Most GTM orgs measure activity because activity is easy to count.I build the instrumentation that shows whether any of it moves revenue.Most GTM orgs measure activity because activity is easy to count.I build the instrumentation that shows whether any of it moves revenue.

The Salesforce acquisition of Contentful is a real infrastructure milestone for the agentic era. It's also only half the battle.

It validates a thesis my team has been building around: structured content is now a core GTM requirement, not a marketing asset. Engineering teams are rushing to make data modular and portable enough for AI agents to consume. That's a critical technical step, but from a commercial standpoint it's only table stakes.

The real challenge — and where most GTM organizations will stumble — is measuring what happens after the agent consumes the content. If an AI agent delivers information ten times faster but doesn't measurably shift attach rates, cycle times, or win rates, you didn't build a strategy. You built a faster pipe.

Structuring the data is a technical milestone. Tying it to verifiable commercial outcomes is the actual transformation.

RevOps can see the pipeline. They can tell you what's stuck, what's aging, what's converting. What they usually can't tell you is why.

Enablement can coach the behavior. They can build the skills, train the methodology, develop the content. What they usually can't do is prove any of it changed a deal outcome.

Both functions are increasingly getting the same question from leadership: are our sellers doing the things that actually move deals, and can you prove it?

Neither side can answer that alone. The real leverage is in the translation layer between seller behavior and pipeline movement. Whoever builds that connection — whether it sits in RevOps, Enablement, or something new — has the most valuable seat in the GTM org.

Most GTM teams are deploying agentic AI faster than they can evaluate it, and almost all of them are measuring the wrong thing. Adoption tells you the feature shipped. It does not tell you it worked.

Start with the metric, not the tool. Establish the human baseline first against that metric, at the segment level, over a stable window. Skipping that is the most expensive mistake — any later improvement gets credited to the tool when it may belong to seasonality or a comp change.

Introduce the agentic layer as a treatment to a defined cohort that resembles a comparable cohort left on the baseline. Hold the rest constant so the delta can be assigned to the intervention. Name the confounders. Be honest about which is which.

Run it small and run it yourself. End on a decision: expand, adjust, or stop. A study that produces a dashboard instead of a decision has failed.