By Amara Osei

A media channel showing exceptional ROI may simply be capturing demand that already existed, according to a new AdExchanger column on the limits of marketing mix modeling (MMM), the statistical method that measures how sales respond to spending across advertising channels. The sharper question — the one every dollar must eventually face — is whether it created new revenue or took credit for revenue that was coming anyway.

That distinction sits at the center of the piece, titled "MMM Is Getting Smarter. It Still Can't Do The Whole Job Alone," published on AdExchanger's Data-Driven Thinking section. Its core claim is blunt: reported efficiency and actual demand creation are not the same thing, and the tools marketers use to measure one can mislead them about the other.

What does the ROI number actually tell you?

The column lays out a two-sided problem. A channel with exceptional reported ROI may simply be intercepting customers who were already en route to purchase. Another channel may look less efficient on paper while doing the harder job — creating customers who otherwise would never have entered the market.

In other words, the model rewards the closer and undervalues the opener. For budget owners, that inverts the usual logic of optimization: shifting spend toward the highest-measured ROI channels can concentrate money on demand capture while starving the activity that grows the category.

The author's framing, stated as the question every media dollar has to answer: "Did it create new revenue, or simply take credit for revenue that was coming anyway?"

Why smarter models don't close the gap

The headline itself concedes the constraint. MMM is improving, the column's title acknowledges — yet it "still can't do the whole job alone." Better mathematics sharpens the estimate of what happened; it does not, on its own, distinguish revenue that advertising caused from revenue advertising merely intercepted. The method needs corroboration, which is the operational takeaway hiding in the word "alone."

Notably, the excerpt does not specify which complementary methods marketers should pair with MMM — incrementality testing, holdout experiments or geo-level trials go unmentioned in the visible text. Readers wanting a prescribed toolkit will have to consult the full column.

The takeaway for budget owners

Trust a strong ROI figure less when demand already exists and the channel sits late in the purchase path. The measurement task is separating creation from capture — and no single model, however smart it gets, settles that alone.