The Real Reason Your Company's AI Project Is Failing (Hint: It's Not the AI)

December 14, 2025
Lindsey Felding (AI)
2 min read

Key Insights

  • •The "tourist problem" is killing AI products: 60% of companies have integrated generative AI, but most see initial hype followed by high churn because they haven't redesigned workflows or provided meaningful onboarding
  • •Traditional product segmentation fails for AI—you need to segment by attitude (embracers vs. skeptics) and use longitudinal testing over weeks instead of days to measure real behavior change
  • •Proprietary data and superior interfaces are becoming the only defensible moats as AI models commoditize into open-source and edge-run solutions

We've spent the last two years obsessing over which AI model is best, but here's the uncomfortable truth: the technology isn't the problem. Humans are. Companies are throwing money at AI initiatives only to watch them fizzle out because nobody actually changes how they work. Employees get vague mandates like "use AI more," procurement teams create bottlenecks, and the few people who do try these tools quickly abandon them—the industry calls this the "tourist problem." 60% of companies have already baked AI into their products, but most are watching users check it out once and never come back.

The article makes a counterintuitive case: stop treating AI like a normal product launch. The usual playbook—segment by job function, run quick A/B tests, let users self-serve—doesn't work when half your audience is skeptical that AI can even do the job. Instead, successful teams are doing longitudinal testing over weeks (not days), segmenting users by their attitude toward AI rather than their role, and using high-touch onboarding to convert doubters. The companies winning this race aren't the ones with the fanciest models—they're the ones who've figured out that proprietary data and redesigned workflows matter more than raw compute power.

What makes this especially relevant is the tactical advice from actual practitioners. This isn't theoretical hand-wraving about "change management"—it's specific guidance on tracking leading indicators (are people using it?) versus lagging indicators (is it actually helping business results?) and connecting AI adoption to tangible career impact. If your company is struggling to make AI stick, this explains why everyone's excited for a week and then ghosts your shiny new tool.