Get the MMM-Lite Starter Kit

Free kit

Privacy-first measurement on a budget. A ready-to-run media-mix model (Python), the exact weekly data schema, an incrementality-test pre-registration template, and the monthly decision rules — the working parts of the system AAJ builds for Seed–Series B clients.

Free. No spam — just the kit (and the occasional growth playbook). Unsubscribe anytime.

The MMM-Lite Starter Kit — Python model, data schema, incrementality test, decision rules

What's inside

Four working parts, in one folder

Everything you need to answer one question every month: where does the next $1,000 work hardest?

The model
A ready-to-run media-mix model in Python — adstock, diminishing returns, honest out-of-sample fit checks, mROAS per channel, and a reallocation suggestion for the next month.
The data schema
The exact weekly data schema the model expects. Fill it, run it. No guesswork about columns, units, or aggregation.
The test pre-reg template
An incrementality-test pre-registration template you lock before launch — hypothesis, geo split, primary metric, decision rule. No peeking, no goal-moving.
The decision rules
The monthly playbook: 10–25% reallocations toward the highest marginal ROAS, and the rule that keeps you honest — a live test outranks the model whenever they disagree.

Is it for you?

Built for teams with real spend and honest questions

You'll need 26–52 weeks of weekly data, spend that varied, and someone comfortable running a Python script. No data science team required — the health checks tell you when to trust the outputs and when not to.

FAQs

Frequently Asked Questions

Common questions about the MMM-Lite Starter Kit.

What is MMM-Lite?

MMM-Lite is a lean, privacy-first marketing mix model that reads aggregate weekly spend and outcomes to estimate each channel's contribution and marginal ROAS. No user-level tracking, no cookies — just spend, revenue, and a few controls. It's the same approach AAJ uses inside the six-week sprint, packaged so a small team can run it themselves.

How much data do I need?

You'll need 26–52 weeks of weekly data, with spend that has actually varied across channels (constant spend teaches the model nothing). Weekly granularity is the sweet spot for most Seed–Series B teams — daily is noisier, monthly loses signal.

Is the code really free?

Yes — MIT-style, use it, adapt it, keep it. The paid version is the six-week MMM-Lite Engagement where we build it on your data with a live incrementality test running alongside. See /services/mmm-lite-sprint.

Measure without tracking anyone.

One email. The full kit in your inbox.

Want it built for you? See the MMM-Lite Engagement

What this resource helps you do

A ready-to-run media-mix model, the data schema, an incrementality-test template, and monthly decision rules — where does the next $1,000 work hardest?

Get the kit from AAJ's free marketing templates library.

Get the MMM-Lite Starter Kit

Four working parts, in one folder

The model

The data schema

The test pre-reg template

The decision rules