Customers don't care about your AI feature
New research on AI messaging from 767 software users
👋 Hi, it’s Kyle Poyar and welcome to Growth Unhinged, my weekly newsletter exploring the hidden playbooks behind the fastest-growing startups.
As everyone rushes to position their products as “AI-powered”, Kristen Berman from Irrational Labs stepped back to ask: do people actually care? And Kristen and her team surveyed 767 software users to find out.
Kristen is co-founder and CEO of Irrational Labs where she’s helped the top companies (think: Google, Intuit, TikTok, LinkedIn) use behavioral science insights to understand user psychologies and build products that customers use, love and pay for. Want more consumer psychology and Kristen? Follow her Product Teardowns newsletter (very engaging short videos!) or email her (kristen@irrationallabs.com) to learn about working with her team.
🎁 PS: I just teamed up with Maven to curate interactive, cohort-based growth courses from my favorite experts — most of whom you might notice from this newsletter. Explore the full Growth Unhinged x Maven collection here and get $100 off with the code KYLExMAVEN.
Imagine your team is launching a new generative AI feature. Everyone has worked really hard for the last 6 months to scope, design, build, and test this new feature. You’re excited to finally launch, but now you have a decision to make: How much do you share that you’re using AI? You are considering a marketing landing page that leads with AI-powered or AI-enhanced. Should you do it?
It might seem like the more you emphasize AI, the more innovative and valuable your product will appear. After all, it’s a strategy everyone has used—from Intuit Assist to UserTesting.com. And adding AI to your product's name seems like a surefire way to make your company and your product appear on the cutting edge of innovation.
As it turns out, this isn’t necessarily true. In fact, my team’s latest research has flipped this assumption on its head.
Labeling your product as “AI” might not be the boost you think it is.
It doesn’t necessarily build trust, justify a higher price, or convince your customers it’ll perform better. And if you misuse it, you may end up doing the one thing no marketer or product manager should do—turning your audience off.
The surprising science behind AI labels and what our research revealed
With AI well beyond buzzword status and creeping dangerously close to cliche, the Irrational Labs behavioral science team set out to explore its actual impact on customer perceptions.
The question: Does labeling a product as “AI-powered” inspire trust and excitement, or does it risk blending into the noise of tech jargon?
To answer this, we surveyed 767 participants and conducted a controlled experiment. Of the 767, 73% had a bachelor’s degree or higher, 48% were men, and 60% earned above $60k annually (the average income in the US).
Participants were presented with marketing landing pages for four real-life products featuring AI-driven capabilities.
In one condition, the product descriptions prominently included terms like "AI-powered" or "generative AI.”
In the other condition, those descriptions focused on the features and benefits without directly mentioning AI.
By comparing how participants responded to these two approaches, we sought to uncover how explicit AI-labeling influenced perceptions of product performance, trust, and willingness to pay.
Here are a couple of examples of what this looked like:
What we found:
Performance expectations: Rather than enhancing perceptions, the term “generative AI” significantly lowered expectations of a product’s potential impact. This decline may reflect growing skepticism due to high-profile disappointments—tools that overpromise and under-deliver, such as unreliable chatbots or lackluster generative content.
Willingness to pay (WTP): Labeling a product as AI-driven did little to justify a higher price. Customers were unwilling to pay more unless the tool demonstrated clear, compelling benefits. Simply including “AI” in the description wasn’t enough to convince users to part with their money.1
There was one exception. When Superhuman was labeled with AI, it got a bump above no mention of AI. We’re not sure why. Our best guess is the AI description was helpful to explain why it was called Superhuman!
Trust perception: The impact on trust was largely neutral. For most brands, explicitly mentioning AI didn’t increase customer confidence in the tool’s reliability.
Our takeaways? Explicit AI-labeling is no guarantee of success. It doesn’t increase performance expectations, willingness to pay (in most cases) and trust.
One caveat—we’re testing with “average people.” Our sample is not early adopters. 68% of people in this survey say they have used AI at least a couple of times. Only 15% of people in the study use it daily. Likewise, if you are targeting early adopters (or Bay Area tech workers) these findings may not apply. That said, most companies will need to go beyond the Bay Area to scale.
So why do companies still overuse “AI” in their marketing?
If customers aren’t responding to the AI tag, why is it still so prevalent? Behavioral science suggests it’s less about customer needs and more about external pressures.
Investors and analysts often expect AI-heavy branding to drive market buzz and boost valuations. It’s also possible companies are optimizing for winning the SEO game by being the best AI company in a certain niche, which would require using “AI” 10 times on your landing page.
And don’t sleep on good old-fashioned FOMO. The fear of falling behind competitors can push teams to overemphasize AI, even when it risks disconnecting from what end users actually care about: solving their problems.
How you should market your AI feature
We’ve been hard on the AI label, but don’t worry—there’s good news. Explicitly branding with AI isn’t the only way to sell your innovation. The study’s findings point towards smarter, more effective tactics grounded in user psychology and behavioral science.
Here are four ways to say AI without saying “AI”:
1. Lead with benefits, not jargon
We’re going back to basics. Customers shouldn’t need to decode how your product works; they should instantly understand what’s in it for them. In a classic but still spot-on history lesson, Apple nailed this when it launched the iPod. Rather than leading with how many gigabytes of storage it could hold, Apple went right at the customer, promising “1,000 songs in your pocket.”
