What Is Base Demand?
Base demand is the share of business outcome that an MMM cannot attribute to any specific channel in the model spec. It is the intercept in the regression: the level of revenue or volume that would occur if every modeled channel went to zero. Conventionally it captures brand equity, organic word of mouth, category-level demand, and anything else not explicitly variable in the model.
The catch is that "anything else not explicitly variable" includes channels the model has not been told about. AI search, dark social, podcast, and other dark-funnel sources are all silently absorbed into base demand when the modeler does not provide explicit variables for them.
Why Base Demand Matters
Base demand is the most diagnostically informative variable in any MMM. A base demand that is stable and consistent with brand equity intuition is healthy. A base demand that is rising faster than any explanatory variable in the model is hiding something. In the AI search era, the most common thing it is hiding is AI-driven referrals.
The most common red flag in 2026 MMM reviews is a base demand contribution that has grown 5 to 15 percentage points over the past two years with no corresponding rise in brand equity or category demand. The growth is almost always AI search showing up in the only place the model has for it: the unattributable base.
How Base Demand Works
In a standard MMM, base demand is captured as a combination of the intercept term and the contribution of controls (seasonality, macro indices, brand-equity covariates). The decomposition splits attributable revenue across modeled channels and assigns the unattributable residual to base. The split is mechanical given the model spec.
Changing the model spec changes what counts as base. Adding the AI visibility variable, for example, will reduce base demand contribution because AI-driven revenue moves out of the residual and into the new explicit channel. This is the entire point of adding the variable.
In Practice
Monitor base demand contribution as a trend, not just a level. Rising base over multiple quarters is a diagnostic that the model is missing a channel. Investigate by hypothesizing what the missing channel might be and adding it. AI search is the most common modern culprit; podcast, private-community influence, and dark social are other candidates.
A defensible base demand for a mature brand is typically 30 to 60 percent of total revenue, varying widely by category. A base demand below 20 percent suggests the model is over-attributing to specific channels; a base demand above 70 percent suggests the model is missing channels.
How Presenc AI Helps
Presenc AI is the typical answer when an MMM's base demand is rising mysteriously. The platform provides the AI visibility data that moves the AI portion of dark funnel out of base demand and into a discrete channel, which both reduces base demand contribution to a defensible level and surfaces the AI channel's actual contribution for budget allocation.
