Measuring out-of-home ROI with branded search traffic

By way of @nicvjayne, the story of MailChimp’s billboard advertising:

When return on investment is measured by delight instead of sales or conversions, there’s a lot more freedom to be creative, to be bold, or maybe even to be creative and bold. It was liberating to begin the project knowing that our metrics were much closer to Bhutan’s Gross National Happiness Index than the kind of metrics that analysts or shareholders of a public company might expect.

In their words: “we just wanted to make MailChimp users smile.” Which begs the question, how many MailChimp users did they make smile? Their answer: count the number of tweets. Given their success, MailChimp can afford this luxury and do things “just because.” But suppose you wanted to measure the impact of billboards or, more generally, out-of-home advertising?

One place to start is search traffic. More specifically, branded search traffic broken down by geography. Since MailChimp initially targeted San Francisco and NYC, incremental lift can be estimated by comparing those markets versus other representative metros. Here’s an example analysis with fake data. By measuring changes in branded search traffic, we can establish a baseline rate of change and assume that incremental gains above and beyond were impacted by out-of-home ads.

While this type of model cannot promise exact ROI, it’s a useful framework for analyzing non-digital performance of brand advertising. The same concept can be used to quantify the impact of upstream marketing channels (discussed previously) or any formats that are easily segmented by geography (e.g. radio and television).