Every industry has developed an operating system for its most important work. Software engineers have GitHub, DevOps pipelines and automated testing. Finance teams have ERP systems. Sales organisations run on CRM platforms. Marketing teams orchestrate campaigns from a single workspace. Yet one of the most expensive, complex and business-critical activities an organisation can undertake—a major data project—still doesn't have an operating system. Instead, most data projects are held together with spreadsheets, emails, PowerPoint decks, documentation scattered across SharePoint, and countless meetings. Success depends less on technology and more on whether enough experienced people can keep thousands of moving parts aligned.

After twenty years leading data strategy and transformation programmes, I've come to believe that this is the fundamental reason why data projects so often overrun on cost, time and scope. The problem isn't that data is difficult. The problem is that we're trying to manage extraordinary complexity using tools that were never designed for it.

Consider a Typical ERP Implementation

Imagine a mid-sized manufacturing business replacing its legacy ERP platform after twenty years. On paper, the objective sounds straightforward. Deploy the new ERP, migrate the existing data and switch the business over. In reality, the ERP software is rarely the biggest challenge. The real complexity lies in the information that has accumulated over decades of operating the business. There may be millions of customer records, product catalogues, supplier information, contracts, invoices, purchase orders, inventory records, pricing tables and historical financial transactions. Much of this data exists across multiple systems, each designed independently over many years. Some customer names have been entered differently by different departments. Product codes have evolved over time. Fields have changed meaning. Entire databases exist that nobody has touched in years but suddenly become critical because they contain information required for regulatory or operational purposes. Before a single record can be migrated, someone has to understand all of it.

The Reality of Delivering the Project

Most executives imagine data migration as a technical exercise—copying data from one system into another. Anyone who has actually delivered these programmes knows that very little of the effort is spent moving data. Instead, enormous amounts of time are consumed answering questions. What does this field actually represent? Why doesn't this table exist in the new system? Which source should be treated as the master record? Why do two systems define an "active customer" differently? Can this value be transformed automatically, or does the business need to make a decision?

Every answer requires another conversation. Business analysts organise workshops. Data analysts compare schemas. Developers update transformation logic. Project managers revise plans. Documentation is rewritten. Test cases are rerun. New exceptions are discovered. Multiply this process across hundreds of tables and thousands of individual fields, and it becomes obvious why ERP implementations routinely involve dozens of specialists working together for months, sometimes years.

The Hidden Cost Isn't Technology

When organisations budget for data projects, they usually account for software licences, systems integrators and internal project teams. What they rarely measure is the cost of coordination. Waiting for subject matter experts to become available. Waiting for business decisions. Waiting for documentation to be updated. Waiting for someone to explain why two systems disagree. Waiting for approvals before testing can continue.

These delays don't happen in isolation. Each dependency creates another, until the project develops a critical path built almost entirely around communication rather than technology. By the time the ERP implementation finishes, the organisation often concludes that "data was harder than expected." In truth, the technology usually worked. The coordination didn't.

Software Engineering Solved This Problem Years Ago

It's interesting to compare this with software development. Twenty years ago, developers also relied heavily on manual processes. Code was emailed between colleagues. Different versions existed on different machines. Testing was inconsistent and deployments were stressful. Today, that sounds absurd.

Modern software engineering runs on an operating system consisting of version control, automated testing, continuous integration, deployment pipelines and collaboration platforms. Developers no longer spend most of their time coordinating work. The platform coordinates it for them. Data projects have never experienced that transformation. They're still managed largely through documents, meetings and manual processes.

What Would an Operating System for Data Projects Actually Do?

An operating system isn't simply another application. Its purpose is to coordinate everything happening around a complex piece of work. Applied to data projects, an operating system would continuously discover metadata across connected systems, understand relationships between schemas, recommend mappings, capture business decisions, generate documentation automatically and orchestrate testing from beginning to end. Instead of storing information across disconnected spreadsheets, documents and emails, every decision would exist within a single environment.

Every stakeholder would see the same information. Every dependency would be visible. Every change would automatically flow through the rest of the project. Rather than asking project managers to coordinate hundreds of independent activities, the platform itself would manage much of that complexity. Experts would spend their time making decisions instead of documenting them.

This Is Where AI Makes the Biggest Difference

Much of the conversation around AI focuses on making existing tasks slightly faster. Generate a document. Write some SQL. Summarise a meeting. Those improvements are useful, but they don't fundamentally change how projects are delivered. The bigger opportunity is to redesign the entire operating model. Instead of helping someone complete another spreadsheet, AI should remove the need for the spreadsheet altogether. Instead of asking analysts to compare thousands of columns manually, AI should perform the comparison automatically and only involve humans when judgement is required. Instead of updating documentation after the project changes, documentation should update itself. That's a much more ambitious use of AI—and one that has the potential to transform the economics of data delivery.

Why We Built DataSync

When we founded DataSync, we weren't trying to build another ETL platform or another data quality tool. There are already excellent products that solve those individual problems. We started with a different question.

What if data projects had an operating system?

A platform that could connect every stage of delivery—from discovery and metadata analysis through mapping, transformation, documentation, collaboration and testing—while using AI to automate the repetitive work that consumes so much of a project's time. If software engineering can deliver applications faster because it has an operating system, why shouldn't data teams have the same advantage?

Looking Ahead

The demand for data projects is only increasing. Every cloud migration, ERP replacement, merger, acquisition and AI initiative depends on moving, integrating or transforming data successfully. At the same time, experienced data professionals remain scarce, and organisations cannot continue solving the problem simply by hiring more consultants. The next generation of data delivery won't come from working harder. It will come from changing the way projects themselves operate.

I believe the organisations that embrace this shift will deliver projects faster, spend less, reduce risk and free their experts to focus on solving business problems rather than managing administration. In much the same way that operating systems transformed personal computing and DevOps transformed software engineering, the next major leap forward for data projects will come from giving them the operating system they've never had.‍

About DataSync

DataSync is building the operating system for data projects. By combining AI with intelligent workflow orchestration, it automates metadata discovery, schema mapping, documentation, testing and collaboration—helping organisations deliver integrations and migrations in days instead of months.

If you're planning an ERP implementation, cloud migration or enterprise integration programme, book a live demonstration to see how DataSync can fundamentally change the way your data projects are delivered.

‍