What happened in the 5%?
Because that’s what we need to learn from.
Yes, yes. Another thought piece about the MIT study indicating that 95% of enterprise AI initiatives have resulted in zero ROI. When I read this article, I was simultaneously not surprised at all and kind of shocked.
I wasn’t surprised because, like most of the people writing about AI and tech, I know much of what’s being said is hype. Between AI tools having a terrible user experience, being wrappers over OpenAI API calls, and being rushed so fast to market that they are basically pre-beta, it seemed inevitable to me that we would come to this place.
What was shocking to me is that the results weren’t just low ROI, or ROI that didn’t meet expectations. It was zero.
I was shocked because every project I’ve worked on to implement AI has been successful in generating ROI.
Honestly, I don’t think this is something to toot my horn about. In some cases, the ROI wasn’t that high, but at least there was some ROI. Considering that some of the companies in the study supposedly include the greatest minds in tech and business, it leaves me wondering: what went wrong?
Here are a few of my thoughts:
What went wrong
FOMO drove too many business decisions. When leaders panic, they make poor choices and jump on bandwagons that promise quick wins. Enter the AI hype, and you get the perfect setup for disappointment.
The hype around AI has created a huge gap between expectations and reality. Because the promise of AI is ease of use, speed, efficiency, and the ability to do everything, people purchase tools expecting exactly that. And they expect it to extend to onboarding and user experience. They’re now finding that not only can AI not deliver everything it promises, but even when it comes close, it still requires a lot of effort.
Because AI was sold as a magic button, executives believed they could build an AI-first workforce and replace people without digging deeper. They skipped key steps to make their teams AI-ready, which is essential for adoption and for realizing any value from their investment.
Combine these three things and you get where we are now.
What could go right
The MIT study shows that the 5% of organizations that did see ROI got there by taking a different approach. They focused on narrow, high-impact processes instead of trying to transform everything at once. They chose AI systems that could learn, adapt, and fit smoothly into existing workflows. They partnered with external vendors who understood their business, rather than relying only on internal builds. And instead of chasing the flashiest front-office use cases, they leaned into back-office automation where savings were immediate and measurable, often by reducing outsourcing and agency costs rather than cutting staff. In other words, they treated AI as a tool to strengthen workflows, not as a shortcut
Here’s what the 5% did that you can replicate:
Data
Leading with data to get a clear understanding of the problems you’re trying to solve with AI, and what success would look like, puts you on the right track to choose the right tools at the right time. In addition, taking the time to fully vet tools gives you the proper information to match the AI tool to your goals. They used data to:
Determine the problems they were trying to solve.
Review tools that might meet those needs.
Define success metrics.
Strategy
Taking that data and developing a comprehensive strategy with those goals in mind will increase your chance of success and your ability to generate ROI. This includes planning for tool selection, procurement, implementation, training, and ensuring that your team is AI-ready. There stratgies included:
Picking the proper tool that best aligned, and being willing to pay for the right tool rather than trying to build their own solution.
Focus focused on the highest value workflows, regardless of optics.
Invested in training and implementation.
Then AI
Only then did they move to implementation. Data and strategy must precede AI for successful outcomes.
Unsurprisingly, these companies also were less likely to cut their workforce. Instead, they were able to find ROI in reducing outsourcing and agency costs.
Process drives outcomes
This approach is what helped one of my clients save as much as $350,000 per quarter. It’s what helped another client boost their visibility in tools like ChatGPT and Claude by 30%. They took the time to figure out what they wanted, developed a clear strategy that included the right tools, and then they implemented. It took longer, yes — but in the long run, it pays off.
In the coming weeks, we will start to see the impact of the study and how it changes attitudes toward AI in business. Many people will decide to abandon AI because they don’t see a pathway to success. That’s why I think it’s important for us to focus on what went right in the 5% — because that’s where the real opportunity lies.
