After a recent piece on AI fatigue, several conversations surfaced a theme most organisations are still avoiding: Making a business case for data before the cost gets out of hand.
That cost is already accumulating. It shows up in the team spending half the week reconciling figures that shouldn’t disagree, in the board report everyone treats as directionally accurate but nobody quite trusts, and in the quiet workarounds that have become so embedded in daily operations that people have stopped questioning them. The problem is not that organisations lack data – it is that the data they hold is fragmented, inconsistent, and in many cases physically inaccessible.
The price of poor data is no longer theoretical
The financial toll of bad data has been quantified repeatedly, and the numbers are difficult to dismiss. A 2025 report from the IBM Institute for Business Value found that over a quarter of organisations estimate they lose more than $5 million annually due to poor data quality, with seven per cent reporting losses of $25 million or more[1]. Gartner’s research paints a similar picture, estimating that poor data quality costs organisations an average of $12.9 million per year[2]. These are operational drags that touch every department, from finance and compliance to customer service and strategic planning.
What makes these figures particularly alarming is how they compound in an era of increasing AI adoption. AI organises inconsistency, at volume, with confidence. Feed a large language model or predictive analytics engine with contradictory customer records, inconsistent product classifications, or incomplete regulatory data, and the output will look polished but be fundamentally unreliable. A Dimensional Research survey found that 96 per cent of organisations encounter data quality problems when training AI models[3]. As AI investment scales, the cost of unresolved data problems scales with it – and the damage becomes harder to trace back to its source.
For organisations holding vast archives of paper records, legacy files, and unsorted digital content, the risk is even more acute. Dajon Data Management works with businesses across regulated sectors to digitise, classify, and migrate these holdings into structured, accessible formats; transforming dormant records into data that can actually support AI readiness and informed decision-making.
Why governance stalls when IT owns it
The instinct in most organisations is to treat data governance as an IT problem: Delivered by the data team, reporting to a steering committee nobody is excited about, and measured by technical metrics that rarely connect to commercial outcomes. This is largely why these programmes stall.
IT-led governance does not change how sales define a customer. It does not resolve which system wins when two sources disagree. It does not establish the retention rules that determine whether records should be digitised, archived, or securely destroyed. Only business ownership does – and it only materialises when leaders understand what they are losing in real terms.
Research from Info-Tech suggests that up to 75 per cent of governance initiatives fail because ownership is unclear[4]. Grant Ostler of Workiva puts it plainly: When data is scattered across disparate systems and manual processes operated in silos by different departments, the ability to make informed decisions becomes challenging at best – and applying new technology on top of poor-quality data can only create an efficient path to inaccuracy[5].
The business case must come first. That means a credible number on the problem – quantified in hours lost, decisions delayed, and regulatory exposure – followed by clear improvement metrics and a governance framework that makes change stick. Without that foundation, technology investments will continue to underperform, and AI programmes will be built on data nobody fully trusts.
This is where professional data management support can be transformative. Dajon helps organisations establish the practical groundwork for governance: Cataloguing what exists across physical and digital estates, identifying duplication and inconsistency, and creating the structured data foundations on which sustainable governance depends.
The GRC gap that AI is about to widen
The deeper issue is how Governance, Risk and Compliance (GRC) is failing to keep pace with AI. A 2026 benchmark study from Cisco found that 93 per cent of organisations are planning further investment in AI governance to keep up with the complexity of AI systems and customer expectations[6]. Meanwhile, Gartner predicts that by 2026, 50 per cent of large enterprises will have formal AI risk management programmes in place – up from less than 10 per cent in 2023[7].
But you cannot govern what you cannot see. Organisations talk enthusiastically about their shiny new systems for managing data – be it HRIS, ERP, CRM, ECM, or SharePoint – and there is no shortage of ambition around AI roadmaps. Yet one of the primary sources of the GRC gap is far more prosaic: The data itself. Much of it is incomplete, poorly classified, or locked away in formats that no modern system can reach.
Dajon’s data migration and data cleansing services are designed precisely for this challenge, helping organisations reconcile conflicting records, deduplicate across platforms, and ensure that the data feeding new systems is accurate, complete, and fit for purpose – whether the destination is a cloud-based ERP, a regulatory reporting platform, or an AI-driven analytics environment.
