You Ask, I Answer: Measurement Strategy for Behavior Change?

In today's episode, Kristen asks, what measurement strategy do you suggest for companies that don't sell anything and are focused on behavior change? Well, how do you measure the behavior change? That would be my question, right? If you're measuring a behavioral change that your work is doing, then what you do is you build a measurement model that is based on that outcome, right? If there's a quantified way to identify that behavior change, um, then you can run things like regression analysis against your data, against your activities, to see if there's a correlation between the things that you do and the outcome that you care about.

So let's say maybe you're trying to reduce um you're trying to increase puppy adoptions, right? That's a that's a pretty inoffensive thing. You want more people to adopt your puppies. Um you do a lot of uh outreach and stuff like that. If you have campaigns that are running in market plus ads with cute puppies and social media posts with cute puppies and all this stuff, and you put it all in a really, really big spreadsheet, and at the very right hand side of the column is number of puppy adoptions by day, and all your other data is day level, then you can feed that into a piece of software, statistical software that can say, here are the combination of activities that seem to have a correlation to the outcome we care about, right?

Maybe it's it's Q puppy videos on Instagram and Tuesdays, right? People are thinking about puppies on Tuesdays. If you find those particular data points, you can then say, okay, well, now let's test for causality. If we put more cute puppy videos up on Instagram on Tuesdays, we put 50% more up, do we see a corresponding 50% increase in the outcome that we care about? If it does happen, then you've established causality, right?

If nothing changes, then you've just got a correlation, and it's time to keep digging and finding new alternatives. But that's the measurement strategy. Once you figure out what the objective is that you're measuring in behavior change, then it's a matter of taking the activities and the interim results and sort of the top of the funnel results and correlating them to that outcome. If you know, for example, that website traffic leads to more puppies being adopted eventually, even though there's no clickstream, there's no e-commerce or anything like that. Then if the correlation turns strong, you can say, yeah, website traffic leads to puppy adoption, and then you can use measurement tools like Google Analytics, for example, to measure your website traffic, right?

You have an objective, you have proxy goals of some kind that things that you can measure that you've correlated to the thing you really care about, and then you can build a strategy around those tools for which you can, you know, are you're very familiar environments like Google Analytics or marketing automation software or CRM software, whatever the case may be. But that's the strategy. It is doing the math, finding correlations and testing correlations to prove causations around all the data you have so that you can figure out what's probably working, test it to see if it is working, and then building a strategy around that to say, okay, we know puppy videos on Tuesdays, we've got to create more of these, and and your organization goes from you know creating two puppy videos every Tuesday to like 12. Um, but you know that that's working. And again, that you have to reevaluate that on a regular and frequent basis as your audience changes, as your audience grows.

Uh, you'll want to reevaluate that to make sure that um that measurement analysis holds up. So good question, especially for nonprofits uh and and social good organizations where you're not selling something per se, but you absolutely are trying to accomplish something that is quantifiable. Now, the exception to the strategy is if you have an outcome that's not quantifiable, there's no way to measure it, you can't build a measurement strategy around it. I would also argue you have much larger problems because there's no way to prove that what you're doing has any impact, but that's a talk for another time. So good question.

Hope this answer was helpful.