You Don’t Have an AI Problem. You Have an Execution Problem.

Inspired by the webinar: AI Changed the Buying Journey. Most Websites Aren't Prepared.

I recently was on the RTM Nexus webinar panel, moderated by Tim Zawislack, with two leaders I deeply respect: Breanna Fowler from Dell Technologies and Girish Joshi from LG Electronics. The topic on the marquee was how AI is changing the buying journey.

The conversation we actually had was about how AI is exposing the buying journey we’ve all been ignoring for fifteen years.

Somebody had to say it out loud, so I will: AI didn’t break your commerce stack. AI handed you a flashlight and walked you into a closet you’ve been refusing to open. The skeletons in there aren’t new. The light is.

If you’re a VP of Ecommerce or a CDO at a $250M+ retailer, you already know what’s in that closet. Conflicting product copy across thousands of PDPs. Variants that don’t match the campaign that drove the click. Checkout errors no analytics tool ever flagged. Dashboards full of insight nobody can act on fast enough to matter.

The question isn’t whether AI will help you. It’s whether you’ll fix the thing AI just made impossible to ignore.

AI Isn’t Surfacing New Problems. It’s Surfacing the Process That Created Them.

Every brand we work with has the same first reaction when AI starts auditing their site: how long has that been there?

Years. The answer is almost always years.

What’s new isn’t the problem. It’s the floodlight. AI can scan ten thousand PDPs in the time it used to take a junior merchandiser to spot-check ten, and the junior merchandiser, four in the morning with one bloodshot eye, was only ever going to get you so far. Girish made that point on the panel and it stuck with me, because he’s right: most of the issues AI finds were never on a human’s job description in the first place.

Here’s the harder truth I keep telling our customers: the skeleton isn’t the problem. The closet is.

These issues piled up because there was never a continuous process to catch them and act on them. Enterprise commerce teams have built dozens of intelligence tools: analytics, testing, personalization, attribution, journey analytics. Each one is a smoke detector wired to its own panel, in its own room, calling its own fire department. Insight comes in. Action goes out the same way it has since 2010: through a ticket, a sprint, a queue, a dev backlog.

That’s the structural problem. AI didn’t create it. AI just made it loud.

What is the AI execution gap?

The AI execution gap is the distance between the moment AI surfaces a useful insight and the moment that insight actually goes live on the site. For most enterprise retailers, that distance is four to six handoffs and three to six weeks. The AI didn’t fail. The operating model around it did.

The McKinsey 71%: It’s Not the AI. It’s Everything Underneath.

McKinsey’s recent agentic-AI-in-merchandising study surveyed merchants and found that 71% say their AI merchandising tools fall short of expectations. Not 7%. Not 17%. Seventy-one percent.

These aren’t small brands. These are large enterprise retailers writing seven-figure checks for AI that, by their own admission, isn’t delivering. It’s like buying a Ferrari and parking it at the end of a dirt road. Beautiful machine. Nowhere to go. The driver isn’t the problem. The road is.

The instinct is to blame the AI. That’s the wrong instinct.

The AI surfaced the right insight. It pointed at the right experiment. It even made the recommendation. What it couldn’t do was ship the change. It couldn’t update the PDP. It couldn’t push the variant live. It couldn’t get past the dev backlog.

Intelligence with no path to execution is a CT scan with no surgeon. The image is beautiful. The patient is still bleeding.

This is what Breanna and Girish were getting at from different angles. Breanna pointed out that data quality and content governance are the real bottleneck for showing up consistently in AI search results, when an LLM scrapes two conflicting product descriptions for the same SKU, the consumer doesn’t pause to figure out which one is right. They leave. Girish made the related point that AI’s biggest unlock isn’t doing existing work faster, it’s doing work humans were never going to get to in the first place. Both true. Both downstream of the same root cause: the operating model never caught up to what the technology can now do.

Pre-AI Math: 4 to 6 Handoffs. 2 to 3 Systems. One Window That’s Already Closed.

Picture the standard enterprise commerce team today.

Someone spots a conversion drop in a PDP segment. The team knows what to test. The hypothesis is sound. The variant is straightforward.

Then the real work begins.

