Opinion

The AI mirage: Why data discipline is the new KPI for UK construction

Rob Norton

UK director, PlanRadar

28 April 2026

Information fragmentation persists and is a major problem that needs to be addressed before a sensible AI conversation can take place, says Rob Norton.

Abstract image for information management change management piece — Information fragmentation persists even on well-run projects. Image: Cammeraydave | Dreamstime.com

The construction industry has many questions about AI: which tools to trial, which workflows to automate, and which efficiencies to unlock? But the real question we need to ask ourselves is this: do we have the information AI can actually learn from?

For most UK construction teams, the honest answer is ‘no’. Until we fix that, the AI conversation will remain largely academic.

Information fragmentation persists even on well-run projects. Checklists live in one place, photos are filed in another, and approvals are stored somewhere else entirely. Teams rely on a mixture of spreadsheets, email threads and messaging apps to move information around.

The result is a fragmented record that no human, let alone an AI system, can reliably figure out.

This is a data discipline problem with real consequences. Our latest research shows that a substantial portion of project time is currently consumed by manual coordination – nearly half of construction professionals spend 11 or more hours per week on administrative tasks.

This represents more than one full working day every week spent on the exact type of tasks, such as consolidating scattered communications and coordinating approvals, that professionals believe AI should be streamlining. Capturing more data only compounds this burden when teams spend their energy cleaning up fragmented records rather than acting on them.

What AI actually needs

AI adds genuine value only when site data is consistent. It can summarise field notes, group defects by type, highlight patterns and flag safety risks before they escalate. But it requires clean, structured input to do any of that. When the basics are missing, AI has to guess, and guessing destroys trust in decisions around cost and safety.

Structured data on a live project is straightforward in practice. It means consistently capturing the issue category, precise locations pinned to drawings, who owns the next step, and a clear definition of what ‘closed’ actually looks like, typically involving photographic evidence. When those fields are incomplete or inconsistent across a site, AI cannot reconcile the gaps.

High-performing teams commit to a single agreed-upon process for logging issues. They work from shared drawings with clear naming conventions. They keep required fields short and enforce uniform close-out standards. That consistency separates teams that generate useful data from those that generate noise.

A digital platform provides the foundation for these workflows. It structures information at the point of capture rather than asking someone to organise it retrospectively. The platform creates the conditions; the people using it do the thinking.

The shift that matters

AI has real potential in construction. Realising it depends entirely on the quality of information feeding it. The habits that make AI work are the same ones that make projects run better, regardless: capturing data well, structuring it consistently and building records that hold lasting value.

For those already on this journey, these gains are no longer theoretical. Our research shows that two-thirds of project teams using AI-integrated tools are already saving at least two hours per week, per project. This measurable return on productivity is the KPI that will ultimately define which teams secure a genuine competitive advantage and which ones are left chasing an AI ‘mirage’.

Rob Norton is UK director at PlanRadar

Keep up to date with DC+: sign up for the midweek newsletter.

Topics:

AI PlanRadar

Story for Digital Construction Plus? Get in touch via email: [email protected]