Most website owners worry about engagement, which they should do, because if people aren’t engaged with your site, they won’t do the things that you really want them to do.
More on that in a bit, but first, let’s say you wanted to find the most engaged users on your website. How would you do it?
You could start with a simple assumption about what the word “engagement” means, and you could assemble a list of factors that might go into an engagement score. Then you’d assign point values to those factors, and you’d sum those points for each visitor. This is a relatively easy task for a customer data platform.
You might start a project like this by defining what you mean by “engagement.” It’s something like “frequent and repetitive use of your service.” So you’d make a list of things that indicate that. But some of those factors might have more significance than others.
For example, landing on an article page could be worth one point in an engagement score, but scrolling to the bottom is 3 points. Using the search function is 5 points, and answering a quiz is 10 points. And so on. This simply becomes a summing exercise for your CDP. It’s very straight-forward.
But are you really capturing the right data? You have to wonder whether you have the right collection of factors and whether you’ve given them the right point value. Should search really get 5 points? Why not 4, or 6?
And how do you know if your calculation yields the right answer? You’ve decided that people with top scores — based on your point system — are the most engaged, but how do you verify that?
It would be nice if there was some objective way to determine engagement so you could test your scoring system. But there isn’t. So what you have to do is back up one step and ask why you care about engagement in the first place. Why is engagement valuable to your company?
If you run an ad-based site, customer value might be as simple as page views. If you sell things, it’s how much does somebody buy, or do they subscribe, and then renew. Or maybe you want people to come to your restaurant or bar.
“Engagement” itself is not the final metric. It has to correlate with something that puts money in the bank.
Now we’ve taken a step beyond engagement. What we really want is some engagement-like score that correlates with a high-value customer.
That’s a job for AI.
Rather than assigning your own values to some limited list of factors, you start off with as many factors as you can imagine that might be related to high-value customers. Be creative with your list, because the thing that really matters might not be obvious.
Then you let AI find the patterns and figure out what scores to assign to each factor, testing those scores against your actual goal, which is more page views, or more purchases, or visits to happy hour, or whatever it is you really want.
If you’re interested in pursuing something like this, give me a call. I can help you get started, and I have some friends who specialize in this sort of thing.
