The Problem With Almost Every Retention Program
your segmentation is a goddamn mess
Hi Team!
Spent a few days in Panama end of May, and 10 days in Israel at the beginning of June. Got hit with jet lag, a stomach bug, and the strong willingness to be a proper homebody. (at least until the Toronto trip in July and the Sicily/Malta trip in August 🙃)
Something that has come up a lot in conversations over the last few weeks with brands I’m chatting with is segmentation and its importance in overall retention efforts.
Most DTC brands have gotten really good at segmenting customers by order history. You’ve got your one-time buyers, your repeat customers, your high AOV spenders, maybe a VIP tier if you bought more than three times or crossed some dollar threshold.
The problem is that none of those segments tell you anything about the person. A C-suite executive and a college student can both be two-time purchasers who spent $80. Your segments treat them the same because your data only knows what they did, not who they are.
So you send them the same email, the same ad, the same offer, and one converts while the other churns.
This week, I want to talk about why segmenting on purchase behavior is a weak foundation for retention and growth, and why the brands pulling ahead are segmenting on identity instead: occupation, lifestyle, life stage, the stuff that actually explains why someone bought in the first place.
Let’s get into it.
This week’s newsie is brought to you by one of the coolest new tools I’ve personally ever seen and one that explicitly solves the problem we’re talking about today. Meet OuterSignal.
OuterSignal helps brands transform basic order data into fully researched, actionable customer profiles.
Right now, all you get is a name, email, and address. Imagine if every order also had age, gender, occupation, social following, property value, and 100+ other signals.
With OuterSignal, you can:
1. Segment your existing customers by data-driven personas – including occupation, life stage, lifestyle, and buying motivations– allowing you to personalize email, SMS, and customer experience to each persona
2. Know why each customer bought, so you can predict who’ll repeat, win back the right churned customers, and stop blasting everyone the same offer.
3. Spot VIPs in your customer base (celebrities, athletes, investors, influencers, buyers), and treat them like it.
Leading brands like Create, Jolie, Magic Mind, and Mizzen+Main trust OuterSignal to power their customer personalization and increase their ROAS by up to 160x.
OuterSignal will research 1,000 customers for free, and is offering friends of Eli 50% off their first two months when they upgrade.
Tell them Eli sent you!
Your segments are SKU buckets wearing a costume
Most brands segment like this: anyone who bought Product A goes in Bucket A, anyone who bought twice goes in the repeat bucket, anyone who spent over $200 goes in the high-value bucket. You might dress it up with a name like “Core Enthusiast” or “Premium Loyalist,” but under the hood it’s still a filter on order count or SKU.
But something most retention folks learn over time is that buying twice doesn’t tell you if someone’s about to buy a third time or ghost you forever. It tells you what someone did after they decided to buy, but it doesn’t tell you why they bought, what problem the product solves in their life, or whether they’ll buy again.
A menswear brand could have 50,000 customers who each bought a dress shirt, and some of them are executives who wear it to board meetings, some are grooms who need it for a wedding, and some are retail workers who need something cheap and wrinkle-free for their shift.
Same SKU, totally different reasons for buying, completely different retention curves. Messaging all of them the same way means you’re treating the person who needs you every week the same as the person who needed you once.
The brands doing this differently segment on who people are
Mizzen+Main sells performance menswear. They could segment on repeat buyers or AOV, and they did for years. But when they enriched over 100,000 historical customer records with actual identity data, they found something better: a massive cluster of C-suite executives who had purchased in the past but weren’t responding to broad campaigns.
They built an audience around that profession, that lifestyle, that identity, and ran it on direct mail over BFCM. It became their highest-performing campaign in company history. Over 20x ROAS, $50 AOV lift above baseline, their number one campaign for the entire peak season. Same product. Same platform. Different understanding of who was buying it.
Gratsi, the DTC Italian wine brand, ran a controlled A/B test on this exact question. They took a single email campaign and split it: half the list got a generic message, the other half got persona-matched creative based on identity data like Active Retiree, Mindful Achiever, Creative Entrepreneur, Health-Conscious Professional. Same subject line, same send time, same list size. Open rates were nearly identical, which means the audiences were consistent.
What changed was what happened after the open. The persona-matched emails drove 47% more revenue, 54% higher click rates, 56% more subscription conversions. The generic email said “It’s Always a Good Time for Gratsi.” The retiree email said “You’ve Earned the Right to Slow Down.” The health-conscious professional got “Zero Sugar. Zero Guilt. Zero Hangovers.” Same wine. Same list. Just talked to them like humans instead of order IDs.
Identity explains behavior, behavior doesn’t explain identity
If you only know someone bought your product twice, you have no idea if they’re about to buy a third time or if they’re done. You’re flying blind. If you know they’re a new parent, a remote worker, a retiree, a competitive athlete, suddenly you have context. You know what problem your product solves in their life, how to talk to them, what else they might need, when to expect them to come back.
A beverage brand might see that someone subscribed and then canceled after two months. If you only have purchase data, that person looks like a churn risk or a lost subscriber. If you know they’re a shift worker who drinks your product before early morning warehouse shifts, you can message them differently than the yoga instructor who drinks it pre-class. One of them churned because the flavor got boring, the other churned because their schedule changed. You can win back one of those people, but you need to know which is which.
You already have the list, you don’t know who’s on it
Magic Mind, the productivity drink brand, installed identity enrichment on their customer base and found out that Kim K had been a customer for nearly two years. No partnership or sponsorship. She had just been buying organically. They sent a care package, she posted it to her Instagram story, hundreds of millions of impressions. The going rate for a sponsored post from her account is in the millions. They got it for the cost of a care package because they knew who was already buying.
Every brand has some version of this. You have customers who are executives, influencers, journalists, investors, people with audiences or purchasing power or strategic value that goes way beyond their LTV. But if you’re only looking at order data, they’re invisible. They look like every other two-time buyer who spent $60, and you’re sending them the same abandoned cart email as everyone else.
Behavior is the what, identity is the why
If you’re only segmenting on the what, you’re optimizing for a pattern you don’t actually understand. You’re building retention programs and acquisition audiences on top of a foundation that can’t tell you why someone bought in the first place or whether they’ll buy again.
The brands winning on this are enriching their customer data with occupation, lifestyle, life stage, and household attributes. They’re segmenting their email lists by persona, not by product. They’re building lookalike audiences based on who their best customers are as people, not what they happened to purchase. And the results aren’t incremental: Gratsi saw 47% more revenue from the same list, Mizzen+Main’s executive audience did 20x ROAS and became their top audience over the biggest shopping weekend of the year, Jolie’s identity-based lookalike drove 412% more sales than the average of everything else they were running.
You can keep segmenting on order count and SKU mix. Most brands are still doing that because it’s easier and the data’s already in Shopify. But easier doesn’t mean better. And the gap between the brands segmenting on purchases and the brands segmenting on people is getting wider every quarter.
Your customers are not their order history. They’re people with jobs, routines, problems your product solves in ways you probably don’t fully understand yet. Segment like it.
That’s it for this week!
Any topics you’d like to see me cover in the future?
Just shoot me a DM or an email!
Cheers,
Eli 💛







What is best way to pull this information out?