Log a ticket. Wait for the sprint. The sprint is full. Bump it. The variant needs a frontend change, so it routes through engineering. Engineering needs assets. Design routes through brand. Brand needs sign-off on the copy. Legal sends edits back. Engineering rebuilds. QA flags an issue. Back to engineering. Three to six weeks later, it ships.

By then, the customer segment has moved. The traffic pattern has shifted. The window is closed.

Pre-AI, getting an experiment live is a relay race where every runner stops to fill out paperwork before passing the baton. Four to six handoffs across two to three different systems and dev queues. That’s not a failure of strategy or ambition. It’s a structural execution problem AI now has the capacity to solve, but only if you let it.

The brands pulling ahead right now are doing one thing in common: they’re collapsing the loop.

The Brands Winning Right Now Collapsed Ownership.

Every winning AI program I’ve seen inside enterprise commerce has the same shape, and it’s not a center of excellence presenting from 30,000 feet.

It’s an executive, usually a VP or above, who owns a number a CFO can see on a dashboard. Revenue. Conversion. Time-to-launch. Cost-per-experiment. Not “AI adoption.” Not “AI strategy.” Real metrics tied to real money.

That executive isn’t asking for permission. They’re shipping experiments. They’re failing fast. They’re learning faster. And they have explicit cover from the top to make mistakes that move the business forward.

Where I see AI programs stall is exactly where you’d expect: spread too thin across too many functional teams, owned by everyone and therefore by no one. Endless meetings to prep for the meeting. AI as a side project for people whose actual job is something else. The committee will be ready to launch right around the time the next wave of AI lands and the committee gets renamed.

Cloud was a 10-year wave. Mobile was a 7-year wave. This one is moving in quarters, not years, and it’s the first one where the cost of being late isn’t a missed opportunity. It’s irrelevance. Treating it like a side project will produce side-project results.

We’re in the Top of the First Inning. The Smart Money Is Already at the Plate.

The narrative around AI in commerce has been dominated by cost-cutting and headcount reduction. That framing is short-sighted, and it’s about to flip. Doing less with less is not a strategy. It’s a layoff with a slide deck.

The brands I’m most excited about aren’t asking how to do the same work with fewer people. They’re asking how to do work that was never possible before. Test ten times more variants per quarter. Launch personalized experiences across thousands of PDPs without a dev queue. Catch checkout errors in real time instead of in the quarterly post-mortem. Show up consistently in ChatGPT, Perplexity, and Gemini as the source of truth for their own products, instead of letting Reddit and a five-year-old review thread tell their story.

That’s not cost reduction. That’s a revenue story.

Yes, we’re early. Top of the first inning, by my count. But the smart money is already at the plate. Brands that experiment now, fail fast, and ship faster will compound a lead that’s nearly impossible to catch later. The ones still drafting an “AI roadmap committee charter” will look up in 18 months and discover the charter is the most expensive PDF they’ve ever produced.

The Real Question Isn’t Whether to Adopt AI. It’s Whether to Rebuild How Work Gets Done.

Here’s the reframe I’d offer every commerce leader reading this:

You don’t have an AI problem. You have an execution problem AI just made impossible to hide.

The fix isn’t another tool. It isn’t another dashboard. It isn’t another center of excellence presenting next quarter. The fix is collapsing the gap between insight and action, putting the analyst, the optimization lead, the merchandiser, and the deployment surface in one place. Closing the Insight Gap and the Activation Gap inside one workspace, with no developer handoffs between them.

That’s the bet we made when we built Fastr Workspace: a single surface where commerce teams find the issue, decide the fix, and ship it, without filing a ticket. AI is built throughout the workflow, not bolted on as a feature. Every brand we work with arrives with the same complaint: intelligence is everywhere, momentum is nowhere.

You can keep bolting AI onto the old process. Or you can rebuild the process around what AI makes possible.

The plane is taxiing. The only question is whether you’re on it.

Want to keep this conversation going?

Join the RTM Nexus commerce leaders mixer at the Leads Summit on May 19, 6:00 PM Eastern. Spots are limited and the room is small on purpose. If you’re a commerce leader navigating the gap between AI insight and AI execution, come find me there.