It is an uncomfortable question to put to a senior responsible person and the answer is rarely a number. More often, it is a pause.
Most asset-intensive organisations run on registers that have not been validated end to end in years. Many were built in a hurry during implementation or mobilisation, by people who have since moved on. Since then they have been patched through partial surveys and passed between clients and service providers, losing quality at each handover. Formats are mixed: some records follow a recognised standard, some a provider's proprietary version and others are little more than free text.
Meanwhile the estate keeps changing. Decommissioned assets stay listed, new ones are logged inconsistently and the same pump appears three times under three names. Supplier, finance and procurement records drift too. The people who use this data know it, which is why so many keep a spreadsheet on the side.
The research agrees. SFG20's State of Facilities Management Report 2026 found that only 15% of respondents have a fully accurate asset register and half keep at least part of it in a spreadsheet. Reliabilityweb reports that 68% of organisations are dissatisfied with data quality in their maintenance systems.
Drift is a natural result of running a live operation. It carries a cost, often larger than it looks.
What unverified data costs
The damage rarely shows as one line in a budget. It is spread across hundreds of decisions made on information that was not quite right. Yet one record can matter a great deal. An asset missing from the register can go unmaintained, and if it fails, the result can be unplanned cost, compliance exposure and lost capacity. Duplicate or outdated asset and supplier records keep attracting cost and risk.
Across the business, the effect adds up. Inspections planned against an out-of-date list can make compliance look complete while gaps remain. Budgets and capital plans may rest on unvalidated figures. Stock may be held for retired assets and contractors paid against an uncertain baseline. Gartner puts the average cost of poor data quality at $12.9m a year per organisation.
Other losses go unnoticed. Utilities, energy and service charges can still be billed for property, meters or equipment that are out of use or were never there. A payment that recurs without issue is rarely questioned.
Public bodies face the same problem. In 2025 the Public Accounts Committee found that the Office of Government Property has no condition data for 36% of central government property, and that these gaps make it harder to tackle a maintenance backlog of at least £49bn.
The most serious losses arrive at specific moments. In a contract negotiation or renewal, the data sets scope and price. If it is wrong, whoever did not check usually pays for assets that are not there. In a regulated operation, one missed statutory task can bring enforcement action, lost contracts and lasting reputational damage.
Some of these costs build over years, others overnight. At worst, they can put the business and its reputation at risk. All trace back to the same source: data that was trusted but not validated.
Why a new system does not fix it
When the data cannot be relied on, the usual response is a new EAM, CAFM, CMMS or ERP platform. In most cases the existing system works well enough. The problem is the data inside it.
Migration does not fix that. The duplicates, gaps and unverified records move across to the new system, and the move can make them worse. As fields are mapped and formats forced to fit, minor inconsistencies can become conflicts and gaps get filled with defaults that look like real data. By go-live the errors are harder to spot, and the people who knew where they were may have left.
Replacing a system carries its own risk. McKinsey and the University of Oxford reviewed more than 5,400 large IT projects and found they ran 45% over budget on average and delivered 56% less value than predicted.
Bringing the data up to standard usually costs far less than replacing the system and builds confidence in the one you have. Once the data can be trusted, any platform decision can be made on its merits.
What it takes to put it right
Getting data right is detailed, unglamorous work, closer to an audit than a tidy-up. It begins with data forensics: tracing where the data came from and testing each record against independent evidence such as contracts, drawings, invoices and site records. That is where duplicates, gaps and contradictions surface. Resolution and categorisation follow. Records are aligned to recognised international standards, such as Uniclass and ISO 14224, so each asset, supplier or cost is described the same way everywhere. Anything that cannot be confirmed is flagged for review rather than guessed. Every record keeps a trail back to its evidence, which is what makes the data auditable.
Much of the work can be done at the desk. That keeps cost and disruption down and saves site verification for where it matters. In one engagement from a previous senior role, a 99k-line register was consolidated to 67k validated lines, almost a third smaller.
Validated data does not stay that way on its own. Without clear ownership, standards and rules for creating and changing records, it starts to drift again. That is why the work follows the Lean 5S method: sort, set in order, shine, standardise and sustain. Data forensics, resolution and categorisation cover the first three. Governance and training in data management cover the last two.
What you get back is the improvement itself, not a report recommending one: validated data, ready to support decisions, and the means to keep it that way.
The value of data you can trust
The benefit is easy to overlook, because most of it is the absence of problems. Plans that hold. Compliance you can evidence. Budgets, forecasts and contracts based on real figures. No spreadsheet on the side.
You may already suspect your data is less reliable than it should be. Many organisations do, often rightly. The choice is whether to deal with it on your own timetable or when an auditor, board or regulator starts asking.
The opening question still stands. How many of your assets exist, and when was the data behind them last verified? Both answers are worth knowing before the next decision depends on them.