The 6 Questions of AI Marketing Adoption
The right questions drive the right decisions.
Over the year, we’ve been talking with marketing leaders every day. Some of these talks are public, like our podcast or this newsletter, and some of them live in Zoom transcriptions and in our email inbox as replies.
We can’t pinpoint the exact trigger, but we’ve started noticing that every leader we talk to is naming the same problems and proposing the same solutions for team-wide AI adoption in marketing. And we have also noticed that almost all of them were wrong.
They were wrong because they asked the wrong questions. And we all know that to make the right decisions, you first have to ask the right questions. Without some reflection, you can’t really provide a solution. This is true for individual problems and problems that affect entire teams.
This was the motivation behind writing about this topic. Helping you to make the right decisions by pointing you to ask the right questions first.
The classic cases of wrong questions
Before we jump in, let’s talk about the wrong questions. We want to name two. If you read any kind of online content around AI, you can probably guess what these are just by looking at what others are talking about.
1. A case against learning
The first thing is the obvious one. When your superior asks, "What’s our plan for AI adoption for marketing?” your answer points to your people. “We have initiated a learning program for our team members,” or something even more concrete, “We’ve bought X courses and enrolled in Y AI Academy, and we are going to the Z AI marketing conference.” You present the case as if it were a people problem, as it is a matter of “we don’t know enough.”
You are partially right. It’s somewhat a people problem, but it is not something that you can learn, but something you can experience. Your people won’t use AI more or better just because they read a guide, a manual, do a course, or master prompt engineering. There is no secret knowledge. No secret prompt. No hack, no tips, no hidden insight. No need to waste your precious time on consuming content on the best ChatGPT prompts for marketers.
What’s the good approach here? Simple. The solution is not learning but experiencing. By experimenting with AI, treating it as a colleague that you can work with, not a tool that you can wield to work, you can learn by doing. No prompt engineering course can teach you more than fixing your own prompts on real use cases, in real time, by you, based on your own creativity.
We don’t believe learning solves anything. The things that are kinda mandatory to learn about AI can be summed up in a 3- to 5-page article. From there, it’s all up to you to experiment.
Besides, let’s say you enroll in whatever I’m sure is a wonderful AI academy in May. The knowledge you’ll get is at best 2-3 months old. By the time you wrap up the course, the knowledge you’ve learned is already outdated. We know, just search our archives! We were questioning AI video capabilities half a year ago…
This is the reasoning behind our Knowledge Hub. It’s one of the richest, most fully loaded AI marketing knowledge dumps, with endless guides, manuals, courses, and workshops. And all of that content is free. Have a look, if you want to learn, be our guest. But don’t expect that consuming AI content for 2 weeks will get you further than experimenting and building something with AI for 2 hours.
2. The case against tools
Another classic CMO move is to throw resources at a problem you see.
I (Peter) was always so opposed to this that, when I got hired as a CMO, one of my very first moves was to audit the list of tools the team was using. I can tell you, it’s mindbendingly easy quick win. I always found thousands of dollars per month of worth of unused tools that my predecessors signed up for and forgot but still charged the company credit card. Astonishingly careless, really.
So when it comes to “what’s our AI strategy,” you might say, “oh, we’ve signed up for X tools and started to unroll another Y tool to our team with Z features.” The amount of time we get the “please recommend us the best tool for Y” is mind-blowing. Our response is always the same.
Like there is no secret prompt, hidden knowledge, or best course, there is also no almighty tool that exists. The solution is simple, again: go deep with one tool, possibly with some of the AI labs’ models. We, sadly, are not affiliated, but we recommend Claude. Set up custom projects there, use Cowork daily, and build something with Code. Instead of going deep horizontally with multiple tools, just sign your team members up to one tool, full access, unlimited credits, and good to go.
This is even more true than the knowledge problem, by the way. Almost all branded tools use the same AI models that AI labs develop, through their APIs. If you see a tool that does X, it’s probably a Claude/OpenAI or other labs’ API in a UX wrapper. Nothing wrong with it, of course, almost all modern CRM tools are basically cloud-based wrappers for Excel, but you have to keep in mind that the real differences between these tools are blurry. So it doesn’t mean much which one you sign up for.
If you are really into shopping, fine, we are all into giving what the people want. Our AI marketing tech stack list is the most comprehensive list on the internet. We list, review, compare, and even check whether the tools are right for your team. Browse the stack here. It is constantly updated.