The contradiction hiding in plain sight
While debating AI implementation strategies, it comes as a surprise to many organisations that a fundamentally legacy approach is still actively in play. Many are simultaneously signing contracts to store thousands of archive boxes in off-site warehouses, year after year, without scrutiny.
This is particularly entrenched in the public sector. The General Medical Council recently awarded a storage contract covering approximately 34,000 cubic feet of boxed paper records at £313,000. The Health and Safety Executive opted to store 120,000 paper records and 1,800 tapes off-site for seven years at £511,000. Both contracts are publicly listed, entirely routine, and passed through procurement without anyone flagging them as a contradiction to the digital strategies being discussed down the corridor.
These are not isolated examples. Across the public and private sectors alike, organisations continue to pay recurring fees to warehouse records that are inaccessible, unsearchable, and contribute nothing to data quality or the AI-readiness agenda. The storage bill quietly compounds while the data locked inside those boxes remains invisible to the systems and strategies that need it most.
This is one of the areas where Dajon’s secure document digitisation services deliver the most immediate and measurable impact. By converting physical archives into searchable, indexed digital formats – with full chain-of-custody controls and compliance with sector-specific regulations – organisations can eliminate ongoing storage costs, bring dormant records into their data ecosystem, and close one of the most overlooked gaps in their digital transformation.
File servers: The digital equivalent of archive boxes
It is not only paper records that create this problem. File servers and legacy network drives present an almost identical challenge in digital form, preserving decades of digital sediment in its entirety. Shared drives accumulate years of duplicated documents, outdated versions, and unclassified files that no one has reviewed or rationalised.
The parallel with physical storage is striking. Just as archive boxes sit in warehouses untouched for years, file servers quietly consume infrastructure budget while the data they hold remains unsorted, ungoverned, and largely invisible to the organisation’s data strategy. The difference is that file servers are often perceived as cost-free. In reality they carry significant costs in storage infrastructure, backup overhead, security risk, and the opportunity cost of inaccessible information.
Data scientists and analysts already spend between 45 and 60 per cent of their time on data preparation tasks rather than the analysis and modelling work they were hired to do[8]. Poorly managed file estates only compound this problem, forcing teams to sift through disorganised repositories before they can even begin meaningful work.
Dajon’s data management services extend to these digital holdings as well, helping organisations audit, classify, and rationalise unstructured file estates: identifying what should be retained, what should be migrated into governed systems, and what can be securely disposed of in line with retention policies and regulatory requirements.
The window is narrower than it looks
Every year a data strategy is deferred is another year of underdelivering on AI, another year of people working around outputs they do not trust, and a growing stack of boxes (both physical and digital) that nobody has the mandate to touch.
The regulatory environment is tightening rapidly. The EU AI Act is phasing in substantive obligations through 2026, with full enforcement for high-risk AI systems in critical sectors approaching[9]. In the UK, the Data Use and Access Act received Royal Assent in June 2025, introducing targeted updates to the UK GDPR with phased implementation expected throughout 2026[10]. Organisations that have not addressed their underlying data quality, classification, and accessibility will find compliance increasingly difficult – and the costs of remediation will only grow.
What is missing is not appetite. It keeps showing up in strategy sessions, in board papers, and in the AI readiness assessments that consultants produce with admirable frequency. What is missing is the willingness to look at the whole picture – the physical archives, the digital sediment, the governance gaps, and the compounding cost of inaction – and to act on it before the window narrows further.
Dajon Data Management partners with organisations across financial services, insurance, pensions, legal, construction, and the public sector to tackle this challenge end to end: From digitising physical archives and cleansing legacy data, through to managing complex data migrations and establishing the governance frameworks that make lasting change possible. The starting point is always the same: understanding what you have, where it sits, and what it is costing you.
References
- The True Cost of Poor Data Quality IBM[↩]
- Data Quality: Why It Matters and How to Achieve It Gartner[↩]
- AI Data Quality in 2026: Challenges & Best Practices AIMultiple[↩]
- Turn Data Quality Risks Into Revenue with ADM Acceldata[↩]
- How AI will redefine compliance, risk and governance in 2026 Governance Intelligence[↩]
- The struggle for good AI governance is real CIO[↩]
- Data Governance for AI in 2025 Quinnox[↩]
- Your Data Scientists Were Hired to Build Models Amperity[↩]
- AI Risk & Compliance 2026 Secure Privacy[↩]
- Data protection & AI governance 2025-2026 The DPO Centre[↩]