Welcome to the latest installment of our series on segmenting best practices and use cases. Over the last two weeks, we’ve covered a lot of ground, including using segmentation to influence some of the leakiest leaks in the retail revenue bucket: increasing purchase frequency and preventing churn. But guess what. Churn still happens. So today, we’re going to discuss what to make your win-back efforts as effective as possible with good segmentation practices.
How Do You Win Them Back?
Let’s say you haven’t had the chance to optimize your churn prevention strategies (as described in our last post) or that you’ve got out-of-this-world retention rates thanks in large part to your scary effective churn prevention campaigns.
Even still, no one maintains a 100% retention rate. Great customers slip through the cracks, and they’re long gone despite your best efforts. How do you win them back? How do you want to reach out to that group?
You can split them into demographics again, the same story we’ve told throughout the series. That could lead you to target them in certain ways, but it’s a strategy with a lot of drawbacks.
You could look at first purchase or most recent purchase — this could tell you what they’re interested in and what might tempt them back into the fold. This has potential upsides. It might be useful information some of the time. But it's not quite telling you what’s going to reactivate that unique user. Maybe the last thing they bought was what turned them off to your brand! Maybe if you look at their entire purchase history, you might see that the first or last thing they bought was an out-of-character purchase, and therefore likely to not be relevant for a win-back message.
So, do you try to push what they just bought? Do you try to push something else? One way to find the answer is to to get into that experimentation loop to find better and better solutions.
But if you’re trying to win back a customer who’s bought four, five, six times with your brand, you have a ton of information at your fingertips.
Enter the Machine!
There are some really interesting segmentation tactics that you can use, clustering tactics, that think of customers more holistically. Not just in a sense of what did they buy first, what did they buy last, or what did they buy most often. Instead, the machine looks at a series of customers and finds the patterns that create buyer profiles.
Defining “type” as based on the collection of things a customer buys, you can look at your repeat buyers and ask the data, "Hey, what are the different types of shoppers I have?" In a fashion company, maybe there are some types of customers who tend to buy shirts and skirts, but there's another type who buys fancy jeans and boots. There's another type who buys accessories, a few shirts, and a lot of jeans.
The above-mentioned are all different types of customers that really bring to life who your customer is and what you should show them in your win-back campaigns. We refer to these things as persona segmentations.
So, for instance, the results of this kind of segmentation might look like this:
- Customer Type A: 30% jeans, 65% boots, 5% mix of other things.
- Customer Type B: 50% shirts, 48% skirts, 2% mix of other things.
- Customer Type C: 30% accessories, 30% jeans, 20% shoes
What's great about these segments is not only have we grouped customers into these types based on their longitudinal behavioral data, all of the data we've gathered on them from these powerful purchase moments over time, but they're also prescriptive.
If you’re looking at a customer who falls strongly into Type B, and they're fading away or they’re long gone, you don’t have to wonder for very long what products to show them to get the highest probability of their returning. You send an email that highlights your most recent shirts and skirts in stock. The same logic goes for any segments based on purchase history. Because these segments are a pretty sturdy snapshot of your business, you only need as many creative assets and you have segments, and deployment can be automated based on specific triggers.
This is when you get into the real power of the best segmentation. You not only identify people who are behaviorally similar whom you wouldn't have been able to see without running these models, but you get instructions on how to better connect with these customers. You can really see response rates, conversion rates, long-term lifetime value of customers go up when you discover ways to show customers these things that they're very interested in, looking beyond just single point-in-time moments.