Canva says its “Magic Design” feature helps users effortlessly make beautiful designs. Their audience doesn’t need “AI-powered productivity.” They need the result of that AI-powered productivity. They need “beautiful marketing templates ready in 10 seconds.”
Similarly (and despite having “AI” in its name!), Notion AI emphasizes outcomes in its marketing, highlighting the tag, “Think it. Make it.” And then there’s Otter.ai, which has perfectly distilled its value proposition to address a familiar and relatable user pain point: “Never take meeting notes again.”
Example execution: Redesign your landing pages to communicate the benefits in non-jargon language. Headspace says they put all copy through a “reading level” test and won’t ship if it’s too complex. This doesn’t always mean you can’t mention AI, as long as you’re also clear on the tangible benefits in plain language.
2. Make productivity gains tangible
People are tired of vague promises like “enhance productivity” or “unlock creativity.” Show, don’t tell. Numbers create credibility, specificity builds trust, and measurable results stick.
Instead of claiming, “This tool uses AI to revolutionize your work,” try, “Write three times faster.” or “Save 30 minutes a day on scheduling.” Time, money, or effort saved—these resonate far better than a buzzy name.
Take Duolingo, which has said that quantifying the results of its premium features helped drive premium conversion. Measurable outcomes can move users from free to premium tiers.
Other companies are getting it right, too. GitHub Copilot highlights its impact with “developers are coding up to 55% faster.”
Example execution: Before doing a full rollout, run a simple A/B test using a feature flag—one condition gets your new feature and one condition doesn’t get it. Measure the impact—not just to your business but to the user! Use these stats to market the feature when doing your full rollout.
3. Leverage the Endowment Effect
Behaviorally speaking, people value things more when they feel some ownership. Think about Grammarly or Notion. They let users try the feature first, only positioning it as a paid feature once users have tried the feature.
Once you’ve seen that Grammarly can deliver better sentences, it’s hard not to want more! You’ve accepted other Grammarly suggestions, this one is likely just as good.
Our team worked on a similar challenge with the healthcare company TytoCare. When TytoCare shipped medical devices to users, they included instructions encouraging recipients to call a phone number for setup assistance. However, few users made the call, and many faced difficulties with onboarding.
Our team flipped the mental model: Instead of asking users to call for setup assistance, we framed the call as an activation step to unlock their device. This subtle shift created a sense of urgency and value—users didn’t want to risk losing access to “their” device. Combined with other interventions recommended by Irrational Labs, the new approach resulted in a remarkable 90% increase in successful device pairing.
Example execution: Help the user feel like the product you are selling is already theirs. You can do this easily by testing a reverse trial—let your customers try your AI feature (as long as it’s good!) a few times and then ask them to pay for it.
4. When you must use “AI,” be thoughtful
Sometimes, there is no easy way around it—you will have to use “AI” to describe a new tool or feature’s capabilities. Just remember to pair it with a clear use case or branding that connects to the user’s needs.
For instance, Google Lens seamlessly integrates AI while focusing its messaging on the user benefit: the ability to "see" and analyze the world around them. This thoughtful framing highlights the tool's functionality without overwhelming users with technical jargon.
Similarly, Spotify DJ leverages AI to curate personalized music selections based on listening habits and preferences. By emphasizing the role of a "DJ," the messaging connects intuitively with users, allowing the technology to work behind the scenes without being the center of attention.
Creative angle: Pick a name that clearly articulates the value to the user (and doesn’t include “AI”). Then, carefully use “AI” on your supporting materials to ensure your users know how cool you are.
Rethink the AI crutch—it’s holding you back
Behavioral science reminds us of a hard truth: customers don’t think the way marketers hope they will. They don’t automatically associate “AI” with “better.” In fact, some skeptics might assume the opposite. While AI might power the underlying technology, it shouldn’t be the headline—a notion that might seem counterintuitive if you’re in Silicon Valley and used to drinking the Kool-Aid of innovation.
The real value lies in how your tools improve lives, make work more efficient, and help users achieve their goals. This narrative isn’t built on buzzwords but on truly understanding your users’ needs. Ask yourself: “Am I relying on ‘AI’ as a selling point, or helping users see how I solve their problems?”
It’s time to go beyond the “AI” label in crafting your value proposition. Instead, focus on what matters most to your customers: the results they achieve. Audit your messaging, test for real user outcomes, and adjust your language accordingly. Your tools are worth more than their AI component—they’re defined by their impact.
Users won’t remember your tool as “AI-powered.” They’ll remember how it made their lives better.
The willingness-to-pay chart has been corrected for clarity. The original version of the chart stated that there was no difference in willingness-to-pay across products. The corrected version reflects that there is a small difference (~$1), but it is not statistically significant meaning we cannot detect willingness-to-pay differences between AI vs. non-AI labeled products.
 | A guest post by
|
Great article - thanks, Kyle. For those of us old enough to remember the .com boom (and subsequent bust) in the late ‘90s, the state of AI marketing today sure has a 1999 feel to it.
People care about output/outcomes/benefits and not so much about input. AI is an input. Much like fuel for your car/truck etc, customers care about fuel to the extent that it does the job of getting them to where they need/want to.