Many brands aren’t ready for MMM yet, and that’s okay. We routinely turn away companies because they’re not spending enough money, not spending across enough channels, or don’t have the analytics resources in-house to interpret and execute upon a complex model.
If your mix is limited and annual paid media is < $5M, MMM’s organizational lift will probably outweigh the benefit for now. But you still need defensible answers to “did this move the needle?”
For many teams, the first step toward measuring incrementality is running a lift test. We’re firm believers that every single company that is spending anything meaningful on advertising should be running them.
That’s where teams get into the cycle of generating marketing hypotheses, developing testing plans, plugging those into forecasts, adjusting their marketing budgets, hitting (or missing) their goals, resetting expectations, validating their model, and so on… is the basis for eventually moving into more complex measurement methods – and it creates the organizational habits of managing uncertainty and not believing in a single source of truth.
Stop overpaying for lift tests
We’re the biggest supporters of lift tests, but, at the same time, we want to make something clear: brands should not be paying tens of thousands of dollars to run them – not to Recast, and not to anyone else.
The core idea behind lift tests is old and well-understood. You randomly split geos, apply a treatment (spend) to one group, keep another as a control, and compare outcomes. You’re standing on decades of difference-in-differences and synthetic controls – methods that are very well understood and documented across academic literature.
And because the math is standardized and widely open-sourced, there are dozens of open-source packages, and most vendors implement the same core estimators with minor variations.
We believe lift tests should be democratized. That every team – not just the ones with big budgets or dedicated data scientists – should be able to run clean, defensible experiments. And that pricing should reflect the simplicity and maturity of the method, not the markup someone can get away with.
That didn’t sit right with us. We rolled out Recast GeoLift, and we made it free for the first six months and a nominal fee after for a reason – because it’s not that hard to run, and you just shouldn’t be paying more than that.
Our customers have been using them to validate and calibrate their Recast models for the entire time we’ve been around, but, until now, if they wanted to run these kinds of tests, they either needed a dedicated team of data scientists or they had to work with a lift testing vendor.
Recast GeoLift changes that, making lift testing a core part of Recast’s larger incrementality platform.
Again, this is going to be important for many brands, because their first step toward measuring incrementality is running a lift test. Once they’ve built that muscle, they can bring in statistical modeling and start to build a true incrementality system in a single platform.
The thing that makes lift tests so powerful is exactly what makes them accessible: they’re simple. So all we ask is that you question what you’re paying for when working with a measurement, including Recast.
When to graduate to MMM
As a final note, when should you graduate to MMM? A good rule of thumb: your channel mix expands beyond one or two platforms, you’re spending $5M/year on paid media, and you believe that the internal culture is willing to act on insights (reallocate budgets, test new channels, adjust pacing).
That’s when we believe that MMM can do things that lift tests are just not designed for: forward-looking forecasting and the ability to evaluate cross-channel tradeoffs under different spend plans. Lift tests and MMM answer different questions, so the best measurement teams we work with use both together.
TLDR:
- Lift tests are mature and simple, so you shouldn’t pay exorbitant prices for them.
- Across vendors, the math is largely the same (DiD/synthetic control) – always question what you’re paying for here.
- If you’re not ready for MMM (limited channels, <$5M spend, lean analytics), run geo-lift cheaply and regularly to build your experimentation muscle.
- Treat lift tests as the on-ramp to an incrementality system. Then bring in MMM when you’re at the right scale.
