Incrementality & Geo-Experimentation

Find out what your advertising actually caused, with tests designed so the answer can be trusted.

Who it’s for

Brands that want a straight answer to a specific question: does this channel, campaign or change really drive extra sales, or would they have happened anyway?

What you get

  • Test design: the question, the markets or audiences, the duration and the power analysis, agreed before anything starts.
  • Execution support through your existing ad accounts and agencies.
  • Analysis using the method that fits the design: geo experiments, holdouts, difference-in-differences, synthetic control or interrupted time series.
  • A decision memo: what the result means, how sure we are, and what to do next.
  • A testing roadmap, from single tests to an always-on programme that keeps your measurement calibrated.

How it works

  1. Frame the question and the decision it will inform.
  2. Design for power. If a test is too small to detect an effect that matters, I’ll say so before you spend a cent.
  3. Run it cleanly, with guardrails against contamination.
  4. Analyse and report with ranges, not a single verdict.
  5. Feed the result back into your model and your budget.

What I need from you

Access to the ad accounts involved, sales data by geography or audience, and the ability to change spend in the test markets.

Evidence you’ll reach

Level 3 for quasi-experimental designs and Level 4 for controlled experiments.

Evidence levelLevels 3 to 4See the Evidence Ladder
Can small brands run geo tests?

It depends on how many markets you have and how noisy your sales are. The power calculator on this site gives a rough idea; the real design comes first, before any money is spent.

What if a test is inconclusive?

That is a result. It tells you the effect is smaller than the test could detect, which is useful when deciding how much to bet on the channel.