Q: Andreas, you say we should carry out marketing tests where we market in some cities but not others, so we have control groups. This lets us see if the marketing was responsible for the change in results. But what cities should we use? – Perplexed in Peoria

A. I checked with Acxiom (my dear alma mater). They collect data on all US consumers (yes, including you) and use that for digital marketing.

The US Census and Acxiom correlates income, age, education, gender, and household makeup. For example, the Hispanic population is 16.7% of the total US population. So Acxiom looks for cities that have 16.7% Hispanics. We know the basic parameters of the US population (income, age, education, gender, household makeup), so just look for cities with the same percentages.

Acxiom has a list of US cities that match the general US population. They use this for marketing tests for their clients. Here’s a (alphabetical) copy for you:

  • Albany-Schenectady-Troy, NY
  • Birmingham, AL
  • Cedar Rapids, IA
  • Charlotte-Gastonia-Rock Hill, NC/SC
  • Columbus, OH
  • Eau Claire, WI
  • Eugene-Springfield, OR
  • Grand Junction, CO
  • Greensboro–Winston-Salem–High Point, NC
  • Nashville, TN
  • Odessa-Midland, TX
  • Pittsfield, MA
  • Richmond-Petersburg, VA
  • Rochester, NY
  • Syracuse, NY
  • Wichita Falls, TX
  • Wichita, KS

How to use this list? Build your marketing campaign. Pick (say) ten of these cities as your market (it doesn’t matter which; they’re pretty much identical). The others are the control group, i.e., no marketing. Launch your campaign, collect data, and compare.

If you find that your product sells in Rochester, you’ve got a winner: it’ll sell to the US population as a whole.