Getting Creative with Analytics

  • Dstillery is a programmatic data and technology platform – we translate behavioral and geolocation data into digital intelligence for media buying and insights
  • We ran a random experiment. We chose 3 creatives and a PSA.  The creatives were purposely chosen for being extreme. A local NY artist, Fleer. A Klimt. A Rousseau.  We measured clicks and we were encouraged to see differences based on creative (normally clicks are not a good proxy for behavior change).
  • Overall – which one resonated?
  • But the test was about how we individually react – about the micro cultures which exist within each brand.  Why does this matter?  A creative that resonates positively with one audience may resonate negatively with another.
  • This research comes from causal inference experiments run by my Dstillery colleague when he was working in pharma.  Essentially he found that although one drug was successful overall – if you looked at the results by gender – females were actually worse off.
    • HH with children – Football
    • Visual Artists – Football
    • Gamers – Fleer/Bird
  • Can we do this with a brand?  can we test banner ads? can we find a brand that has the ability to run an experiment on digital creative?  We met with 12 different brand CMOs and as you can  imagine – we had great resonance with the idea but harder to put into practice.
  • We have found that creative matters and micro audiences matter.  By targeting to Chobani audiences with optimal creative, we were able to drive a 20% increase in offline sales.