Marketing measurement used to be simple: you bought an ad, someone clicked it, and they bought your product. You tracked the click, claimed the credit, and celebrated.

But the world has changed. Privacy regulations and the end of tracking cookies have made it nearly impossible to follow a single user across the internet. As a result, modern marketers are returning to a powerful statistical method called Marketing Mix Modeling (MMM).

Remember the Marketing Mix Trinity

MMM doesn’t track individuals. Instead, it looks at the “big picture” trends in your data to tell you what’s working. To understand MMM, you don’t need a PhD in economics, but you do need to understand three core concepts that explain how marketing actually works: Incrementality, Adstock, and Saturation. This is the Holy Trinity of Marketing Mix Models.

1. Incrementality: The “True Value” Test

The biggest trap in marketing is confusing “Attribution” with “Incrementality.”

Attribution is like a waiter taking credit for the meal. Just because the waiter brought you the check doesn’t mean they cooked the food. In marketing, attribution often gives 100% of the credit to the last ad a customer clicked (like a “Brand Search” ad), even if that customer was going to buy from you anyway.

Incrementality asks a harder question: “If we hadn’t spent this money, would this sale have happened?”

It measures the lift—the extra sales generated solely because of your marketing.

  • Baseline Sales: These are sales that happen naturally because people already know your brand or need your product.
  • Incremental Sales: These are the sales that only occured because of your ads.

Why it matters

If you only look at attribution, you might think retargeting people who visited your site is your best strategy. But incrementality often reveals that those people were already convinced. Real value often comes from “upper funnel” ads (like video or TV) that introduce new people to your brand, even if they don’t click immediately.

2. Adstock: The “Memory Effect”

Marketing isn’t a light switch. You don’t turn an ad on and get instant sales, then turn it off and see sales drop to zero immediately. People remember things.

This lingering effect is called Adstock. It acknowledges that an ad you saw last Tuesday might still influence your decision to buy something this Saturday.

How We Model Memory

MMM uses math to simulate how consumer memory fades over time.

  1. Geometric Adstock (The Simple Fade): Imagine a radioactive element decaying. Every day, the memory of the ad drops by a fixed percentage (e.g., 50%). This is good for simple, short-term ads.
  2. Weibull Adstock (The Complex Reality): Sometimes, the impact of an ad builds up before it fades. Think of a TV commercial that people talk about a few days after it airs. Weibull Adstock can model this “lagged” effect, where the peak impact only materializes days or weeks after the money was spent.

The Half-Life

To make this concept easy to understand, analysts use a metric called Half-Life. It’s simply the time it takes for the ad’s impact to drop to 50%. A short half-life (a few days) is typical for sales promotions, while a long half-life (weeks) is typical for brand-building campaigns.

3. Saturation: The Law of Diminishing Returns

Imagine you are watering a plant.

  • The first cup of water: The plant drinks it up and grows.
  • The tenth cup of water: The plant is drowning. The water is no longer helping, being wasted pointlessly.

Marketing works the same way. The first $1,000 you spend usually captures the “low-hanging fruit”—the people most interested in your product. As you spend more, you have to work harder to convince less interested people. Eventually, spending more money yields almost zero extra return. This is called Saturation.

The Curves of Growth

MMM plots your spend against your sales on a graph, creating a Response Curve (often using a mathematical formula called the Hill Function). These curves usually take one of two shapes:

Returns start high and drop immediately. This is common for “high intent” channels like Google Search. You capture the demand that exists, but you can’t force more people to search for your product just by spending more.

This looks like an “S” shape.

  • Start: It’s flat. You aren’t spending enough to be noticed yet.
  • Middle: You hit a “threshold,” and efficiency skyrockets.
  • Top: You hit saturation, and efficiency flattens out.
  • Lesson: For S-curve channels (like TV or Social), you need to commit a minimum budget to see results.

The Golden Rule: Optimize for the Next Dollar

So, how do you use this to make money?

Most marketers look at Average ROAS (Return on Ad Spend). They say, “Facebook gave us a 4.0 ROAS, so let’s spend more there!”

But MMM teaches us to look at Marginal ROAS (mROAS).

Marginal ROAS asks: “How much will I get back from the next dollar I spend?”

If your Facebook channel is already fully saturated, the next dollar might only earn you $0.50, even if the average was $4.00. Meanwhile, a smaller channel like YouTube might have a lower average, but a huge potential for growth on the next dollar.

Smart budgeting moves money from saturated channels (low Marginal ROAS) to unsaturated ones (high Marginal ROAS) until the returns are equal everywhere.

Summary: Remember the Holy Trinity

  • Incrementality tells you if the ad actually caused the sale.
  • Adstock measures how long the ad stays in the consumer’s memory.
  • Saturation tells you when you’ve spent too much and are just wasting money.

By balancing these three forces, you move away from guessing and start predicting.