TL;DR

Your Marketing Mix Modeling (MMM) results break revenue into channel contribution plus baseline, score each channel's efficiency on a deduplicated and incremental basis, report how well the model fits your actual revenue, and show where each channel sits on its diminishing-returns curve. Check model fit first to know how far to trust the outputs, then use decomposition, efficiency, and saturation curves to guide budget decisions in the Scenario Planner.

Overview

Marketing Mix Modeling (MMM) estimates how much each channel, and demand you would have earned anyway, contributed to your revenue over time. Your results screen presents this in four views: revenue decomposition, channel contribution and efficiency, model fit, and saturation curves. This article explains how to read each one and what a healthy result looks like, so you can act with the right level of confidence. To turn these results into budget moves, see Using the Scenario Planner.

Key terms

Baseline revenue: Revenue that would occur even with no advertising, driven by organic search, direct traffic, email, and word-of-mouth. A large baseline is a sign of a healthy brand.
Revenue decomposition: The breakdown of total revenue into each advertising channel's contribution plus baseline.
Efficiency ratio: A channel's revenue per dollar spent, measured on a deduplicated and incremental basis.
Model fit: The degree to which the model's predictions match your actual historical revenue.
Diminishing returns: The pattern where each additional dollar spent on a channel produces progressively less incremental revenue.
Saturation curve: A visualization of the diminishing-returns relationship between spend and incremental revenue for a channel; the slope at any point is the marginal return on additional spend.

Revenue decomposition

The revenue decomposition view breaks your total revenue into components: how much came from each advertising channel, and how much was baseline, the revenue you would have earned even with no advertising.

Baseline revenue typically includes organic demand, branded direct traffic, and word-of-mouth effects. For most direct-to-consumer brands, baseline is a substantial portion of total revenue. If your model shows baseline at 40 to 60 percent of total revenue, that is normal, not a sign the model is undercounting advertising.

Channel contribution and efficiency

Each channel's contribution is expressed both in absolute revenue terms and as an efficiency ratio (revenue per dollar spent). Efficiency ratios from MMM are often lower than what your ad platforms report, for two reasons:

MMM deduplicates credit across channels. If the same customer saw both a Meta ad and a Google ad, MMM does not give both full credit.
MMM nets out purchases that would have happened anyway. Someone who would have bought regardless of seeing an ad is excluded from the incremental count.

A lower efficiency number from MMM versus your platform dashboard is expected. It is a more honest read, not a worse one.

What good model fit looks like

Model fit is the degree to which the model's predictions match your actual historical revenue. Triple Whale reports several fit metrics for each model run, including MAPE (Mean Absolute Percentage Error), CRPS (Continuous Ranked Probability Score), Pearson Correlation, and R squared. These are displayed together in the model fit panel with color-coded labels indicating whether each metric falls in a good, acceptable, or low range. As a general guide:

Fit label	What it means	What to do
Green (good)	All four metrics fall within their target ranges.	Channel-level outputs are reliable and can be acted on with confidence.
Amber (acceptable)	One or more metrics are borderline.	Outputs are directionally useful, but interpret channel-level decomposition with some caution.
Red (low)	One or more metrics are significantly outside target range.	Discuss with your Triple Whale team before acting. The underlying cause is usually identifiable and often fixable.

Training and testing periods. Each metric is reported for two windows: the training period, where the model learns the relationship between spend and revenue, and a test period, where the model forecasts the KPI from spend alone so its predictions can be checked against what actually happened. Strong performance on the held-out test period is the best sign that the model will generalize, not just describe the past.

Confidence score. The fit metrics are also combined into a single confidence score for the training and test periods, giving you a quick read on overall model quality before you dig into individual metrics.

Historical fitting plot. The fit panel plots predicted KPI against actual KPI over time, with a shaded band showing the confidence interval around the prediction. A close overlay between the two lines is a visual confirmation of good fit.

Low fit is usually caused by insufficient data history, very consistent spending patterns (no variation for the model to learn from), or major unaccounted-for events during the historical period. It is not always a sign of a modeling problem.

Saturation curves

The saturation curve for each channel shows the relationship between spend level and expected return. The slope of the curve at your current spend level tells you your marginal return on additional investment, how much incremental revenue you would expect from the next dollar spent in that channel.

A steep, upward-sloping curve at your current spend level: you have room to scale, the channel is not yet saturated.
A flattening curve: you are approaching diminishing returns, so additional spend produces proportionally less revenue.
A nearly flat curve: the channel is saturated at your current spend level, so additional spend is unlikely to be efficient.

From results to budget decisions

Once you trust the model and understand where each channel sits, the next step is acting on it. MMM turns these saturation curves and efficiency estimates into concrete budget scenarios, including Expected versus Optimized spend and recommended reallocations across channels. For how to model those scenarios and read the recommendations, see Using the Scenario Planner.

When to use it, and what to keep in mind

Use your MMM results for cross-channel budget planning, for understanding how much of your revenue comes from brand strength versus paid media, and for comparing channels on a like-for-like incremental basis. MMM works at the channel and budget level, so it complements rather than replaces campaign-level and creative-level analysis.

Always check model fit before acting. A green or amber label gives you enough confidence to use the channel-level detail, while a red label is a signal to review the model with your Triple Whale team rather than act on it.

Reading Your MMM Results