Skip to the good stuff :)
Every company is slapping "AI-powered" on everything. IBM decided to test whether it actually works.
They ran a simple A/B test: "Customize your AI" vs. "Take the Tour." The non-AI version won.
63% of consumers are more likely to support brands that show 'human intent' in messaging (Edelman, 2024). IBM seems to have figured this out… the hard way.
This is a signal most marketing teams are ignoring while they race to slap "AI-powered" on every single thing.
AI Fatigue Is Real and Measurable: Here’s the data that proves it
Consumers are signaling a preference for human-first messaging.
53% distrust AI-powered search results and 57% have little-to-no trust in brands to use AI responsibly (Gartner, 2025; Marketing Week, 2025)
The "AI slop" backlash has gone mainstream:
Mentions of "AI slop" jumped from 461K in 2024 to 2.4M by November 2025 (yes, a 9x increase) (Meltwater)
How to successfully include AI in your messaging
IBM's test reveals something important: leading with "AI" in your messaging can create friction.
Not because AI isn't valuable. (it is.) But because:
It's become noise. When every company claims to be "AI-powered," the term loses meaning.
It triggers skepticism. Consumers are primed to question AI claims after lots of overpromising.
It's abstract. "Take the Tour" tells people what to do. "Customize your AI" tells them... not much, really.
The winning CTA was action-oriented and outcome-focused. The losing CTA was feature-focused and assumed people already cared about (your) AI.
The solution isn’t to eliminate AI from your brand messaging, but to keep it focused on the outcome.
This is one test, though. Casey, CMO at DoWhatWorks, has compiled the results from 27,000 A/B tests.
Join us live on January 15th to learn what brands like DocuSign, Peloton, Hootsuite, and iStock are testing (and why!)
Signs AI fatigue is hurting your conversions
CTAs with "AI" or "AI-powered" underperforming simpler versions
High bounce rates on pages leading with AI claims
Drop-off at moments where you introduce AI features
If any of these sound familiar, here are 4 tests to run...
4 AI Positioning Tests to Run on Your Site
Where to run these: Start with your top-performing page. Run in order. Each test builds on what you learn from the last.
For most Series B companies, that's the homepage or a core product page.
Test 1: AI-explicit vs. benefit-focused CTAs
This is your baseline. Test "AI-powered analysis" vs "Get insights in 60 seconds." If the benefit-focused version wins, you've confirmed AI fatigue is affecting your conversions. If AI wins, you've got a differentiated audience. Keep leaning in.
Test 2: "Human + AI" vs. "AI-powered"
Once you know whether AI messaging helps or hurts, test the framing. "Expert analysis, enhanced by AI" positions AI as a tool. "AI-driven analysis" positions it as the product. See which resonates.
Test 3: Generic "AI" vs. specific capability
Now get granular. "AI-powered" is vague. "Instant competitive analysis" is concrete. Test specificity on your feature pages and product descriptions.
Test 4: Transparency vs. no label
This one's for content and support pages. Some brands are testing explicit "Made with AI assistance" labels. Early data suggests transparency can build trust, but context matters. Test it before you roll it out site-wide.
Tired of guessing? Casey's sharing his prioritization framework live — what to test, in what order, based on 27K decisions.
Here’s Why AI Isn’t the Problem
The data shows consumers actually trust AI for personalization (59% according to Optimove, 2025) and efficiency.
They reject it when:
It replaces human creativity
It feels inauthentic
It lacks transparency
Brands that are framing it as a tool that makes their human expertise better not a replacement for it are seeing positive results.
“AI-powered" used to be a differentiator. Now it's noise, or worse, a trust signal in the wrong direction. IBM tested it. AI lost. If you're leading with AI in your CTAs, test whether it's helping or hurting.
Want to see more tests like this?
We're hosting a live session with Casey Hill, Chief Marketing Officer at DoWhatWorks, where he'll break down A/B results from top companies like Netflix, Charles Schwab, and Hootsuite as well as a framework to quickly diagnose conversion leaks.
Casey has analyzed 27,000 tests across CTAs, hero section content, metadata placement, and AI messaging, and is sharing the results with you.
Register for the event and you’ll walk away with plenty of ideas to turn your traffic into revenue in 2026.
See you on the 15th,
Marcel & Jason
