Common misconceptions about MMMs

Everyone here at Prescient AI is a constant student. Yes, our team built our MMM from scratch and we’re all deeply familiar with its nuances and capabilities (and we’re also thrilled to watch our clients score some serious wins with it). But I’m learning about the wider understanding of MMMs every time I talk to a potential new client and hear what they’ve picked up about this technology before finding us—and a lot of it isn’t true. These are the major MMM misconceptions that keep popping up in conversations and why they need to be put to bed for good.

MMM is a “new” technology

Definitely not, but it’s easy to make this mistake because the computing power behind MMMs was limited until relatively recently. We’ve had MMMs since the 1960s, but they were based on paid media of the time—and we all know marketing and the toolkit marketers use has changed dramatically since then.

Ecommerce propelled MMMs forward because we now had some measurement of digital marketing results. To this day, we still don’t have a concrete source of truth for measuring the impact of some of our oldest advertising methods, like billboards and radio ads. But ecomm is different; the customer journey is largely documented (at least until user data restrictions increase) so we can understand—and verify we understand through an MMM like Prescient’s—how a Facebook or Google ad affects sales online. MMMs seem new because of marketing’s evolution, but they’re an old idea.

All MMMs are the same

When you’re vetting multiple MMM providers, you’re not just checking in on the bells and whistles. The small details of a platform matter more than you think; things like whether you need to pay for each seat and the intuition of the dashboard navigation affect your daily experience of the platform, so they should absolutely be considered. But the biggest question to tackle right away is whether the provider’s MMM is the right fit for your brand.

MMMs can be very different from one another. Some are more flexible, some are less flexible. That changes how calculations are made when they report on attribution. This difference in how they report on what likely happened from your paid media spend can then impact what you decide to do next with your marketing budget.

Prescient, for example, differs from other MMMs in four main ways. We go down to the campaign level, we update daily, we show the impact not just on your online store but also Amazon (and soon retail), and our time to value is the quickest in the industry (~36 hours).

All MMMs are different

I know, I know, annoying points to put back to back, but they’re both important to understand. MMMs do differ, but it’s critical to understand that they’re not that different generally speaking. Most MMM providers are using an open source MMM from one of the big tech firms that they’ve altered slightly for their own company. That makes these MMMs sort of like people: we all look a little different, but we share 99% of the same DNA.

That’s not necessarily a bad thing and it’s not something to think of as lazy. Think about trying to design a new mode of transportation in a world that already has planes, trains, and automobiles. It’s quite challenging, and you probably end up with variations on these existing types because they’re proven to do the job even if you think there might be a better way to do it.

In this world of planes, trains, and automobiles, most MMM providers built a new model of a car. Prescient built a submarine—an entirely different type of transportation, able to traverse previously unknown underwater terrain when we had only ever had the ability to travel by land or air. It’s not a perfect analogy, but it does help people understand building an MMM from scratch compared to tailoring an open source MMM.

MMMs only provide high-level insights

This isn’t the case, though there are different depths of insights offered by different MMMs. Many MMMs on the market show channel- or tactic-level insights. Prescient’s MMM offers campaign-level insights. Ours also shows you the impact of your campaigns on other channels—like organic traffic and branded search—and your other commerce channels, like Amazon. That information helps marketers answer some very specific questions about their marketing efforts, like:

• How does this specific Facebook campaign drive Amazon sales?
• How much revenue does this non-branded Google prospecting campaign drive?
• Do people who see this Instagram ad interact with my brand in other ways?
• Do my Amazon ads or Google ads drive more sales on Amazon?
• Am I spending too much on any of my campaigns?
• How much more revenue could I drive by increasing the budget on this campaign?

My MMM is my sole source of Truth 

Although we pride ourselves on our commitment to delivering very high accuracy scores with our models, we just wouldn’t be telling the truth if we said Prescient should be your sole source of Truth. No company claiming their MMM offers the Truth is telling the truth. This industry is about context and data, art and science—and it’s directional, getting you as close to what actually happened as possible.

MMMs won’t be accurate for brands in hyper growth

We hear this one a lot, and it’s understandable. You may be experiencing 30% month over month growth and that brand acceleration is often reflected in a rapidly-evolving marketing mix. Just because your numbers are changing in magnitude more than other brands doesn’t mean MMM can’t understand what’s happening. Some will be better at this than others, though.

Here at Prescient we use “flexible” models. That isn’t an official machine learning term, but it’s a convenient way to describe them. It means our models are more flexible and can capture a wider range of relationships than “non-flexible” models used by open-sourced MMMs. So it doesn’t matter for the model’s purposes if you’re experiencing hyperbolic or slow and steady growth.

Wrapping it up…

There are always misunderstandings and misconceptions about “new” technologies. Since MMMs feel new as commerce brands more widely adopt them, they’re no exception. We consider this a living document, meaning we’ll keep adding to it when we hear more points that need to be better understood about MMMs. Want clarity around something we didn’t address here? Let me know! I’m always thrilled to share more about our use cases and roadmap.