New Research From Contentful and The Atlantic Shows CMOs Talking Big and Investing Small

In new research Contentful conducted with The Atlantic, 96 percent of CMOs said AI is a top priority. Only 65 percent are making meaningful investments in it.

Elizabeth Maxson, Contentful's Chief Marketing Officer, calls it the optimism execution gap. Teams are enthusiastic. Boards keep pushing to do more with less, "which is not a new ask to marketers," she notes. But when you look at the day-to-day work inside most marketing orgs, very little has actually changed.

"We were stuck in experimentation mode," Maxson says. Teams want proof they're using AI, but the real work, embedding it deeply into workflows, is where the gap lives. Her advice is to stop experimenting for its own sake and get specific. "What is the problem they are actually trying to solve? What is the pain point? What is the one thing you can go and fix in your organization?"

Why nobody wants to commit

Part of the paralysis is rational. The tools change so fast that a product that was mediocre three months ago might be excellent today, and whatever you buy now could be obsolete or open sourced by summer. Maxson has made peace with that. "That's inevitable with anything that you buy."

The rest of her answer is community. "I've never been so open about my technology stack, ever in my life," she says. "Everyone is curious and no one has the answers." CMOs who used to guard their stacks like trade secrets now compare notes constantly, because sharing is the fastest way anyone learns right now. She even wants to formalize it: a closed door session where a handful of CMOs put their full stacks on the table and question everything.

For the record, her research found the sweet spot is six to seven tools. At a recent CMO dinner in New York, one attendee with 30 tools thought that number was absurdly low. The person next to her thought it was too many. "It's very unique to the needs of the business," Maxson says.

Geography changes what marketers care about

One finding genuinely surprised her. Asked what they prioritize most, North American marketers said campaigns. EMEA marketers said governance.

In hindsight it tracks, given the regulatory weight around data in Europe. But the gap was stark enough that it's already changing how Contentful markets to marketers, with EMEA content leaning into governance while North American content solves different problems. She sees the same cultural split in her own tools. The Qualified chatbot on Contentful's website gets far heavier use in North America than in EMEA, where visitors are apparently more skeptical of talking to a bot at all.

Slop was last year's word. It's still this year's problem.

Merriam-Webster made "slop" the word of the year in 2025, and Maxson doesn't think the underlying issue has gone anywhere. What's changed is awareness. Her own wake-up moment came when she recognized writing published on LinkedIn that her own GPT had sent her nearly verbatim. "It just shows that you're not sounding unique."

Her rule for the team: "More content doesn't equal better content. Better content equals better content." Early AI adoption got seduced by volume, she argues, and volume is exactly what turns everything into noise.

The Contentful team ran an internal exercise that makes the point. At a company kickoff, they trained marketers to write highly detailed image prompts, down to camera angles and lighting, then had everyone produce ads and content pieces. The result was 80 slides of 80 publishable looking ads. Impressive, and also a trap. None of it was trained on Contentful's brand guidelines, voice, or performance data. The output looked great and belonged to no one. Training models on your brand, she says, "really needs to be a phase" of the work.

The homepage is not a museum

The heart of Maxson's playbook, and the part with actual numbers attached, is personalization. Not the elaborate journey mapping version marketers have talked about for a decade. The simple version.

"Homepages get stuck in: that's my brand, that's the first impression, I can't touch this," she says. "And it's like, no. Go take risks with that."

Her case for urgency is that by the time a human reaches your website now, they've done their research in an LLM. "The use of the website how it used to be is no longer," she says. Visitors arrive with high intent, which makes generic messaging an expensive waste of the most valuable traffic you have.

Some of what Contentful has done, all of it deliberately simple:

Stop selling to people who already bought. The team noticed 26 percent of homepage visitors were clicking "login," meaning existing customers were being greeted with pitches for a product they already own. Those visitors now see what's new instead.

Different personas, different homepage. A marketer and a developer landing on contentful.com see entirely different experiences. Images, copy, CTAs, all of it.

Geo-tag the eyebrow. For its Contentful at the Cinema events, where the company buys out AMC theaters for customer premieres, the site simply greets visitors by city with the local date. That one change drove a 51 percent increase in event engagement.

A Lorem Ipsum stunt. Marketers visiting the homepage saw a banner filled with placeholder text, as if someone had shipped a mistake. It was bait, a "caught you" moment pointing to the personalization product page, which saw over 250 percent higher engagement than ever before. Developers never saw it.

Her customers run the same play. Ruggable personalizes on a single binary: click an ad with a cat and you see rugs with cats, never a dog. Cat people versus dog people, nothing fancier, and conversion rates jumped.

When clients start rattling off every possible combination of products, personas, segments, and regions, Maxson stops them. "Let's just pause." Big buckets. One clear difference. Test it.

The agent is a persona now

Asked how she thinks about a future where buyers send AI agents instead of visiting themselves, Maxson doesn't treat it as science fiction. "Right now I treat an agent as another persona."

That means creating content LLMs can easily read and process, with attention to technical details, even token efficiency, that nobody building websites for humans ever had to consider. And the trend line only goes one direction. 2024 was the first year bot traffic surpassed human traffic, and that share keeps growing.

Search behavior is shifting in parallel. The standard Google search runs about six words. The standard LLM query runs around 37. Her example: the old search was "Nike mom shoes." The new one is "I'm in my forties with four children, I live in the Bay Area, what shoes can I hike the redwoods in that transition to the playground in the afternoon?" And Nike, she notes, will have an article for exactly that, because they've committed to producing content for hyper specific use cases. Detailed questions now get detailed answers, and brands that haven't built for that don't bubble up.

The rise of the full stack marketer

The research also stack ranked the skills marketers consider most important: data analysis first, then digital experience design, then personalization, then writing for AI tools. Prompt engineering as a standalone skill didn't crack the top.

Maxson sees a new archetype emerging that she calls the full stack marketer, someone who combines evidence based creativity, data fluency, prompt optimization, and full experience thinking. But what she actually hires for is softer. Curiosity. Comfort with ambiguity. How candidates use AI in their personal lives, because "the more we use something in our personal life, the more we expect it in our work life."

She's also created a role that didn't exist a few years ago, an AI engineer inside her marketing operations team. In one hour at a company kickoff, that engineer trained the entire marketing org to build custom GPTs, then took over connecting them to real data sources and permissions.

The culture piece matters as much as the hire. The team runs a Slack channel called the AI playground where everyone shares "the good, the bad, and the ugly." It has already paid for itself. A go-to-market leader proudly shared an AI generated version of the dreaded NASCAR slide, the wall of customer logos every field team needs. Customer marketing took one look: none of the logos were approved. Because it was shared in the open, it got fixed and connected to Salesforce's approved logo data before it shipped to the entire field in a first call deck.

Through all of it, she keeps the human firmly in the loop. AI, in her framing, has no morals, no judgment, no lived experience. Her team's phrase for its role is the crappy first draft. "You need to put your true fingerprints on that."

Homework for every CMO

Maxson has called 2026 a course correction year for marketing. Asked what homework she'd assign every CMO friend for the next 30 to 60 days, she offers two assignments.

First: "Let's just have some taste. No one's differentiated right now. Use your judgment. Use your lived experiences."

Second: "Get comfortable being uncomfortable with personalization. Start simple. Start experimenting. Don't hold on to your homepage as if you can never touch it."

In a market where everyone's output looks the same, taste and a willingness to touch the untouchable might be the cheapest advantages left.

Listen to the full episode with Elizabeth Maxson, CMO of Contentful, on the Marketing Trends podcast: YouTube | Spotify | Apple Podcasts