterpri se approach may i ncl ude both organi zati onal and techni cal capabi l i ti es for busi ness i ntel l i gence. At the organi za- ti onal l evel , for exampl e, Procter & Gambl e recentl y consol i dated i ts anal yti cal organi zati ons for operati ons and suppl y chai n, marketi ng, and other functi ons. Thi s wi l l al l ow a cri ti cal mass of anal yti cal exper ti se to be depl oyed to address P&G s most cri ti cal busi ness i ssues. From a technol ogy standpoi nt, many fi rms have had hi ghl y di spersed anal yti cal technol ogy i n the form of many spread- sheets. However, one researcher suggests that between 20 and 40 percent of spreadsheets contai n errors. Fur thermore, the prol i ferati on of user-devel oped spreadsheets and databases i nevi tabl y l eads to mul ti pl e versi ons of key i ndi ca- tors wi thi n an organi zati on. Because of these probl ems, many fi rms are attempti ng to consol i date and i nte- grate thei r technol ogi es for busi ness ana- l yti cs. Adopti ng such an approach means I T organi zati ons must devel op new and broader capabi l i ti es for extracti ng and cl eani ng data, l oadi ng and mai ntai ni ng data warehouses, data mi ni ng, and query and repor ti ng. These tool s hi stori cal l y have come from separate vendors and have been di ffi cul t to i ntegrate. However, the l eadi ng vendors of busi ness i ntel l i - gence tools and applications are beginning to broaden and i ntegrate thei r offeri ngs themsel ves, and to market and sel l them at the enterpri se l evel . Stages of Analytical Competition Anal yti cal competi ti on i s not a bi nary attri bute, whi ch an organi zati on ei ther has or l acks. There are several stages of ana- l yti cal ori entati on that we observed i n the compani es we i ntervi ewed (Fi gure 2). The percentages of organi zati ons at these stages are by no means representati ve of any l arger popul ati on; we i ntenti onal l y sought out compani es at the hi gher end of the anal yti cal spectrum. A random sampl e of organi zati ons woul d probabl y l ook l i ke an i nverted versi on of Fi gure 2, wi th the hi ghest frequenci es at the l ower stages. Stage 1 (Maj or Barri ers) organi zati ons have some desi re to become more anal yt- i cal , but thus far they l ack the wi l l and COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 4 Stage 3 Vision Not Yet Realized Stage 5 Stage 4 Almost There Stage 2 Local Activity Stage 1 Major Barriers Analytical Competitors 0 2 4 6 8 10 12 11/32 F I R M S 6/32 F I R M S 7/32 F I R M S 6/32 F I R M S 2/32 F I R M S Figure 2: Stages of Analytical Competition among Study Organizations ski l l to do so. They face some substanti al barri ersboth organi zati onal and techni - cal to anal yti cal competi ti on, and are sti l l focused on putti ng basi c, i ntegrated trans- acti on functi onal i ty i n pl ace. As a resul t they are not yet on the path to becomi ng analytical competitors. Because we attempted to interview only organizations that compete on analytics, we encountered onl y two Stage 1 organi zati onsa state government agency and an engi neeri ng fi rm (and even that fi rm i s becomi ng more anal yti cal about i ts human resources). However, Stage 1 organizations probably constitute the majority of all large organizations. Stage 2 (Local Acti vi ty) organi zati ons have made substanti al progress i n becom- i ng more anal yti cal , but i t i s pri mari l y l ocal , wi thi n par ti cul ar functi ons or uni ts. Marketi ng, for exampl e, may be i denti fyi ng opti mal customers or model i ng demand, but the exampl e has not spread to other par ts of the company. Thei r busi ness i ntel l i gence acti vi ti es produced economi c benefi ts, but not enough to affect the company s competi ti ve strategy. We found si x of these fi rms. What they pri mari l y l acked was a vi si on of anal yti cal competi - ti on that came from seni or executi ves. Several of the fi rms had some of the same technol ogy as fi rms at hi gher stages of anal yti cal acti vi ty, but they had not put i t to strategi c use. The organi zati ons i n Stage 3 (Vi si on Not Yet Real i zed) do grasp the val ue and the promi se of anal yti cal competi ti on, but they are a l ong way from actual l y suc- ceedi ng wi th i t. We found seven organi za- ti ons i n thi s posi ti on. Some onl y recentl y have ar ti cul ated the vi si on, and have not begun i mpl ementi ng i t. Others have very hi gh l evel s of functi onal or busi ness uni t autonomy, and are havi ng di ffi cul ty mounti ng a cohesi ve approach to anal yti cs across the enterpri se. One mul ti l i ne i nsur- ance company, for exampl e, had a CEO wi th the vi si on of usi ng data, anal yti cs, and a strong customer ori entati on i n the fashi on of Progressi ve, an auto i nsurance company wi th a hi story of technol ogi cal and analytical innovation. But the company onl y recentl y had begun to expand i ts anal yti cal ori entati on beyond the tradi ti on- al l y quanti tati ve actuari al functi on, and there was l i ttl e cooperati on across the l i fe and proper ty and casual ty busi ness uni ts. Stage 4 (Al most There) organi zati ons have the vi si on, and are cl ose to achi evi ng i t. Si x organi zati ons fel l i nto thi s category. Some only recently had adopted an enterprise- wi de approach to anal yti cal competi ti on, and had yet to ful l y real i ze i t i