Iterable CEO Sam Allen is making a case for a more deliberate division of labor between marketers and artificial intelligence. AI can help brands move beyond broad campaign blasts toward individualized customer experiences, he argues, but it cannot supply the business history, competitive nuance, or strategic priorities that make a recommendation useful.
That distinction matters because the marketing conversation is shifting from whether teams should use AI to which decisions they should delegate. Allen’s argument is not a rejection of automation. It is a warning that personalization at scale becomes strategically valuable only when people can question the system, understand its reasoning, and apply judgment before execution.
Table of contents
Jump to each section:
- What Iterable’s argument changes
- Why personalization is becoming a decision problem
- The CMO’s advantage is context
- How marketing teams may need to rebalance
- What marketers should know about explainable AI
What Iterable’s argument changes
Allen frames AI as the technology that can finally make individual customer treatment practical at scale. Marketers have long talked about treating each person as a distinct customer, yet most systems still pushed teams toward segments, fixed journeys, and broadly repeated messages. AI changes the execution economics by making continuous personalization, testing, and optimization more feasible.
The more important part of his argument is how marketers should respond to the system’s output. Allen says teams should interrogate recommendations, asking why a suggestion surfaced and which data informed it. Explainability, in this view, is not a technical extra. It is part of the marketer’s ability to decide whether an automated action fits the brand and the customer.
Personalization is becoming less of a content problem and more of a decision-rights problem.
That reframing puts accountability back into the marketing operating model. A system may be able to choose a message, moment, or audience, but someone still has to define which evidence counts, which tradeoffs are acceptable, and when a recommendation should be rejected.
Why personalization is becoming a decision problem
AI’s capacity to generate and optimize more customer interactions does not resolve what those interactions should accomplish. It can make a marketing system faster without making its goals clearer. That leaves teams with a harder management question: which decisions can be automated because the objective is stable, and which still require human interpretation because the business context is changing?
The common assumption is that more automation naturally produces more relevance. The contrasting reality is that automation can scale the wrong interpretation just as efficiently. The strategic implication is that context must become an explicit part of the decision process, not an informal correction applied after a campaign has already moved.
Iterable’s positioning around AI-driven personalization makes this tension especially visible. A customer engagement platform can activate data across channels and recommend actions, but the quality of those decisions depends on how clearly the organization defines its brand priorities, customer boundaries, and commercial objectives.
A recommendation is not strategy simply because it arrives faster.
The CMO’s advantage is context
Allen pushes back on the idea that marketing leaders should abandon their roles because AI can perform more analytical and executional work. His alternative is for CMOs to expand their influence by becoming better at combining machine output with the context that models lack.
That context includes accumulated knowledge of the brand, the reasons behind past choices, the organization’s competitive position, and the nuances of a market that may not be visible in a dataset. It also includes the ability to recognize when a technically plausible recommendation conflicts with what the business is trying to become.
The marketer’s advantage is not access to AI. It is knowing when the system has misunderstood the business.
This shifts the CMO’s value away from owning every execution step. The role becomes more important at the points where priorities are set, recommendations are challenged, and tradeoffs are explained to the rest of the organization. AI may increase the volume of possible actions, but leadership still determines which possibilities deserve resources and which introduce unnecessary risk.
How marketing teams may need to rebalance
Allen’s view of AI also implies a different talent model. He expects systems to take on more optimization, personalization, testing, and execution, while people concentrate on strategy, creative direction, and original thinking. That is a reallocation of attention, not a simple reduction in headcount.
His leadership principle of hiring people who fill a leader’s gaps points in the same direction. As AI tools widen what each practitioner can produce, homogeneous teams become more exposed to shared blind spots. The useful team is not the one in which everyone prompts the same system in the same way. It is the one in which people bring different forms of judgment and can challenge both the model and each other.
AI does not remove the need for specialists. It changes where specialization creates leverage.
For marketing leaders, the practical question is therefore not which roles AI can imitate. It is which capabilities become more valuable when routine execution is cheaper. Strategic clarity, taste, customer empathy, data literacy, and the ability to originate an idea all become more consequential because they shape a larger automated output.
What marketers should know about explainable AI
Allen’s argument gives marketing teams a useful standard for deciding how much responsibility to hand to AI. The goal is not maximum automation. It is accountable automation that improves the customer experience without obscuring why a decision was made.
Interrogate the recommendation. Ask what data informed an action and why the system prefers it. A recommendation that cannot be examined is difficult to govern and even harder to defend.
Separate scale from strategy. Use AI to increase the speed and range of execution, but keep strategic priorities under explicit human ownership. Faster output should serve a clear position, not substitute for one.
Protect origination. Reserve meaningful space for people to develop ideas that do not begin with model-generated patterns. Optimization can improve an existing direction, while competitive distinction often depends on creating a direction worth optimizing.
Design complementary teams. Hire and organize around different strengths, then use AI to extend those capabilities. The objective is a team that can expose weak assumptions before automation multiplies them.
The broader shift is that marketing technology is moving from a toolkit that helps people execute decisions toward an environment that increasingly proposes and carries out those decisions. That makes explainability a source of operating confidence, not merely a compliance feature.
Brands will still compete on creative quality, customer understanding, and speed. The difference is that those advantages will be expressed through systems capable of acting at a much larger scale. Human judgment becomes more valuable precisely because its consequences travel farther.
The next strong marketing organization may look less like a production line and more like a decision institution. Its advantage will come from knowing what to automate, what to question, and what only its people can originate.