diginomica enterprise data health research - the data is broken and everybody knows it

Earlier this week, diginomica and Maureen Blandford of Serendipitus published the Enterprise Data Health Study – independent qualitative research based on 18 frank conversations with senior enterprise practitioners who spoke under total anonymity. We invited our partners to respond on the record. 

What follows are responses in full, verbatim and unedited, alongside some observations about what the pattern of engagement reveals. All contributors are diginomica partners; readers should weigh that context accordingly. What's important is not just what they said, but how closely it maps to what practitioners told us in private – and in one case, how much further it goes.

A tax that compounds

The finding that landed hardest with respondents was what the report calls the verification tax – the invisible organizational cost of rebuilding, reconciling and rechecking data before anyone will stake their reputation on it. Several respondents didn't just validate it, but extended it.

Raju Malhotra, CPTO at Certinia, puts it most precisely:

What this research captures is an operational cost most organizations have quietly learned to absorb. When practitioners spend up to 70% of their time manually assembling and checking data, that capacity is gone before any real analysis begins. The report rightly identifies this pattern of manual heroics as a 'verification tax' that surfaces as week-long board prep cycles, parallel spreadsheets, and multiple revision rounds. The deeper and more sinister issue is that this tax compounds over time. Decisive action on data requires organizational confidence, and confidence erodes in environments where every number needs rebuilding before it can be trusted. Getting out of that cycle means treating data structure as a governance question from the beginning, with clear ownership and documented provenance, so that verification becomes the exception rather than the default.

Rupal Karia, SVP North America, UKI & MEA at Celonis, picks up the same thread:

The study brings out some hard truths about how companies are really running today. When teams spend up to 70% of their time just pulling and checking data, something is broken. That 'verification tax' is more than just an inefficiency, it's quietly draining the time and talent that should be focused on higher-value work.

This also explains why so many AI initiatives struggle to move past the pilot phase. There is plenty of intent, but very little is actually in production. Without a clear, trusted view of how the business runs, AI is working off incomplete information. The reality is there is no AI without Process Intelligence. You need the business context and view of interactions between systems that give AI what it needs to actually work in the enterprise market.

The good news is that this is possible. Process Intelligence creates a shared version of the truth that connects people, systems, and AI. By getting that foundation right, organizations can stop the manual guesswork, build confidence in their data, and finally see a real return on their AI investments.

Kathy Pham, VP of Open Technology and AI at Workday, brings something the other responses don't – a personal perspective that shifts the 'manual heroics' finding — away from dysfunction and toward something more recognizable as dedication:

The findings in this report offer a refreshing reality check on what [diginomica] describes as a 'trust collapse' in enterprise data. At Workday, we start from the premise that without trust in the underlying data, you can't have trust in the AI that depends on it. Early in my career building data warehouses for hospital systems, I saw that data fragmentation wasn't a sign of institutional neglect, but rather a byproduct of clinical dedication. When practitioners create 'shadow data sets' or implement disparate systems, they do so to fill immediate gaps in patient care that rigid legacy infrastructure cannot address. This 'manual heroics' is a testament to their commitment, yet it inadvertently creates the 'verification tax' this study so accurately identifies, where 30–70% of professional time is lost to assembly rather than analysis. To move into the era of agentic AI, we must honor that dedication by providing a foundation that carries the necessary organizational context to make their data and their decisions truly useful.

As the report highlights, the hardest challenges aren't technical, they're organizational. Tools alone cannot fix a lack of trust between functions, especially when 94% of organizations still struggle with siloed control. At Workday, we partner with customers on these prerequisites by emphasizing transparency and partnership across the enterprise. This includes pilot programs to prove value in a contained environment and a risk management framework led by our Responsible AI Team to address the specific nuances of different data types. By pairing AI outputs with clear confidence signals and transparency tools like AI fact sheets, we move closer to what this report ultimately calls for: helping people make better, more confident decisions without the grind of making broken systems work.

Not a technology failure

Several respondents were direct about where they locate the root cause – and it wasn't the technology. 

Cathy Mauzaize, EMEA President of ServiceNow:

The challenges I hear about from leaders across EMEA every day is reflected in this research: complexity has accumulated over time - layering system upon system until data is fragmented, ownership is unclear, and trust breaks down. Teams compensate by rebuilding and verifying everything manually. As the research shows, in some organizations up to 70% of time is spent on manual data work. The cost is a massive drain on productivity and slower decisions at every level.

This has direct implications for AI. The lack of trust in cross-functional data isn't a technology failure, it's an organizational one – and it limits AI readiness. At ServiceNow, we see that scaling AI responsibly depends on connecting data, governance, and workflows as one foundation. Without that ability to understand and act, AI doesn't solve the problem – it's just expensive, and often unreliable, advice.