n terms of marshal i ng the necessary resources. Others were competi ng on the basi s of anal yti cs, but al so were competi ng on the basi s of other factors, such as mai ntai ni ng strong personal rel ati onshi ps wi th cus- tomers. Onl y a smal l degree of added emphasi s on anal yti cal capabi l i ty woul d pl ace these compani es i n the top rank. The top rank i s Stage 5 (Anal yti cal Competi tors), whi ch descri bes organi za- ti ons that have embarked upon anal yti cal competi ti on as a pri mary di mensi on of strategy. These are the organi zati ons we pri mari l y sought to uncover i n our research, and therefore we i denti fi ed 11 of them. They i ncl ude such l arge and smal l organi zati ons as Apex Management Group (a heal th care actuari al fi rm), Barcl ays Consumer Fi nance, Capi tal One, Harrah s, Marri ott, Owens & Mi nor, Progressi ve, Wal - Mar t, a consumer products fi rm, and the spor ts teams, the New Engl and Patri ots and the Boston Red Sox. These fi rms exhi bi ted each of the attri butes descri bed above as the components of anal yti cal competi ti on. They are al so al l hi ghl y suc- cessful wi thi n thei r i ndustri es, and attri b- ute thei r success at l east i n par t to thei r anal yti cal strategi es. Barcl ays, for exampl e, i ncreased i ts revenue per acti ve account by 25 percent, whi l e reduci ng del i nquent accounts by 23 percent, by fol l owi ng i ts anal yti cal l y ori ented I nformati on Based Customer Management strategy. Whats the Business Value of Analytics? Anal yti cs can be used to pul l al most every l ever of organi zati onal per formance. However, we found several busi ness obj ecti ves and i ssues that were dri vi ng most of the anal yti cal acti vi ty i n the fi rms we studi ed. They i ncl ude the fol l owi ng: Customers or consumerSeveral organi - zations were focused on customer or consumer analytics, whi ch encompass a vari ety of speci fi c obj ecti ves. They mi ght i ncl ude, for exampl e, i denti fyi ng the most prof- i tabl e or desi rabl e customers, or those wi th the l owest ri sk of nonpayment. Customer analytics also include identifying the current customers who are most l i ke- l y to stop bei ng customers. They al so mi ght i ncl ude customer-speci fi c pri ci ng or product/ servi ce offeri ngs based on the customer s past or predi cted future buyi ng frequency and habi ts. Compani es that pursued thi s set of anal yti cs among our study respondents i ncl uded Harrah s, Procter & Gambl e, Progressi ve I nsurance, Barcl ays, and Capi tal One. Supply chainAnal yti cs for l ogi sti cs and the suppl y chai n are wel l -establ i shed i n many l arge fi rms, wi th the pri mary ori entati on usual l y bei ng reducti on of in-process inventory. Supply chain analysis al so mi ght encompass matchi ng demand and suppl y, routi ng shi pments around l ogi sti cal probl ems, reduci ng stockouts and overstocks, al ternati ve suppl y si mu- l ati ons, pl ant and di stri buti on center si ti ng deci si ons, and pri ce opti mi zati on. Among the compani es i n our study, Wal -Mar t i s the l eadi ng exponent of suppl y chai n anal yti cs. Financial performance and cost managementOne domai n of busi ness val ue for anal yti cs can revol ve around per formance management. Moni tori ng and deci si on maki ng on fi nanci al i nfor- mati on i s not often thought of as a COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 5 competi ti ve strategy, but i t can be. At MCI , (the company once known as Worl dCom whi ch recentl y emerged from bankruptcy and was acqui red by Veri zon), managi ng the busi ness wi th accurate i nformati on on costs and their allocations is crucial to the companys strategy. CEO Mi chael Capel l as and the management team reasoned that the company coul dn t restore i nvestor confi - dence or make good deci si ons on prod- ucts and servi ces wi thout a better noti on of the company s costs. Most of MCI s servi ces run over the same network, so al l ocati ng costs to servi ce offeri ngs i s difficult. MCI embarked on a major activity- based costi ng i ni ti ati ve, and devel oped al gori thms for al l ocati ng al l costs. The company needed to repor t segment profi tabi l i ty anyway, and wi th accurate cost al l ocati ons, managers can make effecti ve deci si ons about what servi ces to launch and what resources are needed to suppor t them. Research and new product/service developmentPerhaps the most acti ve use of anal yti cs i n research and product devel opment i s i n the pharmaceuti cal i ndustry. We i ntervi ewed three pharma- ceuti cal fi rms (Mi l l enni um, Novar ti s, and Ver tex), each of whi ch was attempti ng to conquer the overwhel mi ng compl exi ty of rel ati ng chemi cal , cl i ni cal , and genomi c data. Many pharmaceuti cal fi rms have embarked upon di scovery techni ques i nvol vi ng hi gh-throughput screeni ng, whi ch yi el ds an enormous amount of data and a need to anal yze and make sense of al l of the experi mental resul ts. No fi rm has yet mastered al l of these compl exi ti es, but stati sti cal model i ng and anal ysi s i s a focus for anyone who hopes to compete i n the i ndustry. We al so found evi dence of anal yti cs i n new product devel opment i n the fi nanci al servi ces i ndustry. Brown Brothers Harri man, for exampl e, i s empl oyi ng anal yti cal model s for i ts i nsurance i ndustry cl i ents to