Adobe Digital Dialogue

December 23, 2014 /Data Driven Marketing /

Data driven marketing – getting started guide

 Scott Thom­son, Senior Man­ag­er Indus­try Strat­e­gy APAC, Adobe Sys­tems

In most enter­pris­es today, IT and mar­ket­ing func­tions have vary­ing but sig­nif­i­cant invest­ments in data ware­house, busi­ness intel­li­gence, cus­tomer rela­tion­ship man­age­ment and cam­paign man­age­ment sys­tems. All this accu­mu­lat­ed invest­ment in tech­nol­o­gy will be pro­duc­ing cam­paign lists for out­bound call cen­ters, direct mail and poten­tial­ly elec­tron­ic direct mail.

Scott's Data driven marketing blog

Enter­pris­es mar­ket to dif­fer­ent cus­tomer seg­ments about dif­fer­ent prod­ucts. These seg­ments may be deter­mined by prod­uct hold­ings, prod­uct usage, geo demo­graph­ic or behav­iour­al pro­files. The tech­nol­o­gy, process­es & skills to cre­ate cam­paign lists already exist with­in the enter­prise. Known cus­tomer seg­ments can be matched to known prod­uct offer­ings and cam­paigns can be mea­sured in terms cam­paign respons­es and point of sale out­comes.

With very lit­tle addi­tion­al effort, exist­ing cam­paign lists can be re-pur­posed in dig­i­tal. This has two effects: first, it will pro­vide a marked improve­ment over any cur­rent “spray & pray” dig­i­tal media buy­ing strat­e­gy; and sec­ond, it will pro­vide us with a means to cre­ate a known audi­ence and an unknown audi­ence in dig­i­tal. The unknown audi­ence can then be seg­ment­ed based on look-a-like char­ac­ter­is­tics based on a geo demo­graph­ic or behav­iour­al basis. Either way, the known data and look-a-like seg­ments should pro­duce an imme­di­ate uplift over ran­dom reach-based, prospect­ing approach­es.

Scott's Data driven marketing blog (2)

The advan­tage to this approach is that it cap­i­tal­izes on pri­or invest­ments in data ware­hous­ing, busi­ness intel­li­gence, cus­tomer rela­tion­ship man­age­ment and cam­paign man­age­ment as well as pre-exist­ing teams, process­es and mes­sag­ing rules to engi­neer a quick and pos­i­tive uplift result.

The trick to this approach is get­ting a non-per­son­al­ly iden­ti­fi­able ID that will allow you to link your cam­paign lists in 3rd par­ty dig­i­tal sites. This is rel­a­tive­ly sim­ple if you have a dig­i­tal site or app that requires your cus­tomers to login in or authen­ti­cate in some fash­ion. For instance, they might login in to check their account bal­ance, view and pay accounts, make trans­ac­tions, post user gen­er­at­ed con­tent, make com­ments or any oth­er sort of authen­ti­cat­ed activ­i­ty.

When your vis­i­tors authen­ti­cate them­selves, they give you a means to tie their dig­i­tal activ­i­ty to your data ware­house data. You can allo­cate them a unique but not per­son­al­ly iden­ti­fy­ing ID and you can then exchange this iden­ti­fi­er with 3rd par­ty data part­ners allow­ing you to find these same peo­ple on 3rd par­ty sites. The most com­mon means of exchang­ing IDs is by cook­ie sync­ing. Cook­ie sync­ing essen­tial­ly maps your cook­ie iden­ti­fiers to those of anoth­er site and is used exten­sive­ly in the dig­i­tal ad tech­nol­o­gy ecosys­tem. (The process of cook­ie sync­ing is explained in more tech­ni­cal detail by Ad Ops Insid­ers if you are inter­est­ed in under­stand­ing more.)

If you don’t have dig­i­tal sites or apps that encour­age vis­i­tor authen­ti­ca­tion (via login or social sign in), all is not lost. You can still allo­cate a unique but not per­son­al­ly iden­ti­fy­ing ID in a cook­ie. You will be able to tie brows­ing behav­ior on your sites to this cook­ie, retar­get this brows­er on 3rd par­ty sites, aug­ment what you know about this brows­er from 3rd par­ty data sources and cre­ate look-a-like audi­ences. You just won’t be able to tie this brows­er to data or cam­paign seg­men­ta­tion from your data ware­house with­out some form of authen­ti­ca­tion.

In either case, the next steps are rel­a­tive­ly straight­for­ward but are each wor­thy of blog posts on their own…

  1. Trans­fer your exist­ing cam­paign lists on a month­ly, week­ly or night­ly basis to a data man­age­ment plat­form (DMP) using your non-per­son­al­ly iden­ti­fi­able iden­ti­fi­er as a the key
  2. Set up pro­gram­mat­ic buy­ing such as bids on demand side plat­forms to buy against these cam­paign lists
  3. Expose these same cam­paign lists to your dig­i­tal sites & apps to allow test­ing & opti­miza­tion to oper­ate on your data ware­house data and 3rd par­ty data obtained from part­ners.
  4. Aug­ment your audi­ence seg­ments with real-time ana­lyt­ics and aug­ment your ana­lyt­ics with data from your data man­age­ment plat­form, bring­ing your data ware­house data and 3rd par­ty data into your ana­lyt­ics reports.
  5. Use data from your data ware­house, ana­lyt­ics, data man­age­ment plat­form and pro­gram­mat­ic buy­ing to devel­op weight­ed attri­bu­tion mod­els that pro­vide opti­mal media mix­es for com­bi­na­tions of prod­ucts and cam­paign seg­men­ta­tions.

Scott's Data driven marketing blog (3)

The impor­tant thing is to get start­ed. You very like­ly have a lot of infra­struc­ture and process­es already in place and with a rel­a­tive­ly small amount of effort and invest­ment you can get start­ed in data dri­ven adver­tis­ing. Use small tri­als with pre-exist­ing and high per­form­ing seg­ments to get hard num­bers to present to key stake­hold­ers. Use these ear­ly suc­cess­es to dri­ve fur­ther invest­ment and before you know it, you will have trans­formed your exist­ing dig­i­tal mar­ket­ing spend from a broad­cast based prospect­ing mod­el to tar­get­ed media buys gen­er­at­ing uplift in per­for­mance.

Scott's Data driven marketing blog (4)

Data Driven Marketing

Join the discussion

  • By MJ Bartram - 12:42 AM on September 28, 2016   Reply

    Great arti­cle, Scott, par­tic­u­lar­ly the dia­grams. It suc­cinct­ly and clear­ly states the case for data-dri­ven mar­ket­ing, some­thing we’re striv­ing to imple­ment. I’ve shared with my team. (FYI- Love the sketch­es, but note that the word RELATIONSHIP is mis­spelled in all of them as RELEATIONSHIP. 😉

Join the discussion