My name is Charles Mulwa, and I’ve spent more than 20 years working in data leadership – as a database administrator, a data strategist, and, more often than I’d like, as the person called in after something has already gone wrong. In that time, I’ve sat in boardrooms where seven-figure data initiatives got approved on a hunch, and I’ve sat in server rooms at 2 a.m. helping a panicked IT director recover data nobody realized wasn’t being backed up correctly.
The tools have changed enormously since I started my career. Cloud data warehouses are replacing on-premise servers. Dashboards are replacing static reports. Now AI is replacing dashboards. But the mistakes I see companies make? Those have barely changed at all. I could have written this same list in 2010, in 2015, and in 2020 – the technology names would be different, but the underlying failures would be identical.
What follows are the ten mistakes I encounter most often, drawn from two decades of consulting, hands-on database work, and more than a few uncomfortable conversations with executives who wanted a quick fix for a problem that had been building for years.
1. Treating Data as an IT Problem, Not a Business Asset
Early in my career, I worked with a mid-sized distribution company where the entire data function reported three layers down from the CEO, buried inside IT alongside help desk tickets and printer maintenance. Nobody outside the department could tell you who “owned” customer data, so when a sales VP wanted a straightforward report on regional performance, it took six weeks and four escalations to get an answer.
That’s the pattern. The moment leadership hands data ownership entirely to IT and walks away, the company has already lost. Data isn’t infrastructure to be maintained quietly in the background – it’s a business asset that needs business owners, business context, and business accountability. When the CFO doesn’t know who owns revenue data, nobody actually owns it, and decisions slow to a crawl.
2. Buying Tools Before Defining the Problem
I once sat through a vendor demo with a retail client who had just signed a contract for an enterprise BI platform – six figures a year – before anyone on their team could articulate what business question it was meant to answer. Eighteen months later, adoption was near zero, and I was brought in to figure out why a tool that looked impressive in the sales pitch had become shelfware.
This happens constantly. Companies buy the shiny platform, sold on dashboards and AI features, then spend the next year trying to retrofit a use case around it. The tool should always follow the question. When it doesn’t, you end up with expensive software and no better decisions than before you bought it.
3. Confusing Data Collection with Data Strategy
A healthcare logistics company I worked with was proudly “capturing everything” – every click, every transaction, every sensor reading from their fleet. Their data warehouse was enormous. And when I asked their leadership team a simple question — which customer segments were most likely to churn – nobody could answer it. All that volume, and not a single actionable insight to show for it.
Collecting more data is not a strategy. It’s just expensive storage dressed up as progress. A real data strategy starts with the decisions the business needs to make, then works backward to determine what data is actually required.
4. Ignoring Data Quality Until It’s a Crisis
I’ve walked into more than one board meeting that fell apart because two departments presented different revenue numbers for the same quarter — each one convinced their spreadsheet was correct. In one case, the discrepancy came down to a single unresolved duplicate-customer issue that had been flagged by an analyst eight months earlier and never prioritized.
Nobody wants to fund the unglamorous work of cleaning, validating, and governing data — until a moment like that forces the issue. By then, the fix costs far more than prevention would have, and the damage to leadership’s confidence in the data team is often harder to repair than the data itself.
5. Letting Every Department Build Its Own Version of the Truth
Marketing has its numbers. Sales has its numbers. Finance has its numbers. I’ve watched entire quarterly meetings devolve into arguments about whose figures were “right” instead of what actions the business should take. At one manufacturing client, I counted five different definitions of “active customer” floating around different departments – each technically defensible, none of them reconciled.
Without a single source of truth, and the governance to enforce it, companies spend more energy debating the data than acting on it. This is one of the most common – and most fixable – problems I encounter.
6. Underinvesting in the People, Overinvesting in the Platform
I’ve seen companies spend millions on enterprise data platforms and staff them with a single overworked analyst expected to be a database administrator, a report builder, a data scientist, and a governance officer all at once. Predictably, that analyst burns out or leaves within a year, taking institutional knowledge with them.
Technology doesn’t interpret itself. Without skilled people who understand both the data and the business context around it, even the best platform becomes an expensive filing cabinet nobody knows how to use properly.
7. Skipping Governance Because It Feels Like Bureaucracy
Governance has an image problem. I understand why – I’ve seen it introduced badly plenty of times, rolled out as a rigid compliance checklist rather than a framework for making data more trustworthy and usable. One financial services client resisted governance for years, calling it “red tape,” until a data breach forced their hand and they had to build the entire framework under regulatory pressure and public scrutiny simultaneously.
Companies that skip governance early always end up building it eventually – just later, under worse conditions, usually after an incident that could have been prevented.
8. Rushing Into AI Without Fixing the Data Foundation First
This is the mistake I’m seeing most right now, and it worries me more than any of the others on this list. I recently consulted with a company eager to deploy an AI-driven forecasting tool, only to discover their historical sales data had inconsistent product categorization going back years – the kind of inconsistency a human analyst could mentally work around, but that quietly poisons any model trained on it.
Leadership wants the AI initiative, wants it fast, and doesn’t want to hear that the underlying data isn’t ready. But AI built on inconsistent, siloed, or untrustworthy data doesn’t produce insight – it produces confident-sounding nonsense at scale, and it does so persuasively enough that people believe it. The foundational data work isn’t optional. It’s the whole game, and skipping it is the single fastest way to turn an AI initiative into an expensive cautionary tale.
9. Measuring Activity Instead of Impact
I ask nearly every client the same question early in an engagement: “What business decision did this data change last quarter?” The silence that sometimes follows tells me everything I need to know. Dashboards get built. Reports get generated. Meetings get held. None of it matters if no one can point to a decision that changed because of it.
At one company, I found a team producing 40 automated reports a week that, by their own admission, almost nobody opened. We cut it down to six reports tied directly to decisions leadership actually made, and engagement – and trust in the data team – improved almost immediately.
10. Treating Data Leadership as a Technical Role Instead of a Strategic One
The best data leaders I’ve worked alongside in my career aren’t just technically fluent – they’re translators. They can sit with a CEO and talk strategy, then sit with a database engineer an hour later and talk architecture, and let each conversation inform the other. I’ve built my own approach to data leadership around that principle, because I’ve seen what happens when companies hire purely for technical skill and expect strategic outcomes: a brilliant engineer gets promoted into a leadership vacuum, and the business side of the conversation quietly disappears.
Companies that separate the “technical” and “strategic” parts of data leadership into different people – without ensuring those people talk constantly – almost always end up with technically sound systems that don’t actually serve the business.
The Pattern Behind the Pattern
If there’s a single thread connecting all ten of these mistakes, it’s this: companies keep treating data as a technical afterthought instead of a core business discipline. I’ve seen this play out across industries – distribution, healthcare, manufacturing, financial services – and the pattern never really changes, only the specifics.
The fix isn’t more tools, and it isn’t a bigger budget thrown at the newest platform. It’s leadership willing to slow down, define the actual business problem, fund the unglamorous foundational work that never makes it into a press release, and hold the organization accountable to a single, agreed-upon version of the truth.
Twenty-plus years into this work, that’s still the hardest part of the job for the companies I advise – and still the part that separates the organizations that genuinely use their data from the ones that just collect it and hope.
I’m Charles Mulwa, and I’ve built my career helping companies close that gap. If any of these ten mistakes sound uncomfortably familiar, you’re far from alone – and recognizing the pattern is always the first step toward fixing it.