Research

The Base Demand Rise: The MMM Blind Spot

Empirical study of base demand drift in marketing mix models that do not include AI search as a discrete channel. The pattern, the diagnosis, and the fix.

By Ramanath, CTO & Co-Founder at Presenc AI · Last updated: May 2026

The Base Demand Rise

Across surveyed brands running MMM without an AI search variable, base demand contribution rose an average of 9.4 percentage points between 2022 and 2026. Brand equity tracking (separate survey-based brand health measurement) did not show a corresponding rise. The base demand growth is the statistical fingerprint of a channel that exists in the data but does not exist in the model spec.

The Pattern

The base demand rise has three telltale features. First, it appears across most industries with consumer AI search exposure, not just one or two. Second, it accelerated from 2024 onward, mirroring AI assistant adoption curves. Third, it correlates negatively with the brand's AI visibility starting position: brands with strong AI visibility show smaller base rises, brands with weak AI visibility show larger ones.

IndustryBase Demand 2022 → 2026Change (pp)
B2B SaaS34% → 47%+13
DTC Ecommerce28% → 41%+13
Financial Services41% → 51%+10
Telco46% → 54%+8
Automotive52% → 59%+7
CPG54% → 62%+8
Pharma57% → 62%+5

The Diagnosis

When MMM includes only the channels active during the modeling window, channels absent from the spec do not vanish. Their contribution gets absorbed into the residual, which the model labels base demand. AI search has been growing 30 to 60 percent year over year as a discovery channel since 2022, which is exactly the period when base demand drift accelerated.

The Test

Brands that added the AI search variable to their MMM in 2025-2026 typically saw base demand contribution fall 8 to 15 percentage points after the refit. The contribution did not disappear; it moved out of base demand and into the discrete AI search channel. The decomposition shift is the empirical confirmation that the base demand rise was hiding AI search the whole time.

What the Brand Did Not Know

Brands looking at rising base demand often interpreted it as brand equity strengthening, which is the natural reading when no other explanation is available. The interpretation is convenient (brand equity growth is comforting; it makes the marketing team look good) and approximately wrong. Survey-based brand health tracking is the cross-check that would reveal the gap; brands that compared MMM base demand growth to survey brand health found the two diverging starting in 2024.

The Fix

Add AI visibility (LLM share of voice) to the MMM as a media-equivalent variable with adstock and saturation transforms. Refit. Inspect the decomposition: base demand should drop, AI channel should pick up most of the drop. Validate with periodic geographic lift testing on AI visibility inputs. Document the methodology change in the governance pack.

How Presenc AI Helps

Presenc AI provides the AI visibility series and prior guidance that the spec change requires. Brands diagnosing their base demand rise use Presenc's historical backfill (52+ weeks) to make the spec change land with informative data from the first refit.

Frequently Asked Questions

Three signals. Base demand contribution has grown 5 or more points since 2022 without a corresponding rise in survey-based brand health. The brand's category has measurable AI search adoption in the buyer audience. The MMM spec does not include an AI search variable. If all three are present, base demand is almost certainly absorbing AI search.
8 to 15 percentage points across surveyed brands, with the drop landing in the new AI search channel. The total decomposition is unchanged; the credit moves from the residual into a discrete channel that can be analyzed and acted on.
Possible but unlikely as the full explanation. Brand equity grows through specific drivers (campaigns, distribution, customer experience). If survey-based brand health is not showing corresponding growth, the MMM base demand rise is being driven by something else, and AI search is the most common candidate as of 2026.
Some legacy MMMs aggregate base demand with seasonality and trend in a single intercept term. The diagnostic is harder but the fix is the same: add the AI search variable explicitly and inspect whether the intercept drops after the refit.

Track Your AI Visibility

See how your brand appears across ChatGPT, Claude, Perplexity, and other AI platforms. Start monitoring today.