What I've Learned About MMM (Marketing Mix Modeling)

Hi there,

Everyone's obsessed with attribution.

But the best operators I know? They're obsessed with incrementality.

We just had a fascinating lunch-and-learn with the team at Prescient AI, diving deep into Marketing Mix Modeling (MMM). It's something I think a lot of omnichannel brands are still getting wrong -- especially when you're selling across DTC, Amazon, and physical retail.

Here's the uncomfortable truth: relying solely on platform-reported ROAS is basically flying blind.

You get the data… but Meta, Google, and TikTok are all taking credit for the exact same sale. So the question becomes, how do you figure out where your next best marketing dollar should go?

But first, what actually is Marketing Mix Modeling?

If you're not deep in the data weeds, MMM might sound like just another buzzword. But it's actually been around since the 1950s, originally built for CPG brands trying to measure the impact of TV and radio ads before pixels even existed.

At its core, MMM is regression-based modeling. It looks at your historical data (usually the last 2-3 years) to establish a baseline of your sales, factoring in natural peaks and valleys like seasonality or major holidays. Once it understands your baseline, it overlays your marketing spend. If you spent $1M on a campaign and saw a corresponding spike in sales, the model attributes that incremental lift to your marketing.

Unlike attribution, which tries to track individual users (and fails when they jump from TikTok to an in-store purchase), MMM uses aggregate data. It doesn't care about pixels; it cares about what actually moved the needle.

A few thoughts from what we're seeing in the shift toward MMM:

1. The Halo Effect Is Real (and Unmeasured by Attribution)

If you think your Meta ads only drive traffic to your Shopify store, you're missing the bigger picture. In reality, a massive chunk of that impact (often over 50% for omnichannel brands) spills over into Amazon and retail channels like Whole Foods or Target. Attribution models are completely blind to this "halo effect." They assume if someone didn't click and buy on your site immediately, the ad failed. MMM captures this overflow, proving that your digital spend is working much harder than your last-click dashboard suggests.

2. Stop Letting Platforms Grade Their Own Homework

Right now, platforms own the customer journey. If you ask Meta, Google, and TikTok how much credit they drove, they'll all claim 100% of the impact if they had any touchpoint at all. This double-counting leads to wildly inflated ROAS metrics. You need an independent, top-down view to calibrate what's actually driving incremental growth.

3. Seasonality Amplifies Marketing Impact

A dollar spent on Black Friday doesn't yield the same return as a dollar spent in August. Traditional attribution struggles to factor in how seasonal events naturally lift conversion rates. The brands winning right now understand that optimal spend timing requires modeling around these seasonal patterns, turning calendar awareness into concrete revenue growth.

4. The 60/40 Split of Brand vs. Media

When you look at the baseline of your sales, if about 60% is driven by your media efforts, and 40% is your core brand equity, if you turned off all your ads today, your sales would drop by that 60%. Understanding what portion of your business is dependent on marketing driving that business to your brand is critical. This baseline is important for omnichannel brands expanding into retail, where physical presence (aka someone knowing you from walking past your brand in Target) starts building its own kind of brand equity separate from your DTC efforts.

5. The Shift Is Accelerating

Interest in MMM is surging right now. According to recent data from Google Trends, searches for Marketing Mix Modeling spiked by over 200% in mid-2025. Why? Because brands realize that attribution tools alone no longer provide the clarity required for media spend optimization in a complex, multi-channel world.

So why are we spending more time thinking about the MMM space?

Our newest product at Brij has pushed us into the acquisition space. We've realized that Brij directly benefits growth and performance marketers, and our solution is complementary to MMMs.

Specifically, Brij is sending offline deterministic signals to ad platforms. This means when a shopper buys your product at Target and scans a QR code on the packaging, Brij captures that identity signal and feeds it back to your ad platforms, providing the kind of data MMMs have historically had to estimate.

This directly improves the algorithms and adds a new, deterministic dataset to MMMs to better support its predictions.

Curious how others are thinking about incrementality vs. attribution?

Talk to you soon,

Kait

A few extras:

Some of my favorite brands are hiring…

1. Thrive Causemetics - Senior Director, Brand & Consumer Engagement
2. Once Upon a Farm - Senior Director, Innovation
3. Rare Beauty Brands - Senior Brand Manager

References:

[1] Demand for Marketing Mix Modeling (MMM) Accelerating as Marketers Turn Their Attention to 2026