model ri sk-adj usted opti ons for asset al l ocati on. Thi s servi ce had not previ ousl y been avai l abl e i n the i ndustry, and i t has brought BBH a consi derabl e amount of new busi ness among i nsurance fi rms. Si mi l arl y, The Har tford was the fi rst to market an opti ons-based annui ty product that accounted for a si gni fi cant i ncrease i n year-to-year revenues. Strategic planningSeveral of the fi rms we i ntervi ewed are usi ng stati sti cal anal ysi s for the fi rst ti me i n strategi c pl anni ng. Thei r obj ecti ve i s to determi ne what markets and customer types to address wi th what products and servi ces. I n the i nsurance i ndustry, for exampl e, whi l e pri ci ng deci si ons are made on detai l ed anal ysi s of actuari al ri sk and l oss patterns, strategi c pl anni ng often has been purel y i ntui ti ve. One i nsurance fi rm we i ntervi ewed (The Har tford), however, i s usi ng anal yti cs to assess new busi ness oppor tuni ti es, consi deri ng market segmentati on, economi cs, ri sk- adj usted returns, and the cost of capi tal for the oppor tuni ty. Capi tal One i s usi ng detai l ed anal yti cs to assess what fi nan- ci al products to offer customers i n addi - ti on to credi t cards; auto l oans are one exampl e of a product that was tested extensi vel y before a broad rol l out, and i t has become a profi tabl e busi ness for the company. Human resourcesSeveral fi rms men- ti oned they were begi nni ng to do human resource pl anni ng wi th anal yti cs, but the most aggressi ve users today seem to be professi onal spor ts teams. Both the Boston Red Sox and the New Engl and Patri ots, for exampl e, use stati sti cal anal ysi s to i denti fy the most promi si ng pl ayers to draft. The Patri ots have been par ti cul arl y successful i n thi s regard, usi ng anal yti cs to stay bel ow the stri n- gent sal ary cap i n the Nati onal Footbal l League. I n fact, the Patri ots have won three of the l ast four Super Bowl s wi th a rel ati vel y l ow-cost payrol l , the 19th hi gh- est i n the l eague i n the 2004-5 season. The team ranks potenti al recrui ts on the basi s of i ntangi bl e attri butes that other teams don t assess, i ncl udi ng i ntel l i - gence, commi tment, coachabi l i ty, and a wi l l i ngness to subordi nate i ndi vi dual ego to the goal s of the team. I n Maj or League Basebal l , the Red Sox and a few other teams have adopted the anal yti cal approaches pi oneered by Bi l l y Beane, the general manager of the Oakl and A s (descri bed by Mi chael Lewi s i n the book Moneyball). These approaches i nvol ve sel ecti ng pl ayers on such factors as on- base percentage (the percent of the ti me a batter reaches base) and sl uggi ng percentage (the number of bases achi eved per ti me at bat), rather than more tradi ti onal cri teri a such as batti ng average and runni ng speed. There are undoubtedl y other busi ness areas i n whi ch anal yti cs woul d prove to be of value, but the ones above were the most common in our study. It is likely, how- ever, that many decisions previously made on intuition and hope will soon be COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 6 The most i mportant factor i n bei ng prepared for sophi sti cated analytics is the availability of sufficient volumes of high-quality data. addressed with detailed analysis. Procter & Gamble, for example, pulls together an analytical team whenever it considers the supply chain opportunities an acquisi- tion might offer to drive synergy savings. One might hope that more analytical approaches will improve the dismal record of success many companies have experienced in mergers and acquisitions. How Do Firms Become Analytical Competitors? I n order for a fi rm to become an anal yti cal competi tor, the suppl y of and demand for data and anal ysi s must be i n al i gnment. The suppl y i ssues are much easi er to deal wi th and are general l y avai l abl e i n the marketpl ace, al though thei r absence i n a fi rm i s cer tai nl y probl emati c. The suppl y factors for anal yti cal competi ti on i ncl ude the fol l owi ng: High- Quality Data The most i mpor tant factor i n bei ng prepared for sophi sti cated anal yti cs i s the avai l abi l i ty of suffi ci ent vol umes of hi gh-qual i ty data. Thi s i s l ess of a prob- l em today than i t was previ ousl y for many organi zati ons, whi ch have made substan- ti al progress i n accumul ati ng transacti on data the past several years. Whether the data come from ERP systems, point-of-sale systems, or I nternet transacti ons, many organi zati ons have a greater vol ume of data than ever before. The di ffi cul ty i s pri - mari l y i n ensuri ng data qual i ty, i ntegrati ng and reconci l i ng i t across di fferent systems, and deci di ng what subsets of data to make easi l y avai l abl e i n data warehouses (i . e. , havi ng a cl ear strategy for data access). Many organi zati ons remai n hi ghl y fragmented, and have i ssues i nvol vi ng i ntegrati on across thei r di verse busi ness functi ons and uni ts. Even such basi c poi nts as agreei ng on the defi ni ti on of who i s a customer can be probl emati cal across l i nes of busi ness. As we noted above, the l owest-ranki ng fi rms i n our scal e of anal yti cal competi ti on sti l l face si gni fi cant di ffi cul ti es wi th these basi c data i ssues. The l eadi ng fi rms, however, have l argel y overcome