Performance marketers have access to a wealth of detailed conversion data. A purchase can reveal product choices, basket value, discounts, channel, location, and dozens of other attributes.
The limitation is that paid media decisions rarely depend on the transaction alone. A $100 order from a first-time buyer and a $100 order from a customer who purchases every two weeks may look identical in a conversion feed, even though they represent very different customer relationships.
Connect those orders to a persistent profile and they stop matching. One belongs to a customer who buys every two weeks. The other is a first purchase. The transaction was never the difference. The history is.
Start with identity coverage
Connected history starts with recognizing the customer across interactions. How a brand gets there depends on the business. Ecommerce retailers have authenticated accounts. Subscription businesses have identity by design. Physical retail, grocery, and QSR have the harder problem: the transaction happens whether or not anyone identifies themselves. That’s the job loyalty does. It attaches an in-store or drive-through purchase to a known customer.
Enrollment in loyalty programs isn’t the same as identity coverage. What matters is the share of your transactions you can tie to a known customer. Not how many members you have. How many purchases arrive with someone attached.
That coverage supports three kinds of customer data:
- Identity and governance: Identifiers, consent, permitted uses, how profiles relate to each other. This is what makes the rest possible.
- Loyalty program state: Tier, points balance, reward eligibility, redemption history, tenure. This only exists where a program does.
- Derived attributes: Purchase cadence, time between orders, category affinity, time-of-day and location patterns, channel mix, offer response. These are computed from linked events rather than recorded in a source system, and they need identified history rather than a loyalty program specifically.
The third category is where transactions become useful for performance marketing.
Turn history into signals that drive decisions
Derived attributes shouldn’t all be refreshed or activated in the same way. Brands need to calibrate their useful windows against their own customer behavior and the decision each signal is intended to inform.
Consider purchase cadence. Knowing that someone bought coffee this morning is useful. Knowing that they usually purchase every weekday morning creates a different signal. When that customer skips three days, no new event arrives. The purchase that did not happen is the signal. Cadence works only when it is recalculated often enough to register the absence.
Category affinity behaves differently. Because it can be calculated across many purchases, its relevance changes more gradually. One purchase in a new category means little; repeated purchases indicate an emerging preference.
A useful discipline is to work backward from the media decision. If the goal is audience definition, exclusion, or conversion-value optimization, determine which derived attribute improves that decision, how it is calculated, and how current it needs to be.
The two $100 orders diverge here. One fits an established cadence and the other does not. The repeat buyer belongs in a suppression list for acquisition campaigns. The first-time buyer is the one worth paying to acquire.
Make connected history operational
Go back to that drive-through purchase. It only becomes connected history if it resolves to the same profile as the app session and the loyalty account. The identifier was captured at the register. Whether it lands on the right profile is a separate problem, and it is the one that decides whether any of the derived attributes above can be built. That means unified data across web, app, and POS, profiles that reflect recent activity, audience logic that holds wherever the signal is used, and consent that travels with the data rather than sitting in a separate system.
Platforms such as Rokt mParticle provide that connective layer.
Two $100 orders, one of them worth acting on differently.
Written by:
Nick Craig, Head of GTM, Rokt mParticle
The post The power of connected customer history appeared first on MarTech.