The department has a skill map (the Pyramid) and a machine (the stack). What we haven't shown yet is the brain: the model of the customer that decides who gets which experience. This lesson is that model — and it starts with the framework we deliberately don't use.
We've never been big fans of the customer journey as a framework. Our issue with it is the shape it assumes: a straight line, with the customer sitting at one point on it, and the whole job reduced to nudging them one step further along toward higher LTV. Everything is forward motion on a track.
Reality, from where we sit, is messier than that. A customer buys once, goes quiet for two years, then comes back and buys three times in a month. Someone subscribes to email, unsubscribes, and later resubscribes — to both email and SMS this time. People slide backward, jump ahead, and sit still for long stretches. None of that is a line, and trying to force it onto one tends to hide more than it explains.
So we've come to think about a customer less as a position on a journey and more as a location on a map — a region they occupy right now, and one they can move out of and back into far more freely than a line would ever allow. That map is what we call the Datacop Atlas. Explore it below — the state map and each of the four dimensions beneath it open up — and then read on.
The primary dimension: State
The map is divided into regions, and each profile sits in one of them:
- Anonymous — we only know a cookie. No name, no email, just a device that's shown up.
- Known prospect — reachable (we hold an email or a number), but who has never purchased.
- Active customer — purchased within the last 365 days. This is the capital of the map: the region everything else is oriented around.
- Lapsed customer — purchased before, but silent for 365 days or more.
Two honest caveats before we go further.
First, that 365-day line on Active is a default, not a law. The right window is whichever matches a business's median buying cycle. For a grocery store it might be 14 days — if someone hasn't bought in two weeks, they're almost certainly buying somewhere else. For a considered, infrequent purchase, a year can be too short. A year is a reasonable place to start; the number should move to fit the business.
Second, a profile isn't the same thing as a person — and we say "each profile sits in one region" advisedly, because digital identity is messy. Someone who browses from three different browsers and never leaves an email is three separate anonymous profiles in the CDP, not one person in one region. Identity resolution merges what it can, but until it does, a single human can occupy several regions at once. The map still holds; it just runs on the identities you've actually managed to stitch together, not on some perfect one-to-one record of real people.
The regions on their own aren't the interesting part. The movement between them is. We map every use-case to a specific move from one region to another:
- Identify — Anonymous → Known prospect
- Convert — Known prospect → Active customer
- Deepen — Active → Active (keep an active customer active, and raise frequency and value while they're there)
- Decay — Active → Lapsed
- Win-back — Lapsed → Active
Most of those are moves we're trying to cause. Decay is the exception — it's the one we're trying to slow down. Either way, naming the move is the first decision in any campaign. Before the tone, before the offer, before the channel: which movement across the map is this play actually trying to cause? A welcome flow and a "we miss you" flow aren't two templates — they're two different moves, and once the move is named, most of the other choices start to fall out of it.
The secondary dimensions
State tells us which move a campaign is for. It doesn't tell us how urgently to send it, what to say, what to offer, or whether we're even allowed to reach the person where we'd like to. Four more dimensions handle that. If State is the map, these are the layers we read on top of it to turn "a move" into a specific play for a specific person.
Temperature — urgency. How hot a contact is right now: how recently and intensely they've engaged. This sets the tempo. It runs from Burning (act within 24–48h) through Hot (2–7d), Warm (7–30d), and Cool (30–90d) to Cold (90+d). A burning lead wants a message in hours; a cold one can wait. Temperature is what keeps a hard offer from landing on someone who isn't ready, and keeps a ready buyer from slipping away while a campaign waits its turn in the queue.
Persona — the message. Two customers in the same region, at the same temperature, can still need completely different messaging. A deal hunter and a quality investor might both be active customers, but almost nothing you'd say to one lands with the other. Persona decides the angle, the tone, and the proof points. To make that concrete — Trend seeker, Deal hunter, Quality investor, Occasion shopper, Gifter — though those are just examples; the actual set is something we build per brand. It's the dimension that keeps a message from reading as generic.
Value — the investment. Not every contact justifies the same incentive. Value tier sets how much we're willing to put behind a given relationship — the size of the discount, the richness of the reward, the amount of hands-on attention. As an example, it might run VIP (the most generous treatment), Core, Standard, down to Serial refunder, where generosity is deliberately capped — again, an illustration rather than a fixed ladder. It's how margin ends up spent where it actually pays back rather than sprayed evenly across people who return everything.
Channel availability — accessibility. The last question is the most literal: where is a message even allowed to land? Every contact is reachable through a different mix — email, SMS, push, web, retail — depending on the consent and contact details on file. Some channels are open, some throttled, some closed. Throttled sits in between: email is a good example — the consent is there, but the contact hasn't opened or clicked anything in a while, so leaning on that channel risks the sender reputation rather than helping. A play only gets built on the channels that are actually usable for that person, so the perfect message never gets sent down a road that happens to be blocked.
These four don't replace the move — they refine it. The same Win-back splits into very different plays once you layer them: a Burning, VIP, quality investor with SMS open is not the same campaign as a Cool, Standard, deal hunter who's only reachable by email. Same move on the map; two entirely different plays in practice. That's how a single framework can generate a whole library of use-cases instead of a handful of one-size-fits-all flows — the "at scale" from lesson one, made mechanical.
The Atlas, in one sentence
Put the five together and the division of labour is clean:
State decides whether a campaign exists. Temperature decides how urgently it fires. Persona decides what it says. Value decides what it offers. Channel availability decides where it may land.
That's the Datacop Atlas.
Where this course goes next
One thing hiding in plain sight throughout this lesson: none of the five dimensions is a channel. The Atlas describes the customer, and the channel only enters at the very end, as a constraint on delivery. That ordering — customer first, channel last — has a consequence for how eCommerce teams are organized, and it's the subject of the final lesson: there is one customer experience across multiple channels, and teams can be oriented accordingly.
Datacop is a fractional, AI-enabled technical marketing department for $25M–$2B eCommerce brands. The Academy is where we write down how we work.