Michael Kaminsky explains media mix modeling, a method for measuring how different marketing channels drive sales. MMM has evolved over the past 10-15 years as brands shifted from direct-to-consumer to omnichannel retail, requiring new ways to track what actually works.
If you were at a dinner party and you had to describe what you do and what media mixed modeling is, like, what is it? How would you describe it? And then also, how has it changed, in your opinion over the last, I don't know, 10, 15 years? And like, how would you describe modern MMM today?
Awesome question. Yeah. So I think at the highest level, right, the idea behind MMM is we want to use statistics and machine learning, or AI, if you're pitching a vc, but to find the relationship between marketing activity on one side and a business, KPI's new customers acquired revenue on the other side. The idea, the reason why this is sort of powerful and, you know, sometimes shocking for a lot of modern performance marketers is that you want to do that without digital tracking, right? Without trying to track people across the Internet. That in one sense, like, makes the game a lot harder, right, because you have a lot less data to work with. But on the positive side, it gives you insight that you can't get just from digital tracking because it allows you to measure channels that are offline, like TV or radio, and it allows you to compare channels on an apples to apples basis in a way that isn't possible when you're just trying to track clicks or impressions. And so the core idea is we're going to look at aggregate first party data and then we're going to use statistics, econometrics or machine learning to find the relationships in the data that then tell us how effective is that marketing activity.
So your journey to mmm, from my understanding, started at Harry's. Can you just describe kind of your evolution understanding? Wow. This is where the market's moving and this is really an important part of our data analytics infrastructure as a growing brand that was once D2C native and then kind of expanded into other channels and things started to become more complex.
Yeah, absolutely. So Harry's went on this like really exciting journey while I was there of starting out dtc, only they were one of the first really big DTC focused brands and then evolved into, you know, omnichannel, CPG brands selling multiple products online on Amazon, in store at Target, in store at Walmart. Now you can buy them everywhere that razors are sold. And that evolution meant sort of going on an evolution of different types of marketing measurement technology. So I ran the data science team, our team built our internal multi touch attribution tool and we worked really close.
