The Cloud Migration Nobody Owns

Many companies sign the Snowflake, Databricks, or Microsoft Fabric contract. The migration plan is sound and the team is capable. A year later, the AI pilot is stuck, finance still doesn't trust the dashboard, and the cloud bill keeps climbing. If that sounds familiar, you are not alone. Redgate's 2026 State of the Database Landscape report found that 91% of teams, no matter how experienced, hit at least one significant challenge during their move to the cloud.

I led cloud migrations at large enterprises. What I learned is that the technology is rarely why a migration stalls. It stalls because the work falls between roles. IT owns the platform and business leaders own the decisions, but the data in the middle often has no owner at all. That is a structural gap, not a failure of the people doing the work.

The Platform Isn't the Problem. The Data Foundation Is.

In many mid-market companies, operational data is scattered across CRMs, ERPs, marketing tools, application databases, and spreadsheets that no single department maintains. Lift that estate into the cloud as it stands, and you get the same problems on more expensive infrastructure. A sleek dashboard can't fix inconsistent schemas, missing transactions, or duplicate customer records, and neither can an AI model. Building analytics or AI on a broken foundation doesn't solve the problem. It just shows you the problem faster.

The research backs this up. Gartner predicts that through 2026, organizations will abandon 60% of AI projects that aren't supported by AI-ready data, and 63% of organizations in the same research said they either lack the right data management practices for AI or aren't sure they have them. Meanwhile, Flexera's 2026 State of the Cloud report puts wasted cloud spend at 29%, the first increase in five years, as AI workloads surge. Even the best cloud platform can only be as reliable as the data it holds.

The Customers Who Disappeared Overnight

Picture the Monday after cutover. The old report showed 22,000 active customers. The new cloud dashboard shows 18,500. Nobody lost 3,500 customers. Years ago, someone added a rule to a legacy stored procedure: count a customer as active if they bought anything in the last 13 months. Nobody wrote it down. The migration moved the data perfectly. It didn't move the meaning.

QA flags the mismatch, and leadership won't sign off on a cutover they don't have confidence in. So the old system stays on “for one more month.” Nobody can say which number is right, because nobody owns the definition. Nine months later, the company is paying for both platforms while the team reconciles by hand. Sound familiar?

Neither system is broken. They just disagree on what counts as an active customer, and that is a business definition, not an engineering bug. Most migrations are scoped as technical projects, which is a reasonable place to start. But when the scope stops there, those hidden rules either get lost or travel into the cloud unexamined. That is why cloud migration needs bilingual leadership: someone fluent in both technology and business.

The Leadership Gap Most Migrations Run Into

Most leadership teams recognize they need someone in this seat. Then they meet the unicorn problem: an executive with deep data engineering skill and real C-suite strategy experience is rare, and a search for one can take many months. Meanwhile, capable teams keep the lights on while the migration loses momentum.

That drift has a cost. Cloud spend creeps up as unmanaged queries, idle resources, and the wrong storage tiers add up quietly; in Flexera's 2026 survey, 85% of organizations named managing cloud spend as a top challenge. Analysts get pulled into data prep. A 2018 IDC study for Alteryx found that data workers lost about 14 hours a week because they couldn't find, protect, or prepare the data they needed.

Data quality decays silently, too. Duplicates, missing fields, and broken feeds go unnoticed until they skew a model or a board report, and Gartner has estimated that poor data quality costs organizations an average of $12.9 million a year. And many migrations simply stall halfway. Redgate's 2026 research found hybrid cloud has become the long-term reality for 43% of organizations, where the easy workloads move and the complex, high-risk databases get stuck in between.

The fix isn't always a new full-time hire. What matters is that someone with both the engineering depth and the business view owns the data from day one, whether that's an internal leader or a fractional one.

How to Fix It: Two Tracks, Three Stages

The answer isn't to stop everything and spend two years cleaning data. Leadership needs visible wins now, and the business needs a foundation that lasts. So the work runs on two tracks at the same time.

Track 1 earns trust early, and Track 2 makes later wins last. The tools change from client to client. Snowflake, Databricks, and Microsoft Fabric can all work. The sequence is what makes the difference.

Five Signs Your Migration Could Use Dedicated Data Leadership

If two or more of these sound familiar, you are in good company. None of them means your team has fallen short. They are signs the role is missing.

1. Your analysts spend more time preparing data than analyzing it. Hours every week go to cleaning CSVs and fixing broken feeds.
2. Department dashboards disagree on core numbers. Sales, finance, and marketing bring different revenue or churn figures to the same meeting.
3. Key data is trapped in silos. Customer, financial, and product data live in separate systems with no automated integration.
4. Critical decisions run on spreadsheets without a clear owner. Board-level numbers depend on files with no documentation or governance.
5. Your cloud or AI project has stalled. You are paying for Snowflake, Databricks, or Fabric, but the AI use cases never make it to production.

The Key Takeaway

Cloud platforms deliver on their promise when someone leads the data that runs through them: its definitions, its quality, and its connection to business goals. That leader has to be bilingual, fluent in both technology and business. A technologist alone can move the data, but may not know which revenue number the CFO trusts. A business leader alone can set the goals, but may not see why the pipelines keep breaking or the cloud bill keeps climbing.

The migration needs someone who can sit in the architecture review in the morning and the executive meeting in the afternoon, and translate between the two. That role is easy to overlook because it sits between IT and the business, and filling it is often what turns a stalled migration around.

So who is leading your cloud data migration today: a technologist, a business leader, or someone who is bilingual?