It goes with­out say­ing that most ad agency busi­ness devel­op­ment man­agers need to per­suade clients that the agency can real­ly dri­ve online adver­tis­ing sales. But in a world where Big Data has become tru­ly colos­sal, how do you back up those promis­es with con­vinc­ing data? Dig­i­tal ad tech­nol­o­gy start­up Rock­et Fuel may have the solution.

Big Data meets Arti­fi­cial Intel­li­gence 

Launched in 2009, the com­pa­ny went pub­lic in Sep­tem­ber last year and now has offices in New York, Chica­go, San Fran­cis­co, Los Ange­les, Lon­don and Paris (it’s head­quar­tered in Red­wood, Cal­i­for­nia). At a moment when oth­er ad tech firms were see­ing under­whelm­ing IPOs (think YuMe and Tremor Media, for exam­ple), Rock­et Fuel’s shares dou­bled on the first day of trad­ing. Its mar­ket val­u­a­tion then was $1.6 bil­lion. And it’s attract­ed some big names in a crowd­ed and com­pet­i­tive mar­ket­place, includ­ing Amer­i­can express, Dell, Microsoft and Nike.

The more inquir­ing busi­ness devel­op­ment man­ag­er will want to know what Rock­et Fuel has got that its numer­ous rivals lack. Everyone’s into Big Data these days, after all. Accord­ing to its co-founder and VP of Engi­neer­ing Abhi­nav Gup­ta, not only can the firm’s soft­ware ana­lyze mind-bog­gling­ly vast amounts of data in an instant, it learns from its analy­ses too.

He says:

“At the start, we may not have much idea about the mar­ket, but in a few days or weeks, the sys­tem starts learn­ing by itself about what works for this cam­paign. This is our com­pet­i­tive advan­tage, lever­ag­ing machine learn­ing and arti­fi­cial intel­li­gence, on petabytes of data.”

For any busi­ness devel­op­ment man­ag­er who’s uncer­tain exact­ly how much a petabyte is, it’s a ver­ti­go-induc­ing one mil­lion giga­bytes. That’s how big Big Data has become these days.

Dynam­ic cam­paign shaping

Rock­et Fuel does on track­ing data after an ad appears, keep­ing its fin­ger on the pulse of pos­i­tive out­comes (like pur­chas­es) for every adver­tis­ing dol­lar spent, data which imme­di­ate­ly gets fed in its intel­li­gent, active­ly learn­ing soft­ware. And that gives adver­tis­ers a very clear pic­ture of the suc­cess of a campaign.

As Gup­ta puts it:

“If we learn from the con­fir­ma­tion page that the user has bought some­thing, that informs our soft­ware: ‘Hey, good job, go find more users like this one!’ That’s machine learn­ing. Look at each pos­i­tive out­come, and do more of what­ev­er pro­duced that result.”