I spent this week at Possible in Miami listening to operators across the buy side, sell side, retail media, and the AI infrastructure layer talk about where advertising is actually going. A few patterns stood out, and most of them have direct implications for measurement, currency governance, and the structural position of independent infrastructure providers like us.

Below is what I took away, written for the people in our industry who care about where the dollars, the standards, and the audit rights end up in five years.

1. Measurement is the moment

The framing that hit hardest came from Eutec and Hershey: the work of marketing leadership is moving from "cost per deliverable" to "cost per decision." Hershey is now running an agentic MMM that ingests three years of cleansed data and answers optimization questions in two minutes. Questions that previously took the agency four to six weeks.

That speed changes the political economy of measurement. When the model can answer in real time, the conversation moves from "who reported what" to "what should we do next." The measurement layer becomes the operating system of marketing rather than a quarterly post-mortem.

The implication for our industry: measurement bodies that ship slowly, in arrears, with opaque methodology will not survive this transition. The bar is now real-time, granular, and queryable.

2. Outcomes are eating impressions

Terry Kawaja's keynote put a structure on something everyone already feels. The conversion data hierarchy runs from impression to viewable impression to attention to context to audience to intent to outcome. CPMs sit at the bottom of that stack. LLMs and retail media networks are pulling spend up the stack toward intent and outcome signals.

For OOH this is both a threat and an opening. The threat: if we keep selling against impressions while every other channel sells against intent and outcome, we lose budget share. The opening: OOH's mobility, presence, and pre-purchase context signals are intent signals. We just have to measure and price them as such.

3. The open internet thesis is now a structural argument, not a nostalgia argument

Jeff Green's framing was sharper than I have heard him before. The walled gardens have a tight supply chain, near-zero cost of goods sold, and cannot be audited because they sit on both sides of the transaction. The open internet has premium inventory, journalism, sports, music, and entertainment, but it carries the cost of complexity.

His argument: if the walled gardens win on efficiency alone, the entire premium media ecosystem hollows out. The counter is to align interests with the buyer, eliminate supply chain rents, and make the open internet thrive on outcomes.

OOH sits structurally inside the open internet category. Our case for budget share is the same case Jeff is making for the open web at large.

4. Mobility data is a distinct signal class, not a substitute for location

Arity's session was useful because it drew a clean line between location data and mobility data. Location is a scatter plot of pings. Mobility is a journey: origin, route, intent, dwell, destination, and what comes after. The two are not interchangeable, and brands that treat them as such are leaving incremental lift on the table.

For Motionworks this matters because journey data is the substrate underneath impression measurement in OOH. The audiences we measure are defined by motion, not by static presence. The industry vocabulary needs to catch up.

5. Retail media is becoming an infrastructure layer, not a channel

CVS Media Exchange laid out a position I think every measurement provider should pay attention to. Their argument: retail media is not a place to buy ads. It is the signal layer that informs how brands plan, optimize, and measure across every other channel, including CTV, social, and OOH.

The strategic move is to own the closed-loop signal and license it outward. The CFO conversation shifts from "what was my CPM" to "what was my incrementality, and which touch points contributed."

This is the same play available to any infrastructure-grade measurement provider that owns ground truth. The question is who structurally benefits, who controls the standard, and who owns the audit rights. Those questions decide who captures the value in the next decade.

6. Agentic is real, but the value is in decisioning, not workflow

Most of the agentic chatter at Possible was about workflow automation: faster decks, faster optimizations, fewer FTEs. That is table stakes. It does not create a strategic moat for anyone.

The real shift is decisioning. When an agent can ingest a model output, propose a budget reallocation, and route it for approval in minutes, the unit of work is no longer the deliverable. It is the decision. The org chart, the incentive structure, and the agency contract all bend around that.

For measurement providers, the obligation is to expose our data and methodology in ways that agents can actually consume. APIs, structured outputs, and clear methodology beat PDF reports every time.

7. Trust is the only durable currency

Two things stuck with me from the closing sessions. First, Kawaja's point that ad tech has spent a decade extracting value from the middle while publishers and agencies lost ground. Second, the panel comment that the industry will default to distrust as a low-energy state once enough of the content stack is AI-generated.

Both point to the same conclusion. The companies that earn structural trust — through transparent methodology, aligned incentives, and verifiable independence — are the ones positioned to anchor the next phase of the industry. Everyone else is renting credibility from someone else's brand.

For an independent measurement company, this is the entire game.

What I am taking back to the team

Five concrete things on my mind after this week:

  • Tighten the language around journey, intent, and outcome signals in our product narrative. Impressions are no longer the unit buyers benchmark against.
  • Treat the data foundation as a strategic asset. Hershey spent eight months on data hygiene before the model worked. We have spent six years on it. That is a moat.
  • Pressure-test our exposure to agentic consumption. Anything we publish should be machine-readable and queryable, not just human-readable.
  • Push harder on the structural argument for independent measurement. The political case and the commercial case are now the same case.
  • Watch retail media networks closely. The infrastructure playbook they are running is the playbook OOH measurement should be running.

If any of this resonates, or if you disagree with where I have landed, I would like to hear it. The next phase of this industry will be built by the people who are willing to argue about it in public.