them. Previ ous studi es of fi rms anal yti cal capa- bi l i ti es have found even l eadi ng-edge compani es tend to be good at ei ther qual i tati ve knowl edge management or quanti tati ve data management, but rarel y both. Compani es sti l l wrestl e wi th thi s combi nati on, but we found a few more exampl es of fi rms that do both wel l par ti cul arl y i n the real m of consumer i nformati on. Procter & Gambl e hi stori cal l y has been an i ndustry l eader i n customer anal yti cs, but i t al so tri es to devel op a detai l ed understandi ng of consumer behavi ors through ethnographi c (cl ose observati on) and psychographi c anal ysi s. Wachovi a Bank combi nes knowl edge from customer rel ati onshi ps and quanti tati ve data anal ysi s of customers (pri mari l y cus- tomer segmentati on anal ysi s and market- i ng campai gn targeti ng) to determi ne what servi ces to offer a par ti cul ar customer, what markets to target, and what new i ni - ti ati ves to under take at par ti cul ar fi nanci al centers. The i mpor tance of personal busi - ness rel ati onshi ps i s deepl y embedded i n the Wachovi a cul ture, and CEO Ken Thompson i nsi sts i t remai ns there even as the cul ture al so embraces anal yti cs. Par ti cul arl y where customers are con- cerned, i t s i mpor tant to remember that marketi ng and servi ce processes i nvol ve more than the appl i cati on of stati sti cs. A Capable Technology Environment I n order to take advantage of good data, an organi zati on al so needs a capabl e hardware and software envi ronment. Compl ex anal yti cs chew up a good deal of processi ng power, so the workstati ons and servers used for thi s purpose need to be substanti al l y more power ful than those used for conventi onal offi ce tasks. Apex Management Group, for exampl e, a heal th care actuari al fi rm, i s transi ti oni ng to a 64-bi t computi ng envi ronment to deal wi th the compl ex and data-i ntensi ve sta- ti sti cal anal yses i t per forms for i ts cl i ents. An anal yti cs group at a consumer prod- ucts fi rm rented ti me on a supercomputer to do some of i ts more compl ex anal yses. From a software perspecti ve, busi ness i ntel l i gence software offers a vari ety of capabi l i ti es, i ncl udi ng data warehouse management, query and repor ti ng, data mi ni ng, and vari ous forms of stati sti cal anal ysi s. I deal l y al l these capabi l i ti es woul d be wel l -i ntegrated and easy to use. From the end user perspecti ve, ease of anal ysi s, repor ti ng, and data vi sual i zati on were often menti oned as i mpor tant i n the fi rms we i ntervi ewed. For some fi rms focused on real -ti me anal yti cs (such as real -ti me pri ci ng and yi el d management), the speed of data management and anal ysi s i s a cri ti cal factor for software and hardware. Quantitative Expertise Whi l e anal yti cal software becomes i ncreasi ngl y easy to use, fi rms that com- pete on anal yti cs sti l l requi re substanti al quanti tati ve ski l l sei ther i n-house or con- tracted from outsi de. The stati sti cal exper t, i n order to be useful , al so wi l l need to be fami l i ar wi th the busi ness probl ems i n the functi on and i ndustry; the quanti tati ve ski l l s of a good anal yst are rarel y equal l y appl i cabl e across di verse busi nesses. One pharmaceuti cal company, for exampl e, attempted to use several bi oi nformati cs exper ts to pursue anal ysi s of commerci al probl ems i n marketi ng and operati ons, and found they were nei ther moti vated nor exper t at the appl i cati ons. Whi l e sta- ti sti cal anal ysts who al so understand busi - ness i ssues al ways have been di ffi cul t to fi nd, i t i s i ncreasi ngl y possi bl e to hi re analytical expertise outside of a company even from India or China in some cases. However, some fi rms we i ntervi ewed stressed the i mpor tance of a cl ose and trusti ng rel ati onshi p between quanti tati ve anal ysts and deci si on makers. The need i s for stati sti cal exper ts who al so understand the busi ness i n general , and the par ti cul ar business need of a specific decision maker. COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 7 As one manager at Wachovia Bank put it with regard to the relationships his analyti- cal group tries to maintain: We are trying to build our people as part of the business team; we want them sitting at the business table, participating in a discussion of what the key issues are, determining what information needs the business people have, and recommending actions to the business partners. We want this [analytical group] to be more than a general utility, but rather an active and critical part of the business units success. A consumer products fi rm we i ntervi ewed hires what it calls PhDs with personality for i ts anal yti cal groupi ndi vi dual s wi th heavy quantitative skills, but also the ability to speak the l anguage of the busi ness and market thei r work to i nternal (and i n some cases, external ) customers. To fi nd these types of peopl e and devel op these types of rel ati onshi ps woul d surel y be much more di ffi cul t i n an outsourci ng si t- uati on, and vi r tual l y i mpossi bl e wi th the anal ysts hal f a worl d away from the deci - si on makers. DemandThe Critical Factor in Analytical Competition More di ffi cul t to create than suppl y i s the demand for anal ysi s and fact-based deci - si