Business intelligence (BI)

What is business intelligence?

Business intelligence, usually shortened to BI, is the practice of turning business data into reports, dashboards, and analyses that support decisions. Revenue sits in accounting. Customers sit in CRM. Orders sit in ecommerce or ERP. BI brings those numbers together so people can see how the business is actually running.

BI is mostly about looking back and looking at now. What did we sell last quarter? Which customers are buying less than last year? Which invoices are more than sixty days overdue? The answers are usually already present in the systems, but scattered across several tools.

So BI is not just a piece of dashboard software. It is a way of organising data, agreeing definitions, and putting the same trusted numbers in front of the people who need them.

How a BI setup works

A BI solution usually has a chain of parts. A small company may start with only two or three. A larger setup separates them more clearly.

1. Source systems. ERP, CRM, webshop, accounting, support desk, spreadsheets, and operational databases.

2. Data pipelines. Jobs or connectors that extract data from those sources on a schedule or continuously.

3. Data warehouse or lakehouse. A central place where data is cleaned, joined, historised, and kept ready for reporting.

4. Semantic model. The layer that defines measures and business logic: revenue, margin, active customer, churn, open balance.

5. Reports and dashboards. The visible layer where users filter, drill, compare, and discuss the numbers.

You do not always need the full chain on day one. A first Power BI report connected to one source can be perfectly valid. The chain becomes important when there are more sources, more users, more history, and more disagreement about what the numbers mean.

Reporting versus analysis

Reporting answers the same questions in the same shape each week or month. Revenue by region, margin by product group, overdue invoices, support tickets by status. That rhythm is useful for management and operations.

Analysis starts when the fixed report raises a new question. Revenue in the west region is down. Why? You drill from region to customer, from customer to product, from product to order. A good BI model lets a business user follow that trail without waiting for a new export.

This is where self-service analytics comes in. It only works when the shared model is trustworthy. If every department defines revenue differently, self-service becomes self-service confusion.

BI versus Excel

Excel is still useful. A quick calculation, one-off list, ad hoc simulation, or small personal analysis may be faster in a spreadsheet than in a BI platform.

Excel starts to creak when the same numbers must be shared and governed. Everyone has a different copy. Formulas drift. Refreshes happen by hand. Sensitive data spreads through email. Files become too large. The word revenue means one thing in the finance sheet and another in the sales sheet.

A BI solution solves those problems with shared definitions, controlled access, automatic refresh, larger data volumes, and one model that multiple reports can reuse.

BI versus predictive analytics

BI usually describes what happened and what is happening. Predictive analytics estimates what is likely to happen next: which customer may churn, how much stock you need next month, which invoice may be paid late.

The two belong together. Predictive models need clean historical data to train on. If the past is messy, the forecast will be messy too. BI often builds the foundation that predictive analytics later uses.

What BI gives an SME

Faster answers
Monthly numbers no longer depend on two days of copying exports into spreadsheets. The dashboard is ready when the meeting starts.

One version of the truth
Sales, finance, and operations discuss the same margin and revenue definitions. The conversation moves from whose number is right? to what do we do?

Less manual work
Repeated exports, copy-paste work, and fragile formulas disappear from the monthly routine.

Better follow-up
A KPI only helps if someone sees it in time and owns the next step. BI makes that practical, but ownership still has to be agreed.

How to start with BI

Start with questions, not tools. Pick three questions that matter in the business right now: which customers bought less this year, where is margin leaking, which invoices are overdue, which products are returned too often.

Those questions decide which sources you connect, which definitions you need, and who should use the report. A first dashboard that answers three real questions will be used. A dashboard with forty charts and no decision behind it will not.

The tool choice comes after that. In a Microsoft-oriented company, Power BI is often the natural starting point. Tableau, Qlik, Looker, and other BI tools solve similar problems. The deciding factor is rarely the charting tool itself. It is whether the data is connected properly and whether someone owns the definitions.

What to watch out for with BI

Dashboards without a decision
If a report is not used in a meeting, process, or recurring decision, it will fade away. Tie every dashboard to a business rhythm.

KPIs without owners
A red number that belongs to nobody is decoration. Agree who watches each KPI and what happens when it changes.

Tool-first thinking
Buying licences does not clean customer data, define margin, or fix duplicate products. BI work is mostly data work.

Too much at once
A first BI project should prove value quickly. Start narrow, learn from usage, and expand the model from there.