Context · Business model

B2C subscription

Recurring revenue moves the question from acquisition to retained value, and makes pull-forward the main way a good result turns out to be nothing.

What changes here

The same four questions, and what each one runs into in this context.

Trial starts are easy to attribute and are not the outcome. Credit assigned at signup says nothing about month three.

Pull-forward looks like lift in a short window and disappears in a long one. The test window has to outlast the natural purchase cycle.

Payback period and contribution margin matter more than CAC. A cheap subscriber who churns at month two is a loss.

Churn models find who leaves, not who can be kept. Uplift is the only honest target for retention spend.

Planned

Triangulating When Your Evidence Disagrees

An experiment, an MMM, and attribution give three different answers. A structured way to reconcile them without averaging away the information.

Planned

Adstock, Saturation, and Honest Response Curves

Lagged effects and diminishing returns are where most mix models smuggle in their conclusions. How to tell a fitted curve from an identified one.
Planned

What a Marketing Mix Model Can and Cannot Do

MMM is a good allocation tool and a weak causal one. What it needs to be trustworthy, and why Bayesian priors do not create identification.

Planned

From Segment to Audience to Treatment Policy

Segments, audiences, scores, assignments, and policies are five different objects. Conflating them is why lifecycle programs become unmeasurable.
Planned

Prediction Is Not Persuasion

A high-propensity customer may convert without you. A high-risk one may be unreachable. The best target is whoever would change their mind.
Planned

The Case for Doing Nothing

Contact fatigue, capacity, and cost mean no-action is a real treatment arm. Why it belongs in every decisioning system by default.
Planned

Uplift Modeling in Practice

Estimating heterogeneous treatment effects when you have a randomized holdout, and how to validate a policy prospectively rather than offline.
Worked example

Broad Nurture, Propensity, or Uplift?

One trial-to-paid nurture sequence, four targeting policies, and a randomized holdout. Compared on incremental conversions rather than model accuracy, which is the only comparison that changes the answer.

System Planned

Customer State Model

One row per person or account carrying consent, lifecycle stage, eligibility flags, and contact history. The foundation every audience, experiment, and policy reads from.
System

Audience Builder and Capacity Planning

Audience definitions as versioned, reusable objects instead of one-off SQL, and the capacity check that decides whether a test was ever going to work.
System Planned

Consent and Suppression

Lawful basis, channel permission, global opt-outs, and complaint handling as one shared layer rather than a filter each campaign reimplements.
System Planned

Automated Measurement

A standing readout that runs eligibility, assignment, delivery, exposure, outcome, and guardrails on a schedule instead of as a bespoke analysis each time.
System Planned

Next-Best-Action Policy

Turning scores into decisions under real constraints: capacity, contact fatigue, margin, cooldowns, and the explicit option of doing nothing.