Ecommerce Has Swung Too Far Toward AI. Here’s Why the Brands Winning Right Now Are Pulling Back.

TL;DR: I’m a huge proponent of AI… automating the mundane, building tools, moving faster. But I’m watching too many fast-growing ecommerce brands take it too far, and I think the pendulum needs to swing back. Customers can spot AI-generated content and AI-run support faster than ever, and the brands pulling ahead right now aren’t the most automated ones, they’re the ones using AI to buy back time for real, founder-led humanity. Automate the mundane. Protect the human moments.

—

It’s time to write a post about something that’s been bugging me for a long time.

Fast-growing ecommerce brands seem to be on the forefront or bleeding edge of technological advances, including AI adoption.

Don’t get me wrong, I am a HUGE AI proponent. I’ve built AI-powered tools for my own agency, I automate every mundane thing that I can, and I genuinely believe most ecommerce brands are still underusing AI in the parts of their business where it actually belongs. So this isn’t a “put down the robots” post.

But I keep scrolling LinkedIn and Facebook and seeing the same thing over and over: ecommerce brands that have gone all-in on AI in exactly the wrong places.

• AI-generated product descriptions that read like nobody at the company.
• AI-written ad copy that hits every persuasion checkbox and feels like talking to nobody.
• AI chatbots running live chat with zero human handoff, leaving customers stuck hoping someone real will pick up.

And here’s the uncomfortable part: I can tell. You can tell. Your mom can probably tell. Your customers can definitely tell.

The Tell Customers Are Catching

My agency works with a lot of ecommerce brands, and if there’s one skill I’ve sharpened over the last two years, it’s spotting AI-generated marketing on sight. The cadence is a little too smooth. The metaphors are a little too tidy. The “we” statements feel like nobody in particular is talking.

I don’t think I’m special for being able to spot this. I think most consumers are getting better at it too, even if they can’t always articulate why an ad or a post feels off. People aren’t following ecommerce brands because they want another polished commercial in their feed… there are already too many of those.

They follow because they want to feel like there’s an actual human being, who truly wants to help them, behind the products they’re buying.

The Swing Back: Humanization Is Outperforming Polish

Here’s what I’m actually seeing work right now, across the brands I work with: founders picking up their phone and recording a rough, unscripted video. No teleprompter. No perfect lighting. Just someone walking the warehouse floor talking about why a product exists, or being upfront on camera about a shipping delay, or showing the actual mess behind a launch that sold out faster than expected.

These clips shouldn’t work as well as they do. They’re not “content” in the traditional sense; they don’t hit every marketing best practice. But they’re outperforming the polished stuff precisely because AI cannot fake them. A founder standing in front of a camera, saying something true and a little imperfect, is one of the only things left on the internet that guarantees a human made it.

The Line I Draw: Automate the Mundane, Protect the Human

The framework I keep coming back to is simple: if it’s work a computer can do that a human doesn’t need to do, automate it. If it’s work that only a human can actually do well, protect it with your life.

Here are a few examples of where I believe AI belongs in an ecommerce business:

• Inventory alerts, basic order-status replies, first-draft meta descriptions, review triage
• Analyzing post-purchase survey data – mining open-ended responses at scale to surface insights and opportunity areas that would take a person days to find manually
• Piping customer reviews and support tickets into AI to find patterns, recurring objections, the actual language customers use, friction points, then turning that into content angles and ad hooks
• Pulling data sources like GA4, ad platforms, and CRM into something like BigQuery to automate monthly reporting and first-pass analysis, so a strategist starts from insight instead of a blank spreadsheet

Here’s where I think AI doesn’t belong… now or ever…

• Founder & team storytelling
• Escalated customer complaints
• Brand voice and final creative calls
• Real relationship-building with your best customers

The pattern across every one of those “where AI belongs” examples: it’s work a computer can do that a human genuinely doesn’t need to do. Nobody’s job satisfaction depends on manually reading through hundreds of open-ended survey responses looking for a pattern. Nobody built their ecommerce brand because they dreamed of building pivot tables from ad platform exports every month.

One honest note here, since I’d be a hypocrite not to include it: even the AI-generated reports we run internally at my own agency need a human to catch what it gets wrong. AI is going to hallucinate a number or misread a trend occasionally. That’s not a reason to avoid it; it’s a reason a strategist still has to read the report, cross out what’s wrong, and turn the numbers into an actual recommendation. The automation gets you most of the way there. A trained person closes the gap, all in a fraction of the time it used to take.

Case in Point: AI Live Chat

Nothing makes this argument for me faster than being on the other end of a bad AI chat experience.

I’ve been in the endless loop; I bet you have too: I message a brand’s live chat with a real problem, the bot cheerfully misunderstands me twice, and there’s no obvious way to reach an actual person. At that point, I’m slamming on the keyboard: “real person”, “live person”, “real agent”, “customer support”… You’ve been there too. It’s maddening.

I’m not thinking about the bot’s uptime or its ability to answer FAQs instantly. I’m thinking about how badly I want to abandon the purchase and give my money to someone else. It’s not good.

AI can and should absolutely open that conversation. It can handle “where’s my order” and “what’s your return policy” all day, instantly, without needing a human at all. But the second a customer has a real problem, there needs to be an obvious, fast handoff to a person who can say, “This isn’t normal, let me fix it.”

An ecommerce brand that makes a frustrated customer fight through a bot with no escape hatch isn’t saving money; it’s spending goodwill it can’t get back.

What This Means for Ecommerce Brands Right Now

If you’re running a fast-growing DTC brand and scaling ad spend and support volume faster than your team can keep up, I understand the appeal of going “full AI” across content, chat, and social to keep pace. It feels efficient. It looks like leverage.

But that’s usually the wrong lever. The brands pulling ahead right now aren’t winning because they automated the most things; they’re winning because they automated the right things, and freed up a real person to show up more, not less. The founder video, the personal reply to a bad review, the strategist who actually reads the AI-generated report before it goes out the door… that’s where the trust gets built. AI just makes room for it.

Practical Guardrails

• Use AI for the boring stuff, not the brand-defining stuff
• Put your founder or team in front of a camera more, not less
• Never let a bot be the last line of customer support — build the handoff before you need it
• If you can tell something is AI-generated, assume your customer can too
• Automate tasks, not relationships
• Measure more than time saved; track response quality, satisfaction, and repeat-purchase behavior

The Real Competitive Advantage

The next few years in ecommerce aren’t going to be won by the brand with the most AI tools stacked up in their tech stack. In fact, I see those brands crashing and burning hard, needing to rebuild, which could take years.

They’re going to be won by the brands that used AI to buy back enough time to be unmistakably human again.

If you’re trying to figure out where AI belongs in your marketing, and where it’s quietly costing you trust, that’s a conversation worth having with your leadership team and then your front-line team members before your next quarter of growth, not after.

Frequently Asked Questions

Should ecommerce brands avoid AI in customer-facing marketing entirely?

No, the goal isn’t to avoid AI, it’s to be intentional about where it shows up. AI is well-suited to first drafts, data analysis, and behind-the-scenes automation. It’s poorly suited to anything meant to represent your brand’s voice or handle a real customer problem.

How can I tell if my ad or content reads as AI-generated?

Read it out loud. AI-generated copy tends to be evenly polished with no rough edges, leans on generic “we” language, and avoids anything specific or slightly imperfect. If it could have been written by any brand in your category, it probably reads as AI to your customers too.

What’s the easiest place to start pulling AI out of customer-facing work?

Live chat and social content are usually the fastest wins. Add a clear, fast human handoff on chat, and start mixing in unscripted founder or team video alongside anything AI-assisted.