Insights · Finance and operations
Do You Need a Data Warehouse? A Guide for $20–50M Companies
How a mid-sized company can tell whether it needs a data warehouse or data pipelines, what a right-sized setup looks like, and what to fix first.
In short. Many $20–50M companies do not need an enterprise data platform. You likely need a small, right-sized warehouse when reports depend on hand-stitched exports from several systems, numbers disagree between reports, or history gets lost. If one system holds most of your data, better reports or a scheduled export may be enough.
The weekly numbers often start the same way: someone exports from the accounting system, the ERP and a spreadsheet or two, pastes them together, and fixes the lookups that broke. It works until that person is out, or two reports show different totals for the same thing.
Signs a warehouse would help
- Reports depend on manual exports from three or more systems.
- Numbers disagree. Sales in one report does not match sales in another, and nobody is sure which is right.
- History disappears when a spreadsheet is overwritten or a system is changed.
- Questions cross systems, like orders against AP, or jobs against payroll, and take days to answer.
- Excel is the database. Workbooks are slow, fragile and owned by one person.
Signs you do not need one yet
- One system holds most of the data you report on, and its built-in reports cover the need.
- The reports you need are few and stable, and a scheduled export would do.
- Nobody would own the definitions. A warehouse with no agreed meaning of "revenue" just moves the argument.
What a right-sized setup looks like
- Pipelines that pull data from your accounting, ERP, CRM and spreadsheets on a schedule, with checks for missing or odd data.
- One reporting database sized for your company, not for a large enterprise.
- Clean tables with agreed definitions, documented in plain English.
- Reports built on top, in Excel, your BI tool or automated weekly reports.
- Alerts when a load fails or numbers look wrong.
Most of the value comes from the agreed definitions and the reliable loads, not from the brand of database.
What to look at first
- List the reports leadership relies on and where each number comes from.
- Note which reports need manual exports, and how long they take each cycle.
- Find two reports that should agree and check whether they do.
- Decide who owns the definition of your key numbers.
- Check which of your systems can export or connect on a schedule.
If your reporting runs on exports and copy and paste, we are glad to look at whether a warehouse is worth it, or whether something smaller would do. Book a free 20-minute call. See also data engineering.