Measuring ROI: A Nonlinear Process

Twitter is an odd little application, isn’t it? One of the ways I encourage participation on Twitter is to ask a “Question of the Day.” I offer followers the opportunity to voice their opinions on a number of topics.

I provide background for the questions by offering five or six tweets prior to asking them. This strategy seemed to be working well — the number of followers was increasing and participation in each question was at an all-time high.

And then a minirevolt emerged. A segment of the audience informed me that my use of Twitter wasn't appropriate for the microblogging platform. Upon learning this, I decided to ask all of my followers if my use of Twitter was appropriate. Fifty-five percent of the followers who responded said that my use of Twitter wasn’t as effective as it could be.

So I did something different. For the following week, I strictly adhered to the advice offered by my followers, because, after all, these folks were using the microblogging platform, so they must know what works and what doesn’t. That week I simply asked my questions without context, using Twitter as recommended by the audience of the microblogging service.

A funny thing happened when I made the change in strategy: People stopped interacting with the questions. Without pretweet context, the conversation died.

When measuring return on investment, marketers tend to focus on a very linear process. They evaluate best practices, apply their brand perspectives to those best practices and then tie revenue to the marketing process.

Increasingly, however, measuring ROI has become a nonlinear process. Best practices yield a 50 percent chance of providing an acceptable ROI; listening to your customer base yields a 50 percent chance of providing an acceptable ROI. Your strategies could yield a 50 percent chance of providing an acceptable ROI. As a result, you're multiplying a probability of success by a probability of success by a probability of success, yielding a low probability of success.