Most shoes that Amer­i­cans wear are made in Asia and have a six-month cycle from design to sale. That process, includ­ing the man­u­fac­ture of the shoes, has been the same for 60 years, accord­ing to David Roberts, VP of enter­prise trans­for­ma­tion at Under Armour. The retail­er, which gained fame for dis­rupt­ing work­out cloth­ing, is try­ing to do the same with shoe sales.

Under Armour is now sell­ing shoes with 3‑D print­ed mid­soles direct­ly to con­sumers — shoes that are made on demand after an online order and shipped with­in two days. That’s a cul­mi­na­tion of a trans­for­ma­tion process that began with Under Armour pur­chas­ing many of the major play­ers in the fit­ness track­ing app mar­ket, then con­nect­ing those apps to the shoes it was sell­ing to cre­ate a trea­sure trove of Inter­net of Things (IoT) data. The result? A shoe built with the col­lect­ed knowl­edge gained from pre­vi­ous shoes. 

“We wouldn’t be able to do this with­out the data and the insights that we glean from [fit­ness apps],” said Roberts, who spoke at the 2018 BI + Ana­lyt­ics con­fer­ence in Hunt­ing­ton Beach, Calif. He adds that one exam­ple of an insight Under Armour gath­ered from the con­nect­ed shoes is that its cus­tomers often use cross-train­ing shoes for running. 

Roberts explained that the 3‑D print­ed shoes can be pro­duced at 2,400 per hour, com­pared with 2,400 per eight hours at fac­to­ries. The cost to Under Armour moves from $24 per pair out of the fac­to­ry to 30 cents per pair using the on-demand 3‑D print­ing method. Inter­est­ing­ly, the sav­ings doesn’t seem to have made it to con­sumers just yet, as the shoes with 3‑D print­ed mid­soles retail for almost $300, well above the aver­age fac­to­ry-made Under Armour shoe. Per­haps the ini­tial cost of the 3‑D print­ers affect that price in some way. 

Ana­lyt­ics Dri­ve Transformation

In using ana­lyt­ics to build prod­ucts, Under Armour ana­lyzes its own data from its SAP sys­tem but adds non-SAP sources to get a more thor­ough pic­ture of consumers. 

“Much of our oper­a­tions runs on SAP,” said Roberts, ​“But more impor­tant is that much of the data is out­side of SAP and out­side of Under Armour.” 

Under Armour uses SAP Vora to con­nect its Hadoop clus­ter lay­er to its SAP HANA instance. On top of that, the retail­er lever­ages SAP Dig­i­tal Board­room, SAP Busi­nes­sOb­jects Design Stu­dio, and Crys­tal Reports for analysis. 

“You want to make sure to give the busi­ness the tools to answer ques­tions they didn’t think to ask — and make sure they are not behold­en to the IT team to deter­mine what ques­tions should be answered,” assert­ed Roberts. ​“That’s changed the busi­ness, it’s rev­o­lu­tion­ized what’s been done.” 

While build­ing the design for its 3‑D print­ed shoe, Under Armour changed from its tra­di­tion­al approach of seek­ing feed­back direct­ly from con­sumers. Instead, it took to social media analy­sis using the nat­ur­al lan­guage pro­cess­ing capa­bil­i­ties of SAP HANA. Essen­tial­ly, Roberts said, the final prod­uct was com­bin­ing all the suc­cess­ful pieces of pre­vi­ous Under Armour shoe designs. 

Peo­ple Still Mat­ter in Analytics

Agile devel­op­ment dri­ves Under Armour’s trans­for­ma­tion — the com­pa­ny works in six-week invest­ment cycles. The oper­at­ing mod­el that drove the fit­ness app acqui­si­tions was built in six weeks, and now pro­duces five per­cent of the company’s total revenue. 

Under Armour also scores the data it uses by source — not all data is cre­at­ed equal. For exam­ple, the high­est lev­el of trust goes to SAP ERP data fol­lowed by data that the retail­er pays to acquire. On the oth­er end is social media data — which isn’t worth­less, but Under Armour tries to take the nec­es­sary grain of salt when using it for analy­sis that dri­ves prod­uct deci­sions. That’s where the peo­ple, not just the machines, are still important. 

“At the end of the day, you still have to have a per­son look at the ana­lyt­ics,” said Roberts. ​“Tech­nol­o­gy gets you to the table, but you have to decide how to break the data down.”