The most frustrating stall in any AI programme isn't usually about budget or ambition. The strategy is agreed, leadership is aligned, the investment is approved, and nothing moves. More often than not, it comes down to scope.
The data estate needs attention in several places at once, use cases are competing for the same resources, different teams have different views on where AI should go first. Everything feels connected to everything else, and starting in the wrong place starts to feel riskier than not starting at all. Whilst it’s a rational response to a complicated situation, it's also one of the most expensive places an organisation can get stuck.
There's a logic to wanting to fix the data foundations, align the infrastructure, and map every use case before committing to anything, thorough though it is, in practice it tends to produce the opposite.
When scope expands to cover everything, the business case must be built on projected future value rather than anything that's been tested. Timelines stretch, and when projected returns don't arrive on schedule, confidence in the whole programme takes a hit that's usually much bigger than the problem warrants.
Gartner predicts that through 2026, organisations will abandon 60% of AI projects unsupported by AI-ready data. A separate Gartner survey of over 780 infrastructure and operations leaders, published earlier this year, found that only 28% of AI use cases fully meet ROI expectations, with 57% of leaders who experienced failure attributing it to expecting too much, too fast. In most cases the issue wasn't technical. The scope was too broad and the business case too abstract to hold up when early results were mixed.
Trying to do everything at once isn't a more ambitious approach to AI, it's undeniably a more fragile one.
Organisations that consistently get measurable returns from AI don't start by mapping every possible opportunity, they start by asking two questions: where is the business case clearest, and where is our data already good enough to support it? The place where those two answers overlap is where to begin.
That's not about managing down expectations, it's instead about choosing a starting point that can produce something real, something that can be demonstrated, built on, and used to validate what comes next. A smaller, better-defined first initiative isn't a compromise, it's often what makes the bigger ambition achievable.
We see this consistently, the organisations that move fastest at scale aren't usually the ones that launched with the most complex brief, they're the ones that chose their first initiative carefully, proved the value, and used what they learned to move with confidence from there.
This is where sequencing gets genuinely powerful - the data quality work done to support the first use case doesn't get thrown away when that initiative wraps, it directly benefits the next one. The cross-functional alignment built early reduces friction on every decision that follows, and the team arrives at the second initiative meaningfully further ahead than it was at the start of the first.
AI programmes aren't just sequences of delivery; they're sequences of organisational learning. If you treat them that way, capability accumulates, if you don't, and you can find yourself rebuilding alignment, trust, and data foundations at each stage, regardless of how much has already been invested.
A well-chosen first initiative is worth more than its output alone, it sets the conditions for everything after it to be lower risk.
A sequenced programme produces something a large, modelled programme generally can't: actual evidence, not projected return, instead - demonstrated return.
That matters when you're making the case for continued investment - a business case built on results from a completed initiative is a different document from one built on projections from a programme that hasn't started yet. It holds up under scrutiny and makes the next funding conversation easier.
The CFO who has seen AI produce a verified return in one part of the business will be open to a different conversation from the one still being asked to back a forecast.
Starting well isn't a smaller ambition; it's the mechanism that makes larger ambition credible.
Our whitepaper sets out a practical framework for where to start, and how to sequence from there.