No tools or courses will solve your AI adoption. Asking the right questions will.
Question 1
Where is AI pressure coming from?
We all know that you don’t just need to solve the problem. You also need to sell that solution internally. To do that, you need to know who you are selling to. Everything depends on the target audience, and your messaging should be tailored to them. This is a case in internal communications as well, obviously.
You have to know who’s driving the AI adoption conversation. There are 3 main possible drivers, and each requires a different story and messaging, therefore different solutions. There is a difference between “I want AI”, “We want AI”, and “We have AI.”
“I want AI” - The pressure is top-down. The management, the CEO, or whoever is in charge wants you to implement an AI adoption plan. The board, the investors, or the CEO themselves want AI because, well, no one knows why, but McKinsey told them it’s kinda important nowadays, and everyone seems to have AI, so why can’t we?
Addressing that means framing AI adoption as a story that serves as a use case, and attaching some goals and numbers to it. You don’t want to talk about tools or courses here. No one really cares. You want to talk about a use case you already did, expand on it with XYZ goals, and reach these goals by meeting XYZ numbers. All of this is framed into a story of change. Ideally, on a ppt/sheet that they can submit to the board as their own story…
“We want AI” - The pressure is bottom-up. Your team, employees, and colleagues are either already using AI in the background without a plan or want to use AI for their work.
Usually, it's the first. They are already using AI with their private licenses, without you knowing about it (shadow AI). Which is amazing, because your AI adoption has already happened on an individual level; your goal is to spread it evenly across the entire organization. In this case, you will need a governance plan, policies, and anything that formalizes, makes AI usage visible, and makes it manageable.
“We have AI” - The pressure comes from adjacent territories. The product or service you market has AI features and capabilities, or other teams than marketing are using AI team-wide, or sales is shipping outreach with AI-driven workflows.
Regardless of what the case is, the visibility is the same: if you don’t adopt AI, the marketing team seems like a slow horse. In this case, you need to have a plan on how your team can learn from other teams, collaborate on AI adoption, and probably surface AI usage that already happens in the background. Overcorrection shouldn’t feel like a catch-up, more like a sync.
Question 2
Which parts of your function will change first?
Not everything changes at the same speed. That’s the second trap. The marketing leaders we see move well in 2026 are the ones who can look at their org and, with a straight face, tell you which functions are about to get reshaped in the next two quarters, which will take a year, and which will still look mostly the same two years from now.
Here’s a rough starting point, though your specifics will differ:
The functions getting reshaped first are almost always the high-volume, pattern-heavy ones — inbound content operations, paid media iteration, lifecycle email, basic research, briefing workflows, first-draft copywriting. Not because humans can’t do them. Because the economics of having only humans do them stopped making sense about 6 months ago.
The ones on a twelve-month horizon tend to sit one layer up: creative strategy, brand planning, sales enablement, campaign orchestration. Agents are getting close, but the judgment density is still too high for full handoff.
The ones that look the same for a while yet are the judgment-and-relationship functions: partnerships, positioning, customer advisory, and category creation. Not forever. But for now.
A plan that doesn’t make this kind of distinction is just a wishlist. It treats a content operation and a brand strategy as if they’ll evolve on the same timeline. They won’t. You should know and point out the functions that will change first, so you can act accordingly.
Question 3
Where can agents expand?
This is the question that separates the leaders building something from the ones buying something. The default framing, the one your CEO has probably absorbed from whatever podcast they listen to, is that AI makes your team faster. That’s the shallow version. It’s also the version that quietly commoditizes your function.
The deeper framing is this: agents don’t just make your current work faster. They let you cover ground you couldn’t previously staff. That’s, if you ask us, a freakin’ fundamental difference. It’s the difference between your neighborhood hill and Mount Everest.
Examples? The competitor research you always wanted to run weekly, but could only afford quarterly. The account-level personalization your sales team has been asking for since 2021. The thirty-language version of your campaign that only the Fortune 100 could justify before. The customer-signal synthesis that once required a dedicated analyst now runs overnight.
The question isn’t “where can agents save us time.” It’s “where can agents let us do something we couldn’t do at all before.” Those are different plans. The first one defends a headcount reduction. The second one defends a growth story. Your CEO will take either one. The headcount reduction happens whether you have a say in it or not. The growth, well, that’s your job. Hint: You won’t get promoted for saving money on payroll.
Questin 4