Why AI projects struggle to move beyond pilots
AI ambition isn’t the problem. Turning a promising pilot into something the business can rely on is.
Organisations have spent the last few years experimenting with AI, testing use cases and proving what the technology can do.
But moving from a successful pilot to AI that delivers consistent, measurable value across the organisation is proving much harder.
So, we asked our LinkedIn audience a simple question:
What is the biggest barrier preventing organisations from realising value from AI?
The result was interesting because there was no clear winner:
- 30% – Lack of clear use cases
- 30% – Poor data quality
- 30% – Skills and adoption
- 10% – Governance concerns
Rather than pointing to one dominant problem, the poll suggests something more fundamental: AI value depends on several parts of the organisation moving together.
That reflects a theme from TXP’s wider research. Organisations have made progress putting AI strategies and foundations in place but translating that ambition into meaningful business value remains challenging.
There isn’t one barrier to AI value
Perhaps the most interesting thing about our poll is the three-way split at the top.
Use cases, data, and skills and adoption were seen as equally significant barriers. That makes it difficult to treat AI transformation as purely a technology initiative.
A compelling use case can still struggle without the right data. Strong data does not guarantee value if people do not adopt the solution. And enthusiastic adoption cannot compensate for a use case that does not solve a meaningful business problem.
The challenge is how these elements come together.
-
Start with the outcome
A lack of clear use cases received 30% of the vote.
It is easy for AI experimentation to begin with the technology: we have access to AI, so where can we use it?
A more useful question is: what are we actually trying to improve?
That could mean reducing friction in a customer journey, making organisational knowledge easier to access, improving a decision or removing repetitive work.
The distinction matters. As AI technology becomes increasingly accessible, TXP’s research points to proprietary data, internal expertise and institutional knowledge as important sources of differentiation.
The strongest AI use cases start with the business problem, not the technology.
-
Data has to work beyond the pilot
Poor data quality also received 30% of the vote.
Pilots can often be built around carefully selected information. Production is different. AI has to operate within the reality of existing systems, data and workflows.
TXP’s wider research reinforces that distinction. While leaders reported strong confidence in their data foundations, technology integration was still identified as a challenge to orchestrating AI across end-to-end workflows.
Having data is not the same as having connected, accessible and usable data where AI needs it.
-
Adoption cannot be an afterthought
Skills and adoption completed the three-way tie, with 30% of the vote.
AI can work technically and still fail to become part of how an organisation actually operates.
That is why people need to be involved earlier. Employees hold knowledge about processes, customers, systems, exceptions and dependencies that technology alone cannot provide.
This is also reflected in TXP’s research, which found that critical knowledge around AI can remain with specific individuals rather than being documented and accessible throughout the organisation.
Bringing that knowledge into discovery and design helps create AI that reflects the reality of the business rather than asking people to adapt to a solution built in isolation.
-
Governance still matters
Governance received 10% of the vote, making it the least selected answer in our poll.
But that does not make it optional.
Moving AI into live operations introduces questions around ownership, risk, outputs and control. TXP’s wider research also identified governance and risk concerns among the barriers organisations face when scaling AI.
The objective is not to choose between innovation and control. It is to build enough confidence and clarity to move forward responsibly.
The poll points to a bigger issue
Our poll didn’t reveal one clear barrier to fix. That might be the most useful finding of all.
Perhaps that’s the real lesson: AI value rarely hinges on one issue. It comes from getting the use case, data, people and governance aligned.
And that alignment needs to happen well before the technology reaches production. Taking the time upfront to understand the problem, the wider business environment and what success should look like can create stronger foundations for AI initiatives.
TXP’s wider research supports this. Many business leaders said that, with hindsight, more time and budget should have been invested in discovery before their AI initiatives began.
A stronger starting point is to ask:
- What outcome are we trying to achieve?
- Do our data and systems support it?
- Whose knowledge needs to shape it?
- How will it work in practice?
- What needs to be in place for it to scale?
Because the goal isn’t simply to get another AI pilot live.
It is to create something the business can use, trust and scale.
Explore the wider research
These themes are explored further in TXP’s The AI Value Gap 2026: Bridging Ambition and Execution in the Mid-Market, which looks at the journey from AI pilots to measurable business value.