on maki ng wi thi n a company. I n the earl i est stages of anal yti cal competi ti on (Stage 1 and 2 organi zati ons), demand i s created by par ti cul ar busi ness probl ems. As anal yti cs becomes more central to the competi ti ve strategy, demand becomes more general i zed across an organi zati on. Yet, unl i ke the suppl y factors descri bed above, demandthe desi re to use anal yt- i cs as a pri mary competi ti ve factorcannot be bought i n the marketpl ace. The key demand factors we i denti fi ed i ncl ude: Willing Senior Executives Several l ower-stage fi rms we i ntervi ewed that made some use of busi ness i ntel l i - gence sai d the l ack of demand from top- l evel seni or executi ves was thei r si ngl e most si gni fi cant barri er to engagi ng i n anal yti cal competi ti on. These executi ves were more comfor tabl e wi th i ntui ti ve deci - si ons, or weren t aware of the possi bi l i ti es for anal yti cal competi ti on wi thi n thei r i ndustry. Some were not averse to anal yt- i cs, but di dn t have enough personal ana- l yti cal experi ence to base thei r strategi es on anal yti cs and fact-based deci si ons. Wi thout executi ves who want to use data and anal ysi s to make deci si ons, even the best busi ness i ntel l i gence appl i cati ons wi l l not be used. We saw several patterns of i nvol vement by seni or executi ves on the demand si de, whi ch we descri be bel ow. Some organi zati ons l eaders had the desi re to compete anal yti cal l y from thei r begi nni ng. Capi tal One, for exampl e, was created i n a 1994 I PO i n order to appl y the founders i nformati on-based strategy to the credi t card busi ness. Amazon. com was vi ewed by founder Jeff Bezos as competi ng on anal yti cs from i ts star t. I ts concept of personal i zati on was based on stati sti cal al gori thms and Web transacti on data, and i t qui ckl y moved i nto anal yti cs on suppl y chai n and marketi ng i ssues as wel l . Amazon recentl y used anal yti cs to expl ore whether i t shoul d adver ti se on tel - evi si on, and concl uded i t woul d not be a successful use of i ts resources. The vi si on of the founders of these startup businesses l ed to anal yti cal competi ti on. I n other cases, the demand for anal yti cal competi ti on came from a new seni or exec- utive arriving at an established company. At Harrah s, for exampl e, the recrui tment of Gary Loveman as chi ef operati ng offi cer, and eventual l y CEO, greatl y accel erated the companys analytical orientation and l ed to a new basi s for competi ti on competing on customer loyalty and service, rather than bui l di ng the most expensi ve casi no proper ti es. Someti mes the change comes from a new generation of managers i n a fami l y busi ness. At the wi nemaker E&J Gal l o, when Joe Gal l o, the son of one of the fi rm s foundi ng brothers, became CEO, he i ntensi fi ed the fi rm s focus on data and anal ysi sfi rst i n sal es, and l ater i n other functi ons, i ncl udi ng the assess- ment of consumer taste. At the New Engl and Patri ots Nati onal Footbal l League team, the i nvol vement i n the team by Jonathan Kraft, the son of the owner Bob Kraft and a former manage- ment consul tant, hel ped move the team i n a more anal yti cal di recti on both i n terms of on-fi el d i ssues such as pl ay sel ecti on and team composi ti on, and off- fi el d i ssues affecti ng the fan experi ence. The pri me mover for anal yti cal demand doesn t al ways have to be the CEO. At Procter & Gambl e, for exampl e, the pri - mary i mpetus for more anal ysi s i s comi ng from a vi ce chai rman. However, we di d observe two cases i n whi ch a si ngl e func- ti onal executi ve wi th a strong demand for an anal yti cal ori entati on was unabl e to change the cul ture i n that di recti on. At a consumer products fi rm, an anal yti cal l y focused marketi ng executi ve made hi s own functi on more anal yti cal , but was unsuccessful i n movi ng the enti re fi rm i n that di recti on. Another anal yti cal market- i ng and sal es executi ve at an i nformati on technol ogy fi rm was si mi l arl y unabl e to change hi s fi rm s enti re cul ture, al though other executi ves were cer tai nl y aware of hi s strongl y data-based management styl e. I n both fi rms, busi ness i ntel l i gence i s al i ve and wel l , but i t has not yet become a key el ement of strategy. Stimulating Demand Even wi th wi l l i ng executi ves, there i s often a need to sti mul ate demand on an ongo- i ng basi s. Several fi rms have created organi zati onal uni ts for thi s purpose. At Quaker Chemi cal , each busi ness uni t has a business adviseran analytical specialist repor ti ng to the head of the busi ness uni t. The rol e acts as an i ntermedi ary between COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 8 the suppl i ers (normal l y the I T organi za- ti on) and users (executi ves) of data and anal yses. The advi sers not onl y sti mul ate demand by showi ng the busi ness uni t how anal ysi s can be useful but, as i nter- medi ari es, expl ai n busi ness needs to the suppliers and ensure that business-relevant data and analysis will be provided. Wachovi a has a si mi l ar arrangement i n i ts Customer Anal yti cs group, i n whi ch anal yti cal teams are ti ed to par ti cul ar busi ness uni ts and act as par tners i n creati ng and ful fi l l i ng anal yti cal demand. Managers of these groups