Dani Nadel, President and COO, Feedvisor
A consumer watches a creator demonstrate a skin-care product on TikTok. She doesn’t click. Two days later, she searches for the product on Amazon, sees a sponsored ad, and visits its page. Days later, after seeing the same brand on streaming television, she searches for it by name and purchases. A promotion is running, though she would have paid full price.
Which channel drove the sale?
Marketers have spent years building attribution systems to answer that question, and retailer first-party data, authenticated audiences, clean rooms and commerce media networks are finally making more of the journey measurable. But attribution is improving just as the problem shifts from seeing what happened to deciding what to do about it.
By the time she searched the brand by name, the business outcome depended on much more than the channel that received credit.
Media creates demand. Commerce determines what that demand is worth.
Where Demand Starts And Where It Lands
Demand has become portable. It can be created in one environment, strengthened in another, and captured somewhere else. Evaluate each channel independently and the economics distort: Lower-funnel media can take disproportionate credit for demand created across multiple interactions, while earlier investments look unproductive. This imbalance has been debated for decades, usually without the data to settle it.
That movement is now measurable. Skincare brand Topicals reported a 3.7x uplift in Amazon revenue tied to its TikTok activity, invisible to TikTok Shop’s native measurement. Coffee brand Fellow reported a 500% lift in branded searches after adding streaming TV. Unilever’s Emerge launch with Target showed 80% of media-influenced sales happened in physical stores, with more than half of store orders driven by Pinterest. Different journeys, same pattern. Demand often produces commercial value beyond the channel where it originated.
Yet, channel economics still favor the interaction closest to the transaction. Brands can therefore overinvest in capturing demand that already exists while underfunding media that replenishes it.
The money is moving that way at scale. WARC Media forecasts retail media reaching $200.4 billion globally this year, and expects it to represent 55% of alcoholic-drinks media investment globally and 54.9% of food-category spending by 2027.
WARC’s Alex Brownsell notes that retail media excels at converting demand but can be less effective at long-term brand building, while WARC’s testing found memory encoding 47% lower for ads in simulated Walmart and Amazon shopping environments than off-site.
The strategic tension is that conversion-weighted metrics reward brands for harvesting today’s demand, even when long-term growth depends on creating tomorrow’s.
This is where better attribution reaches its limit. It reveals where demand came from and where it went. It cannot determine what the business should do when that demand arrives. The value of a signal comes from the decision it changes, and the highest-value decisions increasingly extend beyond media.
Five Signals, One Decision
The scenario is familiar. Social and streaming investment increased demand for a product. Search volume and conversion are strengthening, and the advertising system recommends additional spend.
Now add the business context:
• Inventory has fallen to three weeks of coverage.
• A major competitor just stocked out.
• Organic rank has significantly improved.
• A promotion is scheduled for next week.
• Contribution margin has tightened.
The value of another advertising dollar just changed in five directions at once. The competitor stockout argues for capturing share. Limited inventory argues for restraint. Better organic rank may reduce the need for paid traffic, while stronger demand may make the promotion unnecessary. Tighter margins reduce the room for error.
Incremental demand does not automatically become incremental profit. A brand reading only the media signal may scale spend into a stockout and fund a promotion against demand that was already converting.
When those signals and decisions stay disconnected, brands pay what I’ve called a disconnection tax. It compounds through individually rational decisions made with partial visibility, so the business grows less profitably.
The Gap Between Signal And Decision
Until recently, the commercial context needed for those broader decisions lived in separate systems. The architecture is changing.
This year, Amazon Ads opened its MCP Server in open beta, allowing AI agents to connect directly to advertising data and controls. Microsoft’s Dynamics 365 Commerce MCP Server entered public preview, exposing product availability, pricing, promotions and order workflows through the same protocol.
Commercial context once fragmented across advertising, commerce and operational systems can increasingly be made machine-readable and reach the decision in time to matter. Both remain early, and neither evaluates profitability nor grants agents authority to act, but they point to a new decision model: Media can be evaluated against the state of demand and the business, expanding the range of possible responses.
Sometimes the answer is more advertising. Sometimes it is pricing, promotion or inventory. Sometimes, the highest-value decision is to let existing demand convert without buying another interaction, a decision an algorithm optimizing media alone won’t recommend.
None of this manufactures certainty. No system knows what a consumer would have done without an exposure, but perfect attribution is not the prerequisite for better decisions.
The Next Shopper Journey
So, which channel drove the sale?
That question tells us only part of what matters. To understand what changed as demand moved, it’s important to look at search behavior, organic position, inventory pressure, competitive opportunity, price sensitivity and the economics of the next interaction.
For years, commerce media has focused on winning the next conversion. The new shopper journey requires a harder discipline: knowing when not to pay for one.
If a creator, streaming or social campaign has already increased branded search, strengthened organic rank and accelerated conversion, the most profitable next action may not be another bid. It may be protecting inventory, avoiding an unnecessary discount or letting demand already created convert on its own.
Where demand converts tells us where the transaction happened. Understanding where it came from tells us what helped create it. But neither tells us what that demand is worth. That depends on what the business does next.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