Jim Desler, VP of Communications at Acumatica, calls it a vendor responsibility:

This research highlights something we constantly hear from our Acumatica Community: AI is only as powerful as the data behind it. If the data isn't accurate, connected, and reliable, even the most advanced tools won't deliver meaningful results. It also reinforces the importance of trust. Real collaboration across teams depends on confidence in the data. Without it, people resort to manual workarounds and silos, which slow decision-making.

For technology vendors, the results of this study are a clear reminder of our role and obligation to foster trust and confidence in data by connecting systems, improving visibility, and supporting the processes that make data accurate and shareable. That's what ultimately enables organizations to get real value from AI.

Data silos aren't always accidental

One response goes somewhere the research itself didn't quite reach. Rowan Tonkin, CMO at Planful, names something that practitioners couldn't say publicly and vendors rarely acknowledge: that silos are sometimes maintained deliberately, because broken data serves certain interests.

The report points to a clear relationship between the overwhelming presence of data silos (94% of organizations suffer from them) and the lack of trust between departments. The leader who said, 'I wouldn't trust a single thing that came from another part of the business without really scrutinizing it,' summed it up perfectly.

I've written before about how data silos are the number one reason organizations fail to meet their overall business plans. When each team builds its own models, KPIs, and incentive programs, the data these systems produce becomes the only data the team trusts – even if it doesn't line up with anyone else's. Some functions even try to protect their silos because they can hide inefficiency and risk and offer negotiating leverage come budget season. If you own the data, you control the narrative. Applying AI across planning can expose the gaps between systems, but C-level execs who want real data visibility and collaboration need to align metrics and incentives across the business to cut out data silos at the root.

"If you own the data, you control the narrative" points to a dimension of the problem that wasn't fully surfaced in the original qualitative research. The practitioners we interviewed described silos as a structural frustration. Tonkin suggests that for some teams, the silo is a strategic asset. Fixing it means redistributing power, and that's a tough conversation for organizations to have.

What the data strategy response tells us

Fred Lherault, Field CTO for EMEA and Emerging Markets at Everpure, goes furthest into the technical detail – and like several others, gets to the organizational reality before arriving at the product answer:

This report illustrates that businesses aren't suffering from a lack of tools, but a lack of trust in their data. When teams are spending their time manually reconciling spreadsheets and re-building reports by hand, you don't have a technology issue, you have a systemic trust deficit in the information that's supposed to power the business forward.

You can't fix this with another point solution or dashboard. You must start with a holistic data management strategy underpinned by an Enterprise Data Cloud – a single, intelligent data plane that continuously discovers data wherever it lives, classifies it in a business context, and enforces governance and sovereignty policies automatically. This has to go hand-in-hand with organizational and behavioural change. There needs to be strong communication and trust between departments, a willingness to collaborate on a single source of truth, and alignment on agreed data practices. That's how you move on from doubt, and the safe and fast application of AI across your data estate.

Given that Gartner finds 63% of organizations lack appropriate data management for AI, and predicts up to 60% of AI projects will be abandoned without AI-ready data, it's safe to say that until organizations treat data as a managed asset rather than a by-product of applications and teams, living in silos, AI will simply add further layers of complexity, and amplify any bad data. Those who succeed in this next phase will be the ones who invest in trusted data foundations first, and AI second - leveraging an Enterprise Data Cloud to restore confidence in data before they ask machines to make decisions on their behalf.

My take

What these responses do collectively is confirm, on the record and with their names attached, what 18 practitioners told us in private. The verification tax is real. The trust collapse is real. The AI readiness gap is real. That confirmation has value – the market knows what the problem is.

What's harder is saying it plainly. Enterprise technology runs on a kind of performed confidence – the assumption that admitting broken data, failed AI initiatives, or zero cross-functional trust is a career risk rather than a starting point. The practitioners who spoke to us said what they said because no one would know it was them. The vendors who responded here said what they said carefully, with legal review and comms approval, because their names are attached. Both things are understandable. Neither produces the honest conversation that buyers actually need.

Planful came closest to naming what usually goes unsaid: that some of this isn't dysfunction, it's design – data as leverage, silos as protection, the incentive to maintain uncertainty because certainty redistributes power. Most technology roadmaps aren't built to address that, because addressing it requires a willingness to have a different kind of conversation entirely.

The research market knows this is happening. The people building and selling enterprise technology know this is happening. What's largely missing is someone willing to stand on the buyer's side and say so – not to burn the market, but to make it more honest. That's what this research is for. It's the beginning of that work, not the end of it.

The Enterprise Data Health Study by diginomica and Serendipitus is free to download.