commented frequentl y that a rel a- ti onshi p of trust between the anal yst and the executi ve deci si on maker i s cri ti cal to the success of an anal yti cal strategy. The enemi es of an anal yti cal ori entati on are deci si ons based sol el y on i ntui ti on and gut feel . Yet these have al ways been popul ar approaches to deci si on maki ng because of thei r ease and speed, and a bel i ef that gut-feel deci si ons mi ght be better. For those organi zati ons wi thout suffi ci ent demand for data and anal ysi s i n executi ve deci si on maki ng, the obvi ous questi on i s whether such demand can be sti mul ated. I f there i s no seni or executi ve wi th a strong anal yti cal ori entati on, must the organi zati on wai t for such a manager to be appoi nted? One answer, of course, i s for an anal yti cal group to bui l d a suc- cessful track record of anal yti cal deci si ons that have pai d offa set of success stori es. However, i t can take several years to build this type of reputation. We di d fi nd some more speci fi c exampl es of attempts to sti mul ate demand. Whether they wi l l ul ti matel y prove successful i s as yet uncl ear. One reasonabl e and l ogi cal approach i s si mpl y to provi de seni or exec- uti ves wi th accurate and ti mel y i nforma- ti on and per formance measures so the facts wi l l be avai l abl e i f they choose to make fact-based deci si ons. At the tel ecommuni cati ons fi rm Veri zon, for exampl e, one executi ve s goal i s not to sti mul ate anal ysi s per se, but rather to sti mul ate a di fferent ki nd of anal ysi s by di fferent groups of peopl e. Veri zon and other fi rms ari si ng out of the Bel l System have l ong been anal yti cal l y ori - ented, but deci si ons were general l y made sl owl y and were pushed up the organi za- ti onal hi erarchy. Shaygan Kheradpi r, Veri zon s chi ef i nformati on offi cer, i s attempti ng to change thi s cul ture through conti nual exposure to i nformati on. He created a form of conti nuous scorecard i n whi ch hundreds of per formance metri cs of vari ous types are broadcast to PCs around the company, each occupyi ng the screen for 15 seconds. The i dea i s to get every- onenot j ust seni or executi vesfocused on i nformati on and what i t means, and to encourage empl oyees at al l l evel s to address any i ssues that appear i n the data. Kheradpi r feel s he i s begi nni ng to see si gns of cul tural change from the use of the scorecard. At one pharmaceutical firm where we interviewed several IT executives, there was generally little demand from senior execu- tives for analytical decision making, particu- larly in marketing. IT managers didnt have access to the decisions marketers were try- ing to make, and the marketing executives didnt know what data or analysis might be available to support their decisions. However, two external events offered opportunities to build analytical demand. One marketing manager discovered a ven- dor who showed how sales data could be displayed graphically in terms of geography on an interactive map. The companys IT executives felt the display technique was relatively simple, and they offered similar capabilities to the manager. A second opportunity was offered by an external study from a consulting firm. One outcome of the study will be a new set of performance indicators. The IT group plans to seize upon the indicators and will offer more analysis and related data to the man- agement team. These IT managers refuse to wait until more analytically oriented senior executives happen to arrive at the company. Several executives we interviewed com- mented that analytics have to be continually sold throughout an organization. Executives change, new business issues emerge, and those who need to use analytical approach- es in their jobs are not always compliant. At UPS, for example, the customer intelligence group determined customer defections could be accurately predicted based on cus- tomer data patterns and complaints, but the sales force needed to be sold on the new approach. When a potential defector is iden- tified through the use of the data, the sales- person should contact the customer to review and resolve the potential issue. Despite a high record of success, only about 75 percent of the salespeople were willing to act on the predictionsthough the per- centage is increasing. Even without full com- pliance, the analytical approach has reduced the frequency of customer defections. Harrahs has developed a centralized real-time yield management system for all of its hotels that needed to be sold to prop- erty managers. The systems decisions are based not only upon the usual room avail- ability patterns, but also on the customers loyalty level. When a customer calls for a reservation, the systems algorithms weigh a number of variables and data (including the number of available rooms, the amount of time before the planned stay, and the amount of business the customer gives Harrahs at this and other casinos) to calcu- late a price to offer to the customer. The system almost always produces higher rev- enues for individual properties when it is employed. Yet property managers usually have to be convinced the system is more effective than traditional approaches to yield management and local decision making. These examples are evidence that in order to build demand for complex analytics, man- agers and