JEDAM DIGITAL
February 2026
7 min read
There’s a question I hear from leadership teams at the start of almost every AI conversation. It goes like this: “We want to use AI — where should we start?”
It sounds reasonable. It’s the wrong question. And the fact that it’s asked so consistently is, I think, the single biggest reason most enterprise AI initiatives fail to deliver meaningful return.
When you start with “how do we use AI?”, you’ve already made a critical error: you’ve centred the technology rather than the outcome. You’re now looking for problems to fit the solution you’ve already decided to pursue — which is exactly backwards from how good business decisions work.
This leads to a predictable pattern. A team is assembled. A tool is selected, often based on what competitors are seen to be doing or what a vendor pitched convincingly. A use case is identified — usually something that sounds impressive in a Board presentation, like “AI-powered customer service” or “predictive analytics dashboard”. A budget is approved. Six months later, the system exists but no one uses it, the ROI is unmeasurable, and the AI initiative is quietly shelved while the organisation concludes that “AI isn’t ready for our sector.”
The technology was fine. The problem was how the initiative was framed on day one.
The right starting question is specific and uncomfortable: “What is a concrete outcome we need that we cannot achieve at the required scale, speed or cost with the people and systems we currently have?”
Notice what this question does. It forces specificity about the outcome — not “better customer service” but “handle 3,000 tier-1 support queries per month without hiring additional agents.” It anchors the problem in current operational reality. And it creates a clear standard for success before any technology is selected.
From that anchor, the question of whether AI is the right tool — and if so, which kind — becomes answerable. Sometimes AI is the right answer. Sometimes it’s a process redesign. Sometimes it’s hiring. Sometimes it’s all three. The businesses that make good AI investments have usually done the work of separating these questions before any technology conversations begin.
Pattern one: The pilot that proves nothing. A proof of concept is built to demonstrate that AI “works” — that it can do the thing it’s theoretically supposed to do. But no one defined what “works” means in business terms before the pilot started. The pilot succeeds on its own narrow terms. The organisation can’t answer whether it should scale. The project stalls indefinitely in “evaluation.”
Pattern two: The wrong stakeholder owns it. AI initiatives that live inside the IT department, or inside a “digital transformation team” that’s disconnected from day-to-day operations, consistently underdeliver. The people who understand the actual business problem deeply enough to define success criteria aren’t in the room. The people building the system are solving an engineering problem rather than a business one.
Pattern three: The data assumption. An AI initiative is scoped and approved based on the assumption that the data needed to make it work already exists in usable form. It doesn’t. Three months of the engagement are consumed cleaning, structuring and integrating data before any AI can be built. The project runs over budget and over time. Leadership loses confidence. A technology that was genuinely viable gets written off because the prerequisite work was invisible until it was in the way.
The organisations that make AI work start with a structured discovery process — not with a technology decision. They identify two or three specific operational problems where the business case for AI is clear: the time savings are quantifiable, the data exists or can be obtained, the process is well-defined enough to be automated, and the outcome can be measured.
They build a small, scoped first initiative with a clear 90-day success criterion. They measure it honestly. They expand from evidence, not from enthusiasm.
This is slower than the approach most organisations take. It’s also the only one that reliably works.
A genuine AI strategy document isn’t a list of AI technologies the company will adopt. It’s a prioritised map of specific business outcomes, the AI capabilities that could deliver them, the data and infrastructure prerequisites for each, and an honest assessment of organisational readiness.
It includes cases where the recommendation is not to use AI — where a simpler automation, a process change or a headcount decision is the better answer. Any strategy that recommends AI for everything is telling you what you want to hear rather than what you need to know.
Most AI initiatives fail before they begin because they start with the wrong question. Change the question, and the odds change with it.
You are currently viewing a placeholder content from Vimeo. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationYou are currently viewing a placeholder content from YouTube. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationYou need to load content from reCAPTCHA to submit the form. Please note that doing so will share data with third-party providers.
More InformationYou need to load content from hCaptcha to submit the form. Please note that doing so will share data with third-party providers.
More InformationYou need to load content from reCAPTCHA to submit the form. Please note that doing so will share data with third-party providers.
More InformationYou are currently viewing a placeholder content from Turnstile. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More Information