Does obvious AI content hurt conversions? The 2026 numbers & when to use AI in marketing.

I don't own nursing shoes, so the fast option was to generate the whole thing, both shoes included. Honest question, not a rhetorical one: does this land the way something I actually made would have?

How many people actually use AI at work?

I counted the other day. Four. Everyone else I know is doing something with their hands or their whole attention, running patient rounds, replacing a water heater, getting a lawn in before the rain, fixing the thing under your car you don't think about until it starts making a noise. AI isn't in their day. They meet it the way you meet a pop-up, something on a screen that showed up without being asked, and mostly they don't like what they see.

My little corner of Campbell River isn't a national survey, so I went and found the real number, and it turns out my four people are pretty much the whole country. Pew has about 21% of US workers saying they use AI at work at all, and only around 10% touching it daily. The stat you see quoted everywhere, the one that sounds like the robots already won, is that roughly 90% of companies "use AI." That second number counts a whole company as an AI shop the second one person opens ChatGPT to reword an email. Four out of five working people are still nowhere near it.

That gap is the thing I can't put down, because the marketing internet has quietly decided the 90% number is the real one. Everyone I follow writes like their reader is already three tools deep and delighted about it. And then I close the app, and the woman next to me at the pharmacy counter is squinting at an ad with that plasticky AI sheen, and she is not thinking "how efficient." She's thinking something closer to: they couldn't be bothered to use a real photo.

She's seen a thousand AI images by now, it's wallpaper. What she reacts to is the message underneath it, some brand deciding she wasn't worth a person's time. Klaviyo asked people about this, and 7% said visible AI content made them trust a brand more, while 31% said it made them trust it less. GWI found more than half quietly back away and engage less the moment they suspect a machine wrote the thing.

Here's the part that got me, because it kills my own first theory. I figured the move was to read the room, AI-heavy audiences get AI and everyone else gets the human touch. The AI-heavy crowd turns out to be the harshest of the lot. Fractl found the share of people who'd trust a favorite brand less for leaning hard on AI doubled in a single year, from 20 to 40%, and Gen Z, the group that basically lives inside these tools, was the most likely to punish it at 54%. Being fluent in AI makes you better at spotting the shortcut, because you know exactly what the lazy version looks like. So "our buyers use AI all day, they won't mind" is the comfortable lie a lot of brands are about to tell themselves.

Now the money, since that's what moves anyone.

What obvious AI actually costs you

Start with the click, because AI creative genuinely wins it. A field study out of NYU and Emory ran fully AI-generated ads on real Google campaigns and got up to 19% higher click-through than the human control. More clicks for free, everyone's thrilled. Then the same study put an "AI-generated" label on those exact ads and watched click-through fall about 31.5% against the human benchmark. Same pixels. The only thing that changed was people knowing.

And the click was never the point. One 2026 benchmark of Meta campaigns found AI creative pulling roughly 12% higher click-through and then converting about 8% worse on anything over a hundred dollars, with the gap stretching to 14% once you're past five hundred. People click the shiny ad, wander in, and buy less right when the purchase matters to them. When someone can tell the ad was AI-made, the same kind of research clocks the premium feel of it dropping around 17%. You get plenty of traffic. What you lose is the sale on the stuff people have to trust before they hand over real money, the expensive stuff, the whole reason you ran the ad in the first place.

Which brings me back to the shoes I keep using as my example, the ones meant to hold up a nurse's feet through a twelve-hour shift. That is a high-trust, don't-get-this-wrong purchase, for exactly the person least charmed by AI. An AI hero shot there reads as one thing to her, we cut corners on the thing you stand on all day. That's the 14% walking back out the door.

So the useful question is where you put the AI, and the line I keep drawing is backstage and front-of-house.

When to use AI in marketing

Backstage is everything the customer never sees as a finished thing. Scanning competitors, banging out a first outline, chewing through three hundred reviews to find the pattern, resizing assets, writing alt text, spinning up ten versions of a headline to test. Use it as hard as you can back there. That's exactly when to use AI in marketing. It eats the hours that never needed your taste, and nobody's trust is riding on how the internal brief got written.

Front-of-house is the surface where the promise actually lives, the product photo, the founder sounding like a person, the testimonial, the words someone reads while deciding whether you're safe to buy from. The closer the work sits to the thing they're trusting you on, the less visible AI it can carry, and the more that human pass has to be real work instead of a coat of paint over a draft nobody touched.

The trap in "there's room for both" is what you do with the time you saved. AI made polish cheap, so polish is worth less than it was, and the effort has to move to the stuff a model still can't fake, the specific detail, the real customer telling their own story. Most of the brands getting roasted right now took the speed as the prize and shipped the raw draft, then acted surprised. The saved hours were supposed to buy something better. They spent them making more.

The nurse is still at the counter, by the way. She clicked. She's browsing more and buying less, and the hero shot she's squinting at needed to be a real photo of a real foot.