Only 7% of Companies Have Turned AI Into Real Money

A couple of days ago, I read a report that nearly made me spit out my coffee.

Adobe ran a survey. They surveyed 150 heads of marketing — the top marketing leader at each company. Half were C-level, SVP, or VP; the other half were Directors and Managers. Every one of these companies pulls in over $100 million a year in revenue, spread across the US, the UK, Canada, France, and Germany. The fieldwork ran from December 2025 to January 2026.

By all rights, these should be the people closest to AI.

And the result?

Only 7% of these companies have actually embedded AI into their workflows and can show measurable business results from it.

7%.

The other 93%? They're using it. They're experimenting. They're pouring money in. But they haven't turned AI into money.

Now that's interesting.

Speed is up, but the people are barely holding on

What does it mean when people say "marketing has gotten faster"?

It means campaigns that used to take a full quarter now have to ship in two weeks. Creative that used to take a week now has to land the next day. Customer attention has been chopped into fragments, channel algorithms shift every few hours, and the window left for marketers to react keeps narrowing by the day.

One number from the report: 90% of marketing teams say their workflows can keep up with high-frequency campaigns.

Sounds pretty good, right?

But the same people — 69% of them — admit they're barely holding this pace together. Straining. Hanging on by a thread. Some of them can't actually keep up at all.

Think about what that means.

A team says "we can keep up," but every person on it is grinding through late nights, running on fumes, burning out. Short-term, they can muscle through. Long-term, something is going to give. And what gives? Missed opportunities.

Another number from the report cuts even deeper:

Over 80% of marketing teams missed at least one major marketing opportunity last quarter.

55% missed 1–5. 26% missed 6–10. And 3% missed more than 10.

Why did they miss them? Not because they didn't see the opportunities — they saw them, but they couldn't react in time. Approvals got stuck. Cross-team handoffs jammed up. The data wasn't there, the creative wasn't ready. They watched the opportunity slip through their fingers.

It's like standing on top of a gold mine, pickaxe in hand, knowing exactly where the gold is — and you just can't swing the pickaxe.

Money is going in, AI is being used, but the value isn't coming out

The next set of numbers confused me even more.

95% of companies plan to increase their AI investment over the next 12 to 24 months.

68% say they're prepared — or highly prepared — to scale AI.

But only 7% have actually made AI work end-to-end and can deliver measurable business results.

Do you see the pattern?

Willingness to spend: full marks. Confidence: full marks. Execution: close to failing.

Why? It's not that AI isn't capable. It's that the way they're using AI isn't working.

The report says most companies use AI like this: content gets generated in one tool, approvals run through a separate process, activation (deploying content to channels) happens on a third platform, and performance data sits in a fourth system. None of them talk to each other.

AI sped up the "production" step, but it didn't speed up everything that comes before and after production. So AI pumps out 50 images for you in 10 minutes — and then compliance review takes five days, and the performance data can't even tell you which image generated which result.

It feels a lot like buying a fully automatic coffee machine for your kitchen. The machine makes a latte in 30 seconds. But your cups are up in the attic, the milk is in the garage, and the sugar is at the neighbor's house. You still have to scramble for half an hour just to get one cup of coffee.

Where does the value cluster? Closest to the money

So in marketing, where is AI actually most useful?

The report's answer is clear. The closer a step is to performance, to conversion, to actual money, the more value AI delivers.

Specifically, three use cases ranked highest:

First, audience segmentation — getting the right people in front of the right message.

Second, experimentation — quickly testing which creative actually performs.

Third, in-the-moment marketing — pushing the right thing to a user in their current state of mind.

As for channels? Social media and paid media come out on top.

Why these? Because they all demand three things at once: precision, speed, and the ability to iterate. AI happens to have leverage on all three.

But here's where it gets interesting.

For "experimentation," 85% of executives think it's high-value — but only 60% of frontline practitioners agree.

A 25-point gap.

What's behind that gap? Executives are building slide decks and drawing on whiteboards; the frontline is at their desks tearing their hair out.

The frontline knows the truth: to actually run experiments, you need clean metadata, consistent asset-naming conventions, and real-time performance signals feeding back in. Without those things, so-called "rapid experimentation" just means AI churns out a flood of output while the team is flying blind. They have no idea which piece of creative is working — or why.

So executives think "AI will help us experiment," while the frontline thinks "let's get the data foundation solid first." They're not even talking about the same thing.

AI takes the lead on content variations — but someone still has to watch it

There's one more number from the report I made a point of noting.

53% of companies expect AI to become the primary engine for producing content variations in 2026.

What are content variations? It's the same campaign, but you need dozens — sometimes hundreds — of versions of the creative for different audiences, different channels, different markets.

This used to be a headcount game. A small army of designers, copywriters, and localization teams, working overtime to tweak images, rewrite copy, resize formats. Now, most companies want AI to carry that load.

But — and pay attention to this "but."

The dominant model isn't "fully automated." It's "AI-led production with human oversight." 46% say primarily AI; 35% say AI and humans split it roughly evenly; only 7% want full automation.

That distribution is actually pretty rational.

Why not go fully automated? Because content variations aren't a "generation" problem — they're a "usable" problem. You have to guarantee brand consistency, compliance, localizability, reusability, and the ability to assemble quickly. AI can generate. Whether what it generates is actually usable is a different question.

AI can scale up output. But whether it can produce usable output at scale — that needs a human watching.

The three things marketers most want AI to help with

The report also asked: what do you value most in an AI platform?

The results:

55% chose "measurement and performance insights." 43% chose "personalization." 42% chose "workflow automation."

That ranking says a lot.

For the past year, the AI conversation has been dominated by "how much content can it generate," "how much copy can it write," "how many images can it produce."

