Attribution

Baseline Surveys: How To Reach Those Hard-To-Measure Spots

by Matt Bahr

You strap on your Apple Watch and head out for a grueling uphill run. You reach the summit and look down at your watch to see a heart rate of 150.

Is that… significant?

You wouldn’t know what “150” meant if Apple didn’t also provide your resting heart rate for reference -- also known as a baseline. And that’s exactly why baselines exist:

  1. To quantify the current state before any changes are made, especially when there is an assumption that the current state is something other than zero

  2. To provide a frame of reference and significance for the impacts of any changes made

Since surveys are our business, we’re going to offer some guidance here on how to use them to set your attribution baselines when evaluating new marketing channels or campaigns. This is especially useful for offline channels, which are exploding in the DTC space lately.

A baseline survey, for our purposes here, informs the natural response rates for choices that may soon shift due to a marketing initiative. You run baseline surveys beforehand (often 30-60 days prior to a new initiative), so you can have a more accurate reading on lift after the fact. For example:

You want to start running podcast ads. You know there are probably some shows talking about your brand already (earned media), and you assume that a well-executed podcast ad may also boost Word Of Mouth responses which ought to be credited at least in part to the ad. Those educated assumptions are exactly what ought to lead to a baseline survey, because you want to know:

  • Whether your current podcast attribution is something other than zero, in which case the baseline will tell you how much of your upcoming campaign’s attribution can be chalked up to measurement error

  • What your current Word Of Mouth attribution trend is, so you can see what kind of secondary effects your podcast campaign has beyond the direct conversion of a vanity URL + referral code

So, 30-60 days before you run your ad on This American Life, you start serving up post-purchase attribution surveys that include these choices among your other options, if they didn’t already exist:

  • Word Of Mouth
  • Podcast

And your baseline survey produces these results:

  • Word Of Mouth: 26.2%
  • Podcast: 0.4%

Of course, you weren’t running any podcast ads yet, so the 0.4% is a mix of organic/earned podcast mentions and response error. Now you’ve got your baselines for these two channels, and after you start running ads on This American Life, your new results look like this:

  • Word Of Mouth: 31.3%
  • Podcast: 6.6%

Interesting! You can make some working assumptions that roughly 0.4% of that podcast response shouldn’t be attributed to your campaign, while the 5% increase in WoM should be at least somewhat attributed to the pod ads (assuming you didn’t launch into other media campaigns at the same time). The more often you bank your baseline results before making adjustments, the more measurable and predictable lift will be, and the more you’ll understand about the ROI of new campaigns and channels.

 Ready to better know your customers?

See Interactive Demo