Advertising does not just sell units. It changes what people will pay.
Most media mix models are used to find waste. That is the smaller half of what they can do. The larger half is measuring whether your advertising is making customers less sensitive to price, because that shows up in margin rather than volume and it compounds.
The narrow question and the wide one
The narrow question is the one everyone asks: which channels produced incremental sales, and where should the next dollar go. Useful, and it is what most engagements deliver.
The wide question is different. Advertising that communicates what a product actually does for someone changes how that person evaluates it later. They arrive at the shelf already believing something. When they compare your product to the one beside it, price is no longer the only axis they are comparing on.
That shows up in a demand model as two separate effects, and it is worth being precise about the difference.
- Base sales rise. The volume you sell without promotional support goes up. People buy at full price, unprompted, because they already decided.
- Price elasticity falls. A given percentage price increase costs you less volume than it used to. The demand curve flattens.
The first is the effect people talk about. The second is usually worth more, and almost nobody measures it.
Why the second effect is worth more
Incremental volume is good but it carries cost of goods with it. A price increase you can hold carries almost none. If advertising has made your customers meaningfully less price sensitive, the same percentage taken on price drops to margin nearly intact.
Volume growth is revenue. Reduced elasticity is margin. They are not the same asset and they should not be measured with the same instrument.
There is a second reason it matters more than it looks. Elasticity is where a brand is tested in a downturn. When consumers trade down, the brands that hold are the ones whose customers believe they are getting something specific rather than a commodity with a logo. That belief was built by advertising, months or years earlier, and it is not visible in any last click report.
How this appears in a model
None of this requires exotic methods. It requires asking the model for things most engagements never request.
Track the base, not just the lift. A media mix model decomposes outcomes into a base and an advertising driven increment. Most reporting fixates on the increment because that is what the current quarter bought. The base is the more interesting series. If it is drifting up over years while media weight holds steady, brand equity is accumulating. If it is flat and every quarter depends entirely on the increment, you are renting demand rather than building it.
Let the price coefficient move. This is the part that is usually skipped. Standard practice fits one price elasticity for the whole estimation window, which quietly assumes it never changes. If you believe advertising affects price sensitivity, that assumption defines away the effect you are trying to find. Allow elasticity to vary over time, or across regions with different media pressure, and you can see whether it moves with sustained brand support.
Use the geographic variation you already have. Media weight is rarely uniform across markets. Regions that carried more brand support over a long period, compared against otherwise similar regions that carried less, is the closest thing to a natural experiment most advertisers have sitting in their existing data. A hierarchical model that estimates regional response is already most of the way to this analysis.
Be honest about what you can identify. These effects are slow. They need years of data rather than quarters, and they are easy to confuse with distribution gains, category trends, or a competitor stumbling. Elasticity estimated on eighteen months of data with little price variation is not a finding. If the data cannot support the claim, the useful output is knowing that.
Why this is what brand equity actually means
Brand equity is usually measured with surveys. Awareness, consideration, favorability. Those instruments are fine as far as they go, and they have an obvious weakness: they measure what people say, and nobody in a survey is standing in front of a shelf deciding whether your product is worth eighty cents more than the alternative.
The behavioral definition is more useful and it is measurable. Brand equity is the volume you sell without promotional support, plus the price premium you can hold without losing that volume. Both are parameters in a demand model. Both can be estimated. Both can be tracked over time.
Stated that way, brand advertising stops being the line item that gets cut first when budgets tighten and becomes an investment with a return you can actually put a number on.
What to do with this
If you are running an annual model refresh and reading only the channel ROI table, you are seeing the smaller half of what you paid for. Three things worth asking of your next model.
- What has the base done over the last three years? Not the increment. The base.
- Has price elasticity changed, and does it track media pressure? If your model assumes constant elasticity, ask why.
- Do high support and low support regions differ in price sensitivity? If yes, that is the argument for brand spend, expressed in the language a CFO already uses.
The last one matters most in practice. Finance teams are not hostile to brand advertising. They are hostile to spending that nobody can connect to a financial outcome. Pricing power is a financial outcome. It appears on the income statement. If your measurement can show that advertising built it, the conversation about upper funnel budget stops being a matter of faith.
We build media mix models that run continuously rather than quarterly, for advertisers between five and fifty million in annual media. If that is you, get in touch.