Data Rich, Activation Poor

The wheel everyone has seen

Most people in marketing have seen a version of the customer data wheel. A circle, a person in the middle, ten or twelve wedges around the outside labeled demographics, behaviors, attitudes, loyalty, and the rest. It is a fine inventory. It tells you the kinds of things you can know about a customer.

What it does not tell you is the part that decides everything downstream. Which of those things you are allowed to use, and which of them move fast enough to act on. A profile can be complete in every wedge and still never become a message, and the wheel has no place to even ask the question. That is how organizations end up data rich and activation poor.

So hold onto the raw material and redraw the frame. The version I use is the Orbit. Same attributes, organized so the logic is visible instead of buried, with the two missing dimensions pulled to the surface. Read it three ways. The four lenses group each attribute by the decision it informs. The rings sort by speed, stable at the edge and real-time at the center. The dashed gate in the middle marks whether a signal is allowed to fire at all.

The wheel and the map

Figure 1 · The flat wheel

An inventory. Every wedge a peer, every attribute equal. It says what you could know. It says nothing about what you can use or when it matters.

Figure 2 · The Orbit: the same data as a map

The closer to the center, the faster the signal moves. The outer rings say who to talk to, the inner ring says when. Nothing crosses the dashed permission gate without consent and a resolved identity.

• Who they are1 Demographics · 2 Geographics
• What they think & feel3 Interests, attitudes & values · 4 Needs & motivators · 5 Category & competitive affinities · 6 Sentiment
• What they do7 Purchase behaviors · 8 Customer history · 9 Engagement & media habits · 10 Intent
• Worth & where they are11 Loyalty · 12 Customer value · 13 Lifecycle stage

A working diagram. The closer to the center, the faster the signal moves. The outer rings say who to talk to, the inner ring says when. Nothing crosses the permission gate without consent and a resolved identity.

What the data is for

Start with the reorganization. Every category on the map answers one of four questions. Who they are. What they think and feel. What they do. What they are worth to you. Identity, psychographics, behavior, and value, framed as questions because that is how you use them.

The grouping is the first useful thing the map does. On the old wheel, geographics and purchase history and loyalty tier all sat as peers, twelve equal slices of the same pie. They are not peers. One is a fixed fact about a person, one is a record of something they did, one is a judgment the business has made about their value. Sorting them by the question each answers tells you what the data is for, which is the first step toward knowing what to do with it. A leader looking at a profile should be able to point at any attribute and say, in one breath, why it is there and what decision it informs.

Permission is infrastructure

The gate is the change the old wheel never showed, and the one worth the most attention.

Run the cleanest profile you can imagine. Every box filled, every attribute current. You can still reach no one. Between the data and any message sits a gate. Consent state, channel opt-ins, identity resolution, and how often you are allowed to make contact. Some of that is permission in the strict sense, and some of it, like identity resolution, is plumbing. The gate does not care about the difference. Nothing crosses to activation without clearing all of it, so read the gate as the sum of four things: permission, identity, contactability, and policy.

For five years the industry braced for the third-party cookie to die. It did not. Google reversed course in 2024 and kept it in Chrome. But the panic had been aimed at the wrong thing, because what actually changed was consent. Whether a signal can be used now turns on whether the customer agreed to it, not on whether a cookie can carry it. First-party data became the more defensible foundation, and only when consent, identity, and provenance travel with it. The patchwork of state and national privacy laws keeps raising the cost of getting that wrong.

Provenance runs under all of it. The same attribute can be declared (the customer told you), observed (you watched it happen), or inferred (a model guessed). The source sets how far you can trust a signal and what you are allowed to do with it. As more of the profile becomes modeled, knowing which is which stops being academic.

There is a sharper edge to this in 2026. The same first-party data now feeds the models that score propensity, churn, and next best action. The moment you train a model on customer data, the question of whether that data was collected with permission stops being a marketing courtesy and becomes a governance problem. Build a model on data you had no right to use and you have wired a liability into the system. The FTC has already ordered companies to delete models trained on data collected without permission, so read that as precedent rather than theory.

Permission is no longer the footnote on the slide. It is infrastructure.

How fast it moves

The flat wheel treats every attribute as an equal peer. For acting in the moment, they are not.

Some data is stable and rarely changes. Where someone lives, their life stage. Some is slow-moving, shifting over months, like loyalty status and lifecycle position. The behavioral ring moves week to week, purchase patterns and engagement. And the real-time center moves by the hour: a search, a cart, a live shift in sentiment. The slow signals tell you who to talk to. The fast signals tell you when.

One layer sits on top of the map rather than in any lens. Propensity, churn risk, and next best action are computed from everything else here, not collected, and in a working profile they are among the largest parts of what you hold. They also inherit the speed of whatever feeds them. A churn score refreshed monthly is a slow signal wearing a predictive label, whatever the vendor deck says.

This is where most organizations quietly fall short. A profile built only on slow data can personalize, in the sense of putting the right name and the right past purchase in the message. It cannot respond to the moment. If your real-time program is running on demographics and last quarter's orders, it is not real-time anything. The fast end of the map, intent and live sentiment and the still-open cart, is where a message actually fires, and it is usually the most under-built part of the stack.

Picture the retail version. You know a customer's loyalty tier, her region, her predicted value. None of it fires a message. The message fires because she abandoned a cart twenty minutes ago, has SMS permission on file, and has not hit her contact cap for the week. The slow data decided she was worth talking to. The fast data, cleared through the gate, decided the message goes now.

Three places to look

Put your own organization's data on this map and look at three things.

The old wheel answered the question of what we could know about a customer. That was the right question for a different decade, when collecting more was the whole game. The questions now are narrower and harder. What are we allowed to use. How fast does it move. What converts to a message a real person receives.

Read the map that way and it stops being an inventory of everything you have collected. It becomes a working diagram of where your next message actually comes from.