Recent research across Forrester, Gartner, and adjacent CX coverage converges on the same seam. Adoption has stopped being the variable that separates leaders from the field. Roughly nine in ten organizations are piloting or running AI in some form, and the number keeps climbing across every function that measures it. The gap that now predicts financial return sits one layer down, at the conversion of insight into customer-facing action. Forrester reports that 49% of B2C marketing decision-makers say their analytics findings still do not translate into decisions. Gartner finds that only 24% of service and support leaders can show positive financial returns on AI, despite putting a higher share of budget into it than any other business function. The pattern reads consistently: organizations have built the capacity to know things faster than they have built the authority, the workflows, and the customer surface to act on what they know.
Three developments this week sharpen where that action gap opens widest.
1. The bottleneck moved downstream, from producing insight to acting on it
For two years the constraint on AI value looked like a knowledge problem. Get the data clean, get the model working, surface the pattern. That work largely succeeded, and it exposed the next constraint. Forrester’s measurement research puts a number on it: adoption of marketing mix modeling and incrementality testing rose sharply year over year, yet half of B2C decision-makers report that the outputs stall before they reach a plan. Insight also decays. The interval between an insight surfacing and a marketer acting on it is where most of its value drains away, because the older the read on a customer, the weaker the action it can justify. Read through the capability framework, this is the point where the Insights pillar hands off to Orchestration, and the handoff is where return is won or lost.
It reminds me of what Manisha Powar of Qualtrics said when I interviewed her on The Agile Brand podcast: “By the time somebody looks at an insight and decides to act on that insight, life has probably moved on much further.” The organizations closing that interval are treating action latency as a first-class metric, alongside the accuracy metrics they already track.
2. The design decision is how much authority to act you grant the system
Once action becomes the constraint, the question shifts to who or what is allowed to take it. Forrester frames AI measurement as moving from surfacing insights for a human to closing the loop directly. That reframing forces a governance choice most marketing organizations have deferred: how much autonomous optimization authority to grant, and where a human judgment call stays mandatory. The capability framework treats this as the Identity and Permissions layer working alongside Trust. Granting an agent the authority to act raises the stakes on both, because an agent that acts leaves a wider failure surface than one that only recommends. The value shows up when the authority is scoped deliberately, with an audit trail for what the system did, not only what it advised.
The economics can move fast once a team can act without friction. Dollar Shave Club built its “250 Years. No BS. Still Free” campaign in-house for $400 using generative tools, and the brand told Marketing Dive it became its most successful campaign to date. The instructive part is not the production cost. It is that a small team held the authority to concept, produce, and ship in days, which compressed the distance between an idea and its execution to near zero.
3. The surface where customers act is migrating to third-party AI
The sharpest reading of the week comes from Gartner. Customers now reach for third-party generative AI roughly three times as often as a company’s own chatbot when resolving a service issue, and use of those third-party tools has nearly doubled in a year while branded chatbot use has stayed flat since 2022. The behavioral shift underneath matters more than the ratio: 58% of customers, and 74% in B2B, use these tools to complete a task, not only to get an answer. Customers have moved their action into a channel the brand does not own and cannot fully instrument. A brand chatbot that answers questions but routes every transaction elsewhere cedes exactly the moment where value and data are created. The Identity, Trust, and Memory conditions a brand spent years building on its own properties do not automatically carry into an interaction happening inside someone else’s model.
It reminds me of what Bryan Gernert of Resonate said on The Agile Brand podcast: “Unless you can take that and drive action, it’s just interesting information.” Applied to the third-party surface, the implication is direct. A brand can hold rich intelligence about a customer and still capture none of it if the customer’s decisive action happens somewhere the brand cannot reach.
Featured insights
Forrester — CMOs Are Asking The Wrong Question About AI And Marketing Roles, July 14, 2026 (Rani Salehi, Principal Analyst)
Salehi argues that CMOs organizing their teams around which jobs to hire or cut are starting from the wrong unit of analysis. AI changes the work first, absorbing tasks and compressing decision cycles, and roles follow from how the work reshapes. Teams designed around today’s role structure lock in yesterday’s workflow. Map the work AI is actually reshaping before drawing the org chart; the roles that survive are the ones the redesigned work requires.
Forrester — AI Moves Marketing Measurement From Insights To Action, July 14, 2026
Even as advanced measurement techniques spread, their perceived actionability has not kept pace. Forrester’s 2026 survey finds 49% of B2C marketing decision-makers saying analytics findings fail to translate into action, held back by the weeks it takes to turn an insight into a plan and by the time-decay that erodes the insight while it waits. Track the interval between insight and action as a performance metric; a measurement program that does not shorten it is optimizing the wrong stage.
Gartner — Customers Are 3x More Likely to Use Third-Party GenAI Than Company-Provided Chatbots, July 8, 2026 (Eric Keller, Sr Director Analyst)
A survey of 3,566 customers finds third-party GenAI tools outdrawing branded chatbots by roughly three to one for service resolution, with 58% of users (74% in B2B) using them to complete tasks. A companion survey of 1,303 leaders shows service functions investing the highest AI budget share of any function while only 24% report positive returns. Audit where third-party AI already mediates your customer interactions, and design your chatbot to complete transactions rather than deflect them elsewhere.
MarTech / Five9 — AI Adoption Hits 90% as CX Deployment Paths Diverge, July 10, 2026
Five9’s 2026 Business Leaders CX Report finds 90% of CX organizations piloting or deploying AI, with respondents split evenly across end-to-end platforms, hybrid environments, and best-of-breed stacks. Self-service automation leads deployment at 42%, while customer-facing personalization trails at 26%. Notably, the report’s self-reported ROI runs far higher than Gartner’s independent measure, a gap worth reading skeptically given the source. Adoption tells you nothing about advantage; the deployment choices underneath it, and an honest read of who is reporting the ROI, tell you where you actually stand.
Key takeaways for CMOs
Adoption has become table stakes and stopped functioning as a differentiator. The organizations pulling ahead have moved the constraint downstream and are engineering the path from insight to action, treating action latency as a metric with the same weight as accuracy.
Authority to act is now a deliberate design decision. Scoping how much a system may do on its own, and preserving an audit trail of what it did, matters as much as the quality of what it recommends. Guided by values holds here: respect for customers and their data sets the bounds on what autonomous action is permitted to touch.
The customer’s decisive action is moving onto surfaces the brand does not own. First-party investment still matters, and it now has to extend to how third-party AI represents and transacts on the brand’s behalf. Auditing that representation belongs on the near-term list, next to redesigning one measurement framework for action and identifying which roles the reshaped work will require.
Original Source:
The Martech Futurist Blog – Greg Kihlström Marketing Technology & Digital Transformation
Original URL:
