Pillar · Marketing · MarTech

Data-driven Marketing — The 5-Phase Process

Data-driven marketing is a way of working in five phases — Collect, Understand, Decide, Automate, Execute — not a campaign style. Jonas Rashedi (Chief Digital Officer, author of "Datengetriebenes Marketing", Springer Gabler 2020, host of MY DATA IS BETTER THAN YOURS / MDIBTY) walks through the full cycle, where most teams fail (hint: attribution), and why tools are never the actual problem — distilled from over 317 podcast episodes with data leaders.

A way of working, not a campaign style. The hardest phase is attribution — and that's a political problem, not a technical one.

The five phases — short version

  1. Collect. First-party data from all touchpoints into one structured layer. Consent-managed.
  2. Understand. Segments, behavioral patterns, customer lifetime value, churn signals. From numbers to insight.
  3. Decide. Decision logic — when to send what to whom, with what content. The connecting tissue between insights and activation.
  4. Automate. Trigger systems that execute decisions in real time. Marketing automation, personalization engines, push services.
  5. Execute. Channels, creative, measurement, feedback loop. Closes the cycle back into Collect.

Why attribution is the hardest phase

Attribution is technically solvable. Multi-touch, last-click, data-driven attribution — the algorithms exist. The problem is political: every team wants to see its share of the conversion, and no model is objective. Sales claims their email triggered the conversion; performance marketing claims their paid ad did; product claims the personalization engine did. All three are partly right, and any model will favor one of them.

The only way out is transparency in the model and continuous calibration. Tell every team which model is being used and why; recalibrate quarterly with hold-out tests; document the assumptions. Hide the model and you are hiding the political decision behind it.

Why tools are not the actual problem

Most data-driven marketing initiatives fail not because the tool is wrong but because the operating model is wrong. Marketing buys an automation platform without IT, data team, and legal at the table. Six months later, the platform has 30 segments that nobody validated, nobody can audit, and nobody trusts. Tool was correct, organization was wrong.

Correct sequence: define the operating model first (who owns segments, who validates triggers, who measures outcomes), then choose the tool to support it. Reverse this and you are buying expensive shelfware.

How long does it really take?

For a mid-market B2C player with three channels and a decent data foundation: 18–24 months to a maturity that deserves the name. First quick wins after 90 days, phases 1–3 of the 5-phase process within a year. If someone promises faster, they are selling tool configuration instead of a way of working.

Do you need a CDP?

From three active channels, and once you want to personalize across identity, yes. Below three channels, a well-configured CRM plus marketing automation tool is enough. More on the CDP pillar.

Where to go from here

The German-language pillar at /wissen/datengetriebenes-marketing/ contains the full depth (eight cluster articles on each phase, attribution, omnichannel). For consulting on a marketing audit: consulting page.