Escaping Pilot Purgatory: Why 90% of AI Projects Never Reach Production

BACK TO OVERVIEW PAGE

Companies walk into the AI era distracted by every shiny tool, every promising pilot, every efficiency hack, missing the fundamental question: Which business objectives does this serve?

The solution requires connecting AI to core business goals, prioritizing based on value and speed, and building a "Six Quarter Walk"—a rolling 18-month roadmap that balances long-term vision with agility in a rapidly changing landscape.

Charlene Li, leading voice on business transformation and author of “The Disruption Mindset: Open Leadership”, presents this methodology to escape pilot purgatory and move projects to production at scale.

The fundamental issue preventing AI adoption from scaling is a disconnect between activity and strategy. Companies are busy, but they aren't necessarily strategic.

When teams focus on generating use cases without this anchor, they end up in "pilot purgatory." They spin up easy experiments because AI makes it simple to do so, but these experiments rarely cross the chasm to full implementation. The reason is simple: "Unless what you're doing is connected back to very important strategic goals, your leaders won't care,” Charlene mentions.

In practice, executives delegate AI because they see it as a technology problem. They give vague instructions to figure out how to make AI work, then ask the wrong question about ROI. The ROI question shifts: If AI supports a strategic goal, the question isn't what's the ROI but how much can be invested in AI to reach that strategic objective in a better, faster, cheaper, and safer way.

A specific flow bridges the gap between high-level strategy and ground-level AI application. The key principle: This is not about inventing new goals, but about dusting off the existing strategic plan and applying a new lens to it.

Start with the organization's existing priorities. What are the top three goals the CFO or CEO is trying to accomplish this year? Is it expanding into a new geographic market? Is it increasing patient volume? Is it launching two new product lines?

Once the objective is clear, identify the problems standing in the way. What friction points exist? What capacity constraints are holding the organization back?

Only at this stage do we look at AI. We select specific applications that hammer the specific "nails" identified in step two.

Here are industry-specific examples of this flow:

By following this sequence, the conversation shifts from "Here is a cool tool" to "Here is how we achieve the goal you already care about."

When prioritizing AI applications, most organizations immediately filter for feasibility. This approach has a flaw: they ask if there is feasibility during the ideation phase.

If a project is strategic enough, solving a massive problem which is preventing the company from hitting its goals, feasibility becomes a solvable variable. You can hire contractors, you can buy tools and you can reallocate resources."If it's important enough, if it's strategic enough, then you will find a way to make this feasible."

Moving past the feasibility mindset, we need a way to prioritize applications that align with strategy. The solution: the "Double S Matrix," which plots potential projects on two axes: Size of Value vs. Speed of Execution.

This matrix categorizes initiatives into four distinct buckets:

These are the projects to fast-track. They deliver significant impact and can be executed quickly. "Make sure everyone is pulling in those same directions."

These are easy-to-implement use cases. While individually small, they can be grouped together. "Look at those collectively and say, what do they have in common? Could we group them together to make them into higher value?

These are big projects that will take time but promise a major payoff. They require careful planning along the roadmap, with investments made each quarter to progress toward the goal.

They offer little return and drag on resources. The advice is not to do these.

A roadmap is only as good as its ability to prove value. But how do we measure the "Size of Value" in the matrix above? The guidance is clear:

To prove that AI is the factor driving improvement in these KPIs, three testing methodologies are suggested, ranked from most to least definitive:

Run a controlled experiment with two groups: one using the existing process, and one using the AI-powered process. "If AI is giving you demonstrably better results, then it's a clear indication that you should move forward."

Compare results from before the AI implementation to results after. This is less precise because other business factors may influence the data, but it still offers directional validity.

This is where AI helps measure itself. Organizations use predictive analytics on historical data to model what would have happened if they continued working the old way. This creates a "counterfactual basis." They then compare this prediction against the actual results achieved with AI to isolate the delta.

The centerpiece of this approach is the "Six Quarter Walk." This is a rolling 18-month plan that balances the need for long-term vision with the reality of rapid technological change."Your roadmap can be written in pencil. You can adjust it”.

The roadmap lays out what the organization will execute on for the next six quarters.

This structure creates transparency and accountability. It manages two types of employees:

The "secret sauce" of the Six Quarter Walk is the quarterly review.

At the end of Q1, you review the state of the world and the technology. You adjust the plans for Q2 through Q6 based on what you've learned, and then you add a new Q7 to the end of the plan.

Critics often argue that doing a strategic review every quarter is impossible. The counterpoint is that you are not changing your strategy, rather you are changing and updating your execution roadmap for AI; your strategy stays the same.

This process forces the organization out of the annual budgeting cycle, which is too slow for the pace of AI innovation. It allows for "budget fluidity"—shifting money toward what is working and away from what isn't, based on real-time data. "Having that quarterly review ensures that everyone understands how far you have come. It's a moment of celebration that staves off change fatigue."

When populating the roadmap, it is crucial to think about value in three specific buckets:

Most organizations fall into the "efficiency trap." They only see AI as a way to cut costs or speed up existing processes. While valid, this ignores the transformative potential of AI.

"I have never been able to convince executives to change their strategic objectives because of a new technology, but they will change them if a new technology opens up new opportunities."

Reinvention is difficult because it requires looking beyond today's best customers. "Your best customers are probably your worst asset in terms of reinventing your business, because you just want to do everything to serve your best customers."

Focus on adjacent customers (moderately satisfied but annoyed), lost customers (understanding why they left), potential customers (those not yet won), and future customers (those representing where the market is heading). Listen to disgruntled customers instead of best customers when exploring reinvention opportunities.

Traditional competitive advantages—intellectual property, scale, market share—are becoming less defensible than the ability to adapt quickly. "Speed is the new moat" means execution velocity now matters more than static resources.

Long-term planning no longer means rigid five-year strategies. It means maintaining a clear strategic direction while keeping execution fluid and responsive. The "Six Quarter Walk" demonstrates this balance: constant visibility into the next 18 months, with priorities shifting based on what's working.

BACK TO OVERVIEW PAGE