Dave Edelman was interviewed as an expert voice for a study on humanized growth in the era of artificial intelligence, run with Google, Oxford University and the ANA CMO Growth Council. His framework sorts company use into three levels: for me, for us, and for our customers.
His argument is that almost all activity today sits at the first two levels and is about efficiency, and that the third level, changing how value is delivered to customers, is where the opportunity actually is.
Key findings
- The first level is individuals using assistants to get their own work done. He is not dismissive of it, but describes it as internal and isolated.
- The second level is a marketing organization changing how it does something end to end, such as launching a campaign, and he says most of that is focused on efficiency.
- The third level changes how the company provides value to customers, and he cites research with BCG and the ANA showing no more than fifteen percent of companies were really doing it.
- What determines how much is possible is not the industry but access to first party data. He is explicit that his consumer goods example is not about the category, but about the company having sold direct and holding the customer’s details.
- The biggest blocker he names is not technology. It is mindset, and the change required is from pushing campaigns to responding to triggers, which alters budgeting and planning as well.
What the third level looks like
His worked example is a telecom customer paying too much because they are on the wrong monthly plan and repeatedly incurring data charges. The company spots the trigger, recognizes it as an attrition risk, and approaches the customer with a personalized pitch to move to a plan that saves them money.
The second capability in that example is testing. Many variations are generated and the testing process is automated, and the goal shifts from finding the best average offer to finding the best offer for a specific individual.
The counter-example is a coffee machine purchase. Three days after buying a machine that came with a hundred pods, he began receiving daily offers for more pods, and a week and a half later offers for more machines, from a company that could see from his address that he was not a business.
What the company could have done instead is the whole argument in miniature: a note asking how he was enjoying it, telling him he could make iced coffee, asking what flavors he liked. Instead, he says, they simply had a campaign to pitch pods and a new name on the list.
Finding new signals
Because the models have to be fed, he argues that good marketers need to work creatively with their technology teams to open up new sources of data and ask whether there are new ways to get signals.
One European telecom company uses a service that turns every call center interaction into digital data about the caller: what they called about, their attitude, whether they mentioned competitors, what they knew about their own spend and tenure.
The simpler route is to ask. He calls this zero party data, and points to a hotel loyalty program that deliberately asks members where they might go on vacation next, when they might travel with family, and what amenity or drink they prefer, then uses it in the service experience.
Where the line sits
Asked who is responsible for the ethics of this, he starts from what consumers actually report. In one annual global survey, seventy-five percent of US consumers said they had given more business to companies whose use of their data they found useful, while two thirds said they had stopped doing business with companies they felt used their data inappropriately.
His conclusion is that the boundary is not fixed and has to be found empirically. Value earns permission, and the same tools that create the risk also make it possible to test more, including multivariate rather than simple comparisons, and to work out what people find creepy.
Why the brand risk is real
On the fear that brands will converge, his position is that machines should serve human creativity. People are still needed for the insight about what resonates emotionally, what looks natural, and what evokes emotion, and the output has to be tied to a human.
He adds a second consequence beyond sameness. If everyone simply uses the tools, the result is not only brands sounding alike but content overload.
His reading of the criticized soft drink commercial is specific. The problem was not that it was machine generated but that it was cold and distant, from a brand whose whole history is about good times with other people. A generated advertisement could have had warmth if someone had thought hard about what went into it.
Advice for the chief marketing officer
His first piece of advice is that the fundamentals do not change. Strategy, brand, and understanding customers all still apply, and the technology can help gather information but cannot tell a company what its brand is or how it adds value to a customer’s life.
He argues the risk of losing that is now higher rather than lower, which means the thinking about how the brand empowers the customer has to be stronger, and has to run into the marketing itself rather than only the creative.
He recalls being told by an incoming chief executive to bang on the table even louder about customers, so that their impact was considered whenever decisions were made. His diagnosis of why that gets lost is the chase for quarterly earnings and for return on a single campaign, and the visible result is falling open rates, rising unsubscribes and rising costs.
He closes with a fork in the road for the function. On one path marketing becomes a production shop taking orders, which the technology simply makes more efficient. On the other, marketing is at the table bringing what the brand is, what can be done for customers, and what is actually possible.
Scope and limitations
This is a single expert conversation of about half an hour, given as an opinion rather than as research. The interviewee says so himself, and the study it fed was still gathering interviews with a validating survey yet to launch.
Source
This page summarises Transcript: Humanized Growth in the Era of AI – Interview with Dave Edelman, by Mariana Peneva, Dave Edelman, March 12, 2025.