A First Look at Causal MMM

I’d like to share something special: Haus is developing an MMM grounded in experiments. We’re calling it Causal MMM (not casual MMM — there’s nothing casual about it) and it’s the MMM we’ve forever wished we had: an MMM where incrementality experiments drive and tune the model. 

Nearly four years ago, I founded Haus with a clear mission: to make privacy-durable causal inference — aka, incrementality — accessible to businesses of all sizes. We set out to create a platform that made rigorous holdout experimentation simple and approachable, and I’m proud to say we’ve exceeded even our most ambitious expectations.

At Haus, we firmly believe that incrementality experiments are the most reliable method for determining causation — and they should be the cornerstone of any modern measurement stack.

But we also understand that marketers often face questions that experiments alone can’t always answer. For example, how can you measure the efficiency of your media spend across all channels, including those you can’t geo-test? Or how do you evaluate hypothetical scenarios to better plan for the future?

For many, the answer might seem like a traditional MMM (marketing mix model). But we’ve lived through the challenges of traditional MMMs — tools that are often too expensive, overwhelming, underwhelming, unactionable, too slow, too opaque, and based on correlational signals that are tough to bet on (or, frankly, make sense of). In our industry survey of leading DTC and e-commerce brands, we heard that brands that already use a traditional MMM consider it one of the least-trusted tools in their measurement stack. The price-to-value ratio is out of whack, and it’s about time something changed.

Haus is that change. Nearly a year ago, we set out to reimagine the MMM experience. We wanted to build an MMM founded on experiment data – an MMM that brands could trust and reliably use in decision-making. A causal MMM automatically and clearly informed by experiments. 

An MMM we would use, trust, and bet on.

With Causal MMM, every single experiment a customer runs feeds the model and continuously improves it. Rather than building a black box MMM, we visibly show brands the tests calibrating the model. This clarity helps build trust in the data and recommendations — it’s adjusted to truth. Causal MMM will even suggest the next best experiments to consider.

Thanks to our early design partners putting Causal MMM through its paces, we’re seeing Causal MMM outperform traditional models on precision, speed, and user understanding. This is the no-nonsense solution we strive for – the experience marketers demand and deserve.

We’re excited to make Causal MMM widely available later this year and hope you’ll embrace this next frontier.

Causal MMM — coming 2025.