What Did That Media Spend Actually Buy?

By Becky Johnson, Advertising Week Writer & Podcast Host

A campaign can deliver the desired result without settling whether every part of the media buy earned its cost. On Modern Marketing and Measurement, Spencer Potts, Co-Founder and CEO of Precise.ai, makes the case for looking beneath campaign totals to examine the buying decisions that produced them.

Attribution helps marketers assign credit, but Spencer wants to ask a further question: What did each audience segment, data source, publisher, or other component contribute relative to its price? If the same outcome could be achieved for less, that is information an agency or brand should be able to use before the next campaign begins.

Taking Apart the Prefab House

Spencer describes Precise as an applied research lab whose first product gives agencies and brands a way to evaluate and adjust media spending while campaigns are running. The work draws on his experience building Madhive with his partner, Adam Helfgott, and an understanding of the decisions taking place inside the programmatic supply chain.

His analogy is a prefab house. A buyer assembles a living room, a dining room, and a kitchen, but each room arrives with decisions already made about its contents. Similarly, a media buyer brings together vendors and services without necessarily seeing every choice made within them.

Spencer wants to examine the parts inside those rooms and determine whether a different combination would work better. In advertising, that means looking more closely at the signals, data, and costs behind the impression purchase, including how those elements work together.

The distinction matters because a campaign result can conceal differences in the value of its components. An audience segment might be part of a successful buy while adding little to the outcome itself. Understanding that contribution could give the buyer a reason to remove the expense or put the money elsewhere.

Useful Before the Budget Is Gone

Spencer calls this approach “decisioning economics,” which he describes as evaluating impact on an outcome relative to cost. Precise is designed to surface that information during a campaign, alongside recommendations about where spending could be more effective.

He offers the example of moving money between channels or changing the mix of suppliers. The recommendation would include a predicted effect and a degree of confidence, allowing the buyer to consider both the proposed action and the expectation behind it.

Those are forecasts, and Spencer makes clear that Precise has to be accountable for what follows. A recommendation is useful only if the system can compare its prediction with the result and learn from the difference.

For a trader, the additional detail could also support a more specific conversation with a vendor. Spencer describes identifying audience segments automatically added to a buy that contribute little to the desired outcome. That would give the trader a basis for requesting that those segments be removed, rather than relying solely on aggregate performance reports.

The System Doing the Checking Needs Checking, Too

As AI agents take on more buying decisions, Spencer sees the same information becoming available to them. An agent working toward a campaign target could consult evidence about which components are contributing value before deciding where to move budget.

That makes the record of the decision important. Spencer says Precise records recommendations and outcomes in a ledger so a company can review what changed, what was expected, and what happened afterward.

He illustrates this with a hypothetical recommendation expected to improve performance by 5% that ultimately delivers 3%. The change may still have helped, but the difference needs to remain visible and inform the system’s next prediction.

“If it’s one black box versus another black box, we’re like, what’s the point?” Spencer says. His argument is that the technology evaluating media decisions should itself be open to examination, with records that can support questions from a client, a procurement team, or a CFO.

Giving Buyers More to Work With

The trader’s job already involves making sense of information from multiple vendors and dashboards. Spencer sees a closer understanding of costs and contributions as a way to give that person more control over the choices being made on the advertiser’s behalf.

Our discussion turns to the role of review within a marketing organization. Bringing in an outside perspective can help identify blind spots, especially when different teams and suppliers each see only part of the activity.

For Spencer, that visibility is central to the purpose of the product. Knowing that a campaign performed is valuable, but being able to explain a spending change gives the buyer something more concrete to bring to the people responsible for the budget.

Go Fish, With Permission

Spencer’s longer-term ambition extends to organizations learning from one another’s data. He describes a permission-based arrangement in which participants retain control over what they share and could be compensated when their information contributes to a monetized outcome.

He compares finding useful information to a game of Go Fish: Does someone else have the piece that would make this combination work better? The idea depends on understanding the contribution of that piece, including the value it creates in combination with information already available.

Spencer presents this as a future direction, estimating that it is probably at least three years away. The immediate focus is on giving agencies and brands more useful evidence about the decisions within their own campaigns.

That brings the discussion back to a practical question about the media buy. If a buyer can see what an expense contributes, act on that information, and review the result, there is a stronger basis for deciding what belongs in the next purchase. A successful campaign becomes something to examine and improve, with a record of how the money was used.