We already know what OneLake is and how the Shortcuts mechanism works. Today I want to go one level deeper - not into technical details, but into what actually hurts in everyday work with data in an organization. And what to do about that pain step by step.
Let’s start with a diagnosis you probably know from your own environment - the monthly ritual - end of the month.
Each department collects its data. Someone sends a file by email. Someone else makes changes and sends it back labeled “v2_final”. A week later, there are two versions in circulation and no one knows which one is current.
Consolidation takes several days, and still errors creep in - a copied formula, a deleted column, a renamed sheet.
The report for management is there - but everyone knows the numbers are “approximate”.
This is not an exception. It is the standard in organizations that have not yet developed good data practices. And there is no shame in that - simply no one has ever clearly explained how to do it differently.
Don’t fight Excel - change how you use it
Excel is not the problem. The problem lies in the operating model - in treating it as a database, when its real strength is being an interface to data.
Before you even think about implementing advanced platforms, it is worth doing a few simple things that immediately improve the situation.
1. Move files from local drives to SharePoint or OneDrive.
This single step eliminates versioning issues - every change is tracked automatically, you can revert to a previous version in a few clicks, and there is one link that is always up to date. No more “v2_final_corrected_final.xlsx”.
2. Separate editing zones.
Who should enter data and who should not touch formulas - this must be enforced in the file itself, not just in people’s heads. Excel has sheet protection for exactly this purpose. Use it.
3. Introduce templates.
If each department delivers data in a different format, with different column names and structure - no system will be able to combine it. A single shared template with defined headers is a prerequisite for any automation.
4. Use group permissions, not individual ones.
Instead of assigning access person by person, manage it through Microsoft 365 groups. When someone leaves the company, one change is enough instead of reviewing hundreds of files.
In many organizations, these four steps are enough to see improvement. But at scale - many departments, many data sources, recurring reporting - it is still not enough.
When Excel stops being enough - Microsoft Fabric and OneLake
The moment when Excel stops being a solution and becomes a problem is usually well known in the organization.
Reports are “finalized manually”. Consolidation takes more time than analysis. Files freeze with larger datasets. Management meetings focus on which numbers are correct instead of what to do with them.
This is where Microsoft Fabric and OneLake come in.
Not to take Excel away from users - that would not work and does not make sense. But to stop using Excel as a database and let it become what it does best - an interface.
Users continue entering data in structured spreadsheets, while Fabric consolidates everything into one consistent, real-time view. No crashes. No manual consolidation. No debates about which version is correct.
How to change your data culture - without going crazy
Transforming the way people work with data does not have to be a revolution. The most effective approach is incremental.
Step one - audit.
Before implementing anything, identify the files that actually drive decisions in your organization. Where are they? Who owns them? Move them to SharePoint and introduce basic governance and versioning. This is a quick win with immediate impact.
Step two - standardization.
Define a common vocabulary and templates for recurring documents. If sales calls a region “South” and finance calls it “Zone C” - no system will ever connect those datasets. This is organizational work, not technical - and that is why it is often ignored. Incorrectly.
Step three - integration.
Once data is structured and cataloged, start connecting those Excel files to OneLake using Shortcuts. Let the system take over data consolidation. Excel remains as an interface for input or final analysis. Everything else happens automatically.
Who needs to be at the table
Changing the data operating model requires three groups at the same time.
Business owners - CFO, Head of Sales, department leaders. Without their involvement, no change in data culture will go beyond a PowerPoint presentation. They need to say clearly: “we do not accept files sent via email anymore”. And they know best what data is needed for decision-making.
Data analysts or engineers. These are the people who technically connect Excel files in OneLake, build data models, and remove technical burdens from business users.
Business users. They benefit from the system - but they also need to change their habits. Without that, even the best platform will remain unused.
Let’s talk about costs honestly
Microsoft Fabric licensing is scalable - you can start small and expand as usage grows. Organizations already using Microsoft 365, Power BI Pro, or Office often already have most of the required licenses. The missing piece is usually Fabric Capacity - computational resources that can be scaled up or paused when not needed.
But let’s be honest about something else: the biggest cost is not licenses. It is people’s time.
They need to learn new habits and stop storing data on local drives. There may be a temporary drop in productivity - this is a natural part of any meaningful change. It is better to plan and communicate it upfront rather than be surprised.
Where to start - practically
The best starting point is one specific process that causes the most pain. In most organizations I work with, this is forecasting and budgeting. That is where the most time is lost on manual aggregation and where the benefits of a new approach are visible the fastest.
A pilot in this area usually takes a few to several weeks and provides a clear answer: how much time is saved, how many errors are eliminated, and whether it is worth scaling further. In practice, the answer is always “yes” - but it is important to validate it within your own organization, using your own data.
That is why I always recommend: start small, not with a company-wide transformation.
Good data is not an IT tool. It is a way of working.