affected users need to be educated. COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 9 Bank of America is facing this issue head-on by incorporating models into its executive development programs that encourage leaders to look at the three Isinsight, intelligence, and ideaswhen looking for opportunities to grow their businesses. The program chal- lenges leaders to look more broadly at the data available to them, both data available internally as well as external data related to customers, competitors, and the broader environment. This program builds on a solid cultural and strategic foundation of using data to drive the business. Analytical Targets: The Fine Line Between Spread and Focus One challenge in using analytical capabilities to advance strategy is maintaining a balance between depth and focus. Several execu- tives commented in our interviews that a focus on particular business problems and outcomes is necessary if an analytical strategy is to be successful. There is only so much analytical expertise to go around, and only so many business problems on which ana- lytical supply and demand can be focused. Harrahs, as mentioned above, focuses its efforts on the management of customer loy- alty, and its management team is reluctant to venture very far outside of that orienta- tion. Capital One briefly diversified its appli- cation of analytics into such businesses as cellular phones and flowers, but concluded credit cards and other consumer financial services should remain its focus. Virtually every firm we interviewed that had built up its analytical capabilities finds demand for them exceeds the supply. Therefore, the use of analytical resources must be prioritized and allocated. Procter & Gamble, for example, ensures that the efforts of its Global Analytics group are devoted to issues that are highly strategic and worthy of the scarce talent. Although Wachovia has invested significantly in ana- lytical resources, it must still go through an annual planning process (with quarterly adjustments) to ensure that its initiatives are well-targeted. Customer and Supplier Use As we noted above, analytical tools and techniques are often used to enhance rela- tions with customers. The most obvious uses of customer analytics are internal, to inform decisions about internal strategies and operations. Quaker Chemical, for exam- ple, uses detailed analysis of its product performance with current customers to win new ones by offering both documentary proof of product quality and evidence of its extensive, experience-based expertise. Yet we found several of the more advanced analytical competitors offer some elements of their data and analytics directly to their customers and suppliers. Perhaps the best- known example is Wal-Mart, which uses its voluminous data and product demand analyses not only for internal purposes, but also to share with its suppliers through its Retail Link private exchange. All suppliers are required to use the system. Wal-Mart is not alone in sharing data. Progressive Insurance, for example, shares pricing dataits own and that of competi- torswith customers. The company also offers customers the possibility of lower rates if they accept a device in their cars that gathers data about driving activity. Some firms share both data and analyses with their customers. Procter & Gamble offers data and analytics as a service to its retail customers as part of a program it calls Joint Value Creation, and to its sup- pliers in order to help them improve their responsiveness and costs. The hospital sup- plier Owens & Minor provides data and analyses for its customers and suppliers, enabling them to access and analyze their buying and selling data, track ordering pat- terns to look for ways to consolidate orders, and move off-contract product purchases to a group contractfor products distributed by Owens & Minor or its competitors. The winemaker E&J Gallo provides its distribu- tors with data and analytics that lets them determine how best to convince retailers to add shelf space for Gallo wines. Finally, the Hong Kong-based Octopus Cards, a provider of electronic stored value cards for public transport, provides retailers with data on the customers who pass nearby the retailers facilities, and runs promotions encouraging customers to use the Octopus Cards for retail purchases. How Long Does the Change Take? Firms desiring to compete on analytics will naturally wonder how long it takes to implement such a strategy. The best advice is to begin working on it now, because it typically requires several years for analytical competitive strategies to come to fruition. Barclays Consumer Finance, for example, embarked upon a five-year plan to apply analytical approaches to marketing credit cards and other financial products to its customers. It takes time to refine the systems that produce transaction data, to COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 10 Weve been collecting data for six or seven years, but it's only become usable in the last two or three, with enough time and experience to validate conclusions based on the data. Manager of Customer Data at UPS make the data available in warehouses, to select and implement analytical software, and to build a robust hardware and com- munications environment. Firms planning to embark upon analytical competition should have a hardware and software