But when it's time to actually pay for a platform, what people want most is to be told "what's working."

Why? Because as content volume has exploded, measurement has become the harder — and more valuable — problem.

Generating 1,000 images isn't hard. What's hard is knowing which 10 of those 1,000 actually perform, why they perform, and how to adjust the next batch.

There's a line in the report that captures it well. AI's value is shifting from "produce more" to "produce better outcomes."

In plain terms: stop asking what AI can make for you, and start asking what AI can help you figure out about what actually works.

Driving revenue on one hand, cutting costs on the other

There's another set of numbers that lays bare the position marketers are in right now.

65% of marketing teams say their #1 mandate for 2026 is "drive revenue growth." 57% are simultaneously being asked to "improve marketing efficiency."

Make more money on one side. Spend less on the other.

Under that "have it both ways" pressure, where are marketing teams putting their money? Into three things: AI tools (41%), measurement capabilities (38%), and content velocity (33%).

Notice — not one of those three is "make more content." Every single one is "make content measurable, make it fast, and make it work with AI."

Why? Because when a company judges marketing on revenue, what marketers need is operational control over the entire content workflow — from production, to approval, to activation, to optimization. Every step has to be accountable.

Want to scale AI? First, look at the cracks in your organization

When it comes to scaling AI, this is the section I felt most.

The report says the industry is highly confident about "scaling AI." But the moment you actually try to scale, the problems all surface.

The top three obstacles:

51% — inconsistent workflows. 47% — lack of central coordination. 44% — insufficient training.

Notice — not one of these is a technology problem.

They're all organizational problems.

Executives frame this as a governance and infrastructure issue. The frontline feels it as "nobody listens to me, I can't get other teams to move." Both are right. The moment AI scales, it exposes every problem that was already hiding inside the organization: messy approvals, inconsistent processes, no clear owner, teams operating in silos.

And here's another cut:

69% of teams get their AI from existing marketing tools. 53% get it from employees' personal subscriptions. Only 47% use company-level enterprise platforms.

What does that tell you?

Most companies' AI is "free-range." Anyone uses what they want, choices are driven by personal preference, and there's no unified entry point. The result: AI optimizes each small team's workflow a little — but across teams and across systems, it's a muddled mess. Outputs don't line up, governance is inconsistent, and impact can't be measured.

There's a line in the report that nails it. Adoption without coordination keeps AI tactical rather than transformative.

In other words: without coordination, AI stays a tactical tool. It never becomes a strategic lever.

The ceiling on grassroots heroes

Finally, let's talk about "AI advocates."

In a lot of companies, there's always one or two people who are uniquely obsessed with AI. They dig in on their own, they experiment, they push. That person is the AI Champion.

The report says in 23% of companies, these champions have strong influence across multiple teams. In 66% of companies, their influence is "moderate" — effective within their own team, diminished the moment they cross a team boundary.

Grassroots energy has a ceiling.

No matter how capable one person is, without executive backing, without budget, without organizational mandate, they can only influence "how some teams work." They can't change "how the whole company works."

I agree with the report's conclusion. The companies that actually make AI work combine "two tiers":

A senior executive sponsor — sets the stage, secures budget, sets the rules, clears the obstacles.

A frontline champion — pushes forward every day, gets the tools working smoothly, brings colleagues along.

When both tiers move together, AI stops being "a bunch of tools" and becomes "a system."

Three tensions that decide whether AI is "tactical" or "strategic" in your company

The report compresses everything into three tensions, and I think the summary is spot-on:

First: speed ≠ sustainability. Being able to sprint doesn't mean you can run a marathon. Keeping up with the campaign cadence in the short term doesn't mean this workflow can hold up over the long haul.

Second: investment ≠ impact. The money is going out the door — but whether it converts into business results depends on whether you've embedded AI into the process, connected performance data, and built a closed loop back into production. None of that can be bought directly with money.

Third: fragmented workflows can't carry a revenue target. Content, activation, and measurement each sitting in their own corner — you can't see it, can't control it, can't account for it. Talking about "AI-driven revenue" in that state is building castles in the air.

To resolve these three tensions, the key move comes down to one thing: feed performance insights directly back into the next round of content production.

Pull the underperforming creative before fatigue sets in. Understand why the top performers work — so you can replicate and amplify them. Every activation is backed by the previous round's data. And governance has to be embedded in the flow, not bolted on after the fact.

When executive will and frontline enablement pull in the same direction, AI stops being "a pile of tools" and becomes "a system that delivers value."

The gap Adobe wants to fill

The report lands on a product at the end. Adobe GenStudio for Performance Marketing.

The logic is clean: you want to connect the "production – activation – measurement – governance" chain, right? Adobe's bet is a tool that stitches that chain together. AI generates on-brand creative variations, those get activated directly to paid and owned channels, performance data flows back in, and it feeds the next round of creation.

Speed and performance feeding each other — instead of dragging each other down.

That's Adobe's pitch. Whether it actually works depends on your company's maturity.

Back to that 7%

By the time I got here, that 7% was still rattling around in my head.

95% of companies are increasing investment. 68% feel ready to scale. But only 7% have actually turned AI into measurable business results.

That gap isn't an AI gap. It's an organizational gap.

The technology is powerful enough. The models are smart enough. The tools are plentiful enough.

What's missing is the "operating system" that ties it all together: a unified workflow, closed-loop data, cross-team coordination, and top-down meeting bottom-up force.

The companies that have made AI work didn't win because they bought more expensive tools.

They won because they were the first to upgrade the tactical act of "using AI" into the systemic capability of "running the business on AI."

That was the biggest thing the report made me feel.

And it's the real dividing line between the 7% and the 93%.