The default analytics project starts by building a warehouse and moving everything into it. It is not the only way, and often not the fastest route to value.

Query where your data lives: why moving your data is not always the answer

Most analytics projects open the same way. Choose a platform, then spend the first few months building a central warehouse and moving everything into it before anyone sees a useful report. There are good reasons to do that, and we do plenty of it. But it has quietly become the reflex, and the reflex is not always right.

For more on this, see our data strategy and leadership work. It is also worth reading alongside how East of England Co-op adopted Pyramid if the topic is new to you.

What querying in place actually means

The alternative is to read the data where it already sits, across your finance system, your CRM, a cloud warehouse and a pile of spreadsheets, and lay a governed model over the top. You can be analysing across the whole estate in weeks, without a consolidation project as the price of entry.

That is the model platforms like Pyramid Analytics are built around. It suits a messy, multi-system estate, which is what most mid-market businesses actually have, and it gets you to value before the budget for a full warehouse has even been signed off.

When you should still build the warehouse

This is not an argument against warehouses. If you have very large volumes, heavy transformation, or a need to keep history that the source systems will not hold, a proper warehouse earns its place, and Microsoft Fabric is an excellent home for it. We cover that call in Do You Actually Need Fabric?.

The honest position is simple. Querying in place gets you to value faster and suits a scattered estate. A warehouse suits scale and serious engineering. The sensible path is often to start with the first and add the second when it is justified, rather than paying for the second before you have proven the first.

Next step

Before you commit to a months-long data move, ask one question: what decision are we waiting on, and could we answer it against the systems we already have? If the answer is yes, query in place first and consolidate later, on its own merits.

If any of this sounds familiar, talk to us about your data.

Related reading

• Pyramid, Power BI or Fabric: Which One, and When
• If you run Pyramid today, here is what to do now
• ServiceNow Has Bought Pyramid Analytics: What It Means