plan for how they will achieve the needed capabilities. It should address such issues as the amount of data to be processed, the number of users of the analytical systems, and the speed of response necessary to meet the business need. Even more time-consuming at most firms is coming up to speed in human capabilities, to optimize business processes based on the outputs of analysis, and, in some cases, to build a sufficient body of data to support reliable predictive results. At UPS, one man- ager of customer data analytics noted that: Weve been collecting data for six or seven years, but its only become usable in the last two or three, with enough time and experience to vali- date conclusions based on data. Several executives at other firms noted that it takes time for managers to understand data and be comfortable with the analytics based on it. An analytical executive at Procter & Gamble suggested firms might begin to keep managers in their jobs for longer periods because of the time required to master analytical approaches to their businesses. One manager of an analytical group in a consumer products firm pointed out that the longevity of analytical capabilities is crit- ical to their value; his firm has been pursuing analytical capabilities for more than 50 years. This executive pointed out that not all proj- ects will be successful, so analytical groups need to build up a broad portfolio of exec- utive relationships, projects, and analytical technologies. He also suggested that short- term, project-based funding of analytical resources is inconsistent with the long-term nature of analytical competition. However, despite the difficulty and expense of establishing these capabilities, many of the firms we have identified as early adopters of analytical strategies are clear leaders in their industries. This suggests the time and trouble necessary to become ana- lytical competitors are definitely worthwhile. Summary This study has provided a glimpse into a new form of competition. Instead of competing on traditional factors, companies are beginning to employ statistical and quantitative analysis and predictive modeling as primary elements of competition. These firms have overcome the historical barriers to gathering and managing transaction data and some of the cultural resistance in organizations accustomed to gut-feel decision making, and are using com- plex analysis and data-intensive decisions to change the way they manage themselves and compete in the marketplace. They have mar- shaled both supply and demand factors for analytical competition, and are employing their capabilities across multiple functions. Opportunities for analytical competition are possible in every industry. Therefore, virtually every firm should consider how it might adopt analytical methods and capabilities. Figure 3 summarizes key action steps that firms should consider in moving toward analytical competi- tion. While not all of the steps will be applica- ble to all organizations, its likely everyone would find some of them appropriate. There is every reason to believe this approach will grow in acceptance. The necessary data will become increasingly available, and the analytical resources are increasingly accessible to all. Yet the move to analytical competitive- ness is typically a journey of several years. Companies that do not rapidly embrace these possibilities risk falling dramatically behind. No business can afford to lose its best cus- tomers, to spend more on logistics and inven- tory, to miss out on opportunities for new products and services, and to hire less capa- ble employees than its more analytically astute competitors. Figure 3: Action Steps for Analytical Competition 1. Begin to build analytical skillsIts often difficult to find individuals with the requisite quantitative and business skills. Organizations should start looking for them as soon as possible, and hire them in sufficient volume to create critical mass. 2. Get your data in shapeAnalytical environments require large amounts of high-quality data. Figure out what data you really need to advance your strategy, make sure its being gathered, and clean it up. 3. Implement analytical technologyYoull need heavy-duty hardware and software to do serious analytical work. Start putting it in place today. 4. Examine your business strategy Analytical competition requires a clear business strategy that is optimized with data and analysis. Your executives should begin to consider what key processes and strategic initiatives would be advanced if the right analytics were available. 5. Find an executive partnerSince the most difficult factor to put in place in analytical competition is demand from senior executives, you should begin to cultivate that demand by finding an exec- utive partner and embarking with him or her on some analytical initiatives. This research report is part of an ongoing research study at Babson on how compa- nies compete with analytics. The research was carried out independently, but was sponsored by SAS and Intel. To learn more about or participate in the research, contact Tom Davenport at tdavenport@babson.edu. COMPETI NG ON ANALYTI CS BA B SON E X ECUT I V E E DUC AT I ON WOR K I NG K NOWL E DGE R E S E A R CH CE NT E R 11 eg Babson Park, MA 02457-0310 USA Phone 1800882EXEC or +7812394354 Fax +7812395266 E-mail exec@babson.edu www.babson.edu/bee