Why Does AI Fail at Outdoor Brands? It's Culture, Not Tech.

TL;DR: AI isn't about tech. It's about change. And almost all outdoor and resort brands trying to adopt it fail. They don't fail because the underlying technology doesn't add value; they fail because this is a change management process: jobs change shape, responsibilities move, and there are often new systems that need oversight, care, and management; not to mention new policies and governance documents that need to be written and enforced.

But because most organizational leaders think about AI as a technology, they run pilots as tool evaluations. Like evaluating a new CRM or Shopify plug-in, or considering replacing their ERP. But even though AI is a technology, making it work is actually about adopting how we all work. And therefore, they almost always fail.

I see the same thing every single time

The more Founders, Brand Presidents, and General Managers I talk to at outdoor gear companies, specialty retailers, and small hotel groups, the more the AI story is the same - let me know if this sounds familiar:

Step 1. Desire to adopt AI is coming both from the bottom up and the top down.

Step 2. Early champions are sprinkled around the team, typically the more technically minded people: think e-commerce manager, data analyst, maybe somebody in inventory management. So you enabled them with a ChatGPT subscription until they told you they need Claude instead. Maybe someone fired up a Zapier or a Make account and built some automations. It's very exciting. As a leader, you can go home feeling like you're working on it, and you can tell your board the same. But...

Step 3. Most often, that's about as far as it goes. Your early champions weave Claude deeper and deeper into their work. Adoption is super uneven across the organization, and everybody's running AI on their own island with no interconnectedness across the organization. And most importantly, the two things you fundamentally want AI to do, either save time or make you more money, are largely unprovable.

MIT's Project NANDA put a number on the broader version of this in 2025: roughly 95 percent of enterprise generative AI pilots showed no measurable P&L impact. Oof.

The issue is that most everybody's take on this is that generative AI doesn't work. And that's the wrong conclusion. The early work was good. The potential is there, but you ran it like a tech evaluation, not a tool that changes how your team works.

Your AI test is an org change wearing a software costume

The core truth, as old as time: adopting any technology is almost never about the technology. It's about change management, organizational design, re-establishing who is responsible for what, and, uncomfortable as it might be, asking people to change how they work. Even more brutal: history shows AI won't pay off unless you change.

Example: Electric motors began entering factories in the 1880s, but productivity did not surge until the 1920s—roughly a 30–40 year lag—because the biggest gains came only after factories replaced steam-era belts and centralized layouts with machine-level motors and workflows redesigned around electricity.

Takeaway: you can introduce the technology, but you only get the benefit after you change how you work.

We've all done this before

I'll simply ask you to remember:

• The POS switch. Remember moving off the cash register onto Square or Lightspeed? The terminal worked the first afternoon. The benefit showed up months later, after you renamed every SKU, retrained closing procedures, and made one person own the inventory counts the system suddenly made visible. The register was never the project. The counts were.
• The CRM that nobody opened. You bought HubSpot or Salesforce because the pipeline lived in your head sales rep's notebook. It only started working the day you ran the Monday pipeline meeting from the CRM screen and nowhere else. Same software the whole time. What changed was the meeting.
• The ERP year. Anyone who has lived through a NetSuite or Odoo implementation knows the software was installed in weeks and the implementation took a year. The year wasn't code. It was arguments about SKU naming, who owns the product data, and which spreadsheet finally had to die.
• The PMS migration. New property management system at the front desk, and for six months the night auditor ran her paper checklist alongside it "just in case." The system was live the whole time. The hotel wasn't running on it until her checklist was rewritten around it.
• Slack, or the two-company problem. When you rolled out Slack or Teams, half the company moved and half stayed on email, and for a year you effectively ran two companies. It didn't settle until leadership declared where decisions live. Adoption wasn't a download. It was a policy.

It's never the tech. It's always the change:

• Change to procedures
• Change to job responsibilities
• Change management to get people comfortable
• Training
• Championing with early adoptions
• Bringing along people who are afraid

The question to ask before your next AI push

My best guess is that AI adoption is maybe 20% IT-related. At least 80% of this, or more, is about:

• leaders having enough understanding of what AI does well and what it doesn't
• gaining alignment on where you can apply it
• what problems it can solve that will free up time and make you more money
• and then leading the change around it

If you do this well, the people who run your business today are the people who are going to run your business tomorrow. You'll need some new expectations. You'll need people who are willing to learn. You need a culture that is okay with trying stuff and having it not work at first. And admittedly, this last one is self-interested: you might need some help from outside building things that your team can use, which is the premise of my work at Loop Trails, AI implementation for outdoor, retail, and hospitality operators: built so your team runs the business, not AI IT.

There's a fundamentally small conversation to be had about what the right tool is and what the future of this technology is; this is the AI nerdery that's consuming all the oxygen in the media. But the fundamental questions I would suggest around AI are exactly the same ones that you should be asking yourself every day:

• What are our biggest bottlenecks?
• What are our biggest opportunities?
• Where do we know that we can make more money but we're not?
• Where do we know that we're burning out people that we shouldn't?
• What are our biggest customer issues that we wish we could solve?

I'm Jordan Williams. I ran consumer brands before I started implementing AI for them. Loop Trails is my practice: AI implementation for outdoor, retail, and hospitality operators, built so your team runs the business, not AI IT.