"Which BI tool is best?" is the wrong question, and the reason so many comparisons end in a feature table that helps nobody. Every serious BI tool can draw a bar chart, connect to a database and refresh overnight. What differs is the purpose each one was designed around, and therefore the kind of business it suits. Over this series we looked at Power BI, Qlik, and Tableau, Data Studio and Metabase. This is the short version: how to match one to your situation.

Choose on five things, in this order: the ecosystem you already live in, how many people build versus read, the shape of your data, the skills you can keep, and how the licence grows with you. Features come last, because they rarely decide it.

Each tool, in a sentence

  • Power BI: analysis for Microsoft businesses, inexpensive for a small group of users, with a formula language worth learning properly.
  • Qlik: free exploration across lots of related operational data, with an associative engine that shows what is missing as well as what matches, and capacity-based pricing.
  • Tableau: visual analysis and presentation for skilled analysts, strongest in Salesforce businesses.
  • Google Data Studio: free, live reporting on Google's own data, ideal for marketing and web.
  • Metabase: lean, open-source BI for a business whose data already lives in its own database, with someone to run it.

The five questions

1. What ecosystem do you already live in?

This settles more decisions than anything else. A Microsoft 365 business gets sign-in, security, Teams and Excel integration with Power BI for free. A Salesforce business gets the same with Tableau. A Google Workspace business with mostly marketing data gets Data Studio at no cost. Fighting your ecosystem is possible, but it should be for a good reason.

2. How many people build, and how many only read?

Per-user tools, Power BI Pro and Tableau's role licences, are cheap for a small audience and expensive for a large one. Capacity-priced options, Qlik's larger tiers or Power BI on Fabric capacity, cost more to start and less per head at scale. Self-hosted Metabase has no per-user fee at all. Count readers honestly: eighty managers who glance at one dashboard are eighty licences on a per-user plan.

3. What shape is your data?

One clean source, such as an accounting package or an ERP, suits almost anything. Many loosely related operational sources that people want to roam through freely is where Qlik's associative model shines. Data already in a well-run database suits Metabase. Marketing data in Google's products suits Data Studio. If the honest answer is "scattered across spreadsheets", fix that first: no tool will make a single source of truth for you.

4. What skills can you keep?

A BI environment lives for years, so ask who will maintain it after the implementer leaves. Power BI skills are the easiest to find in South Africa. Qlik and Tableau expertise exists but is scarcer and more expensive. Metabase needs someone comfortable with servers and SQL. Choosing a tool you cannot staff is choosing a future rescue project.

5. How does the cost grow?

Model three years, not one: users, data volume, refresh frequency and the currency risk on dollar-priced licences. Include the implementer, training and the time your own people will spend. The cheapest tool in year one is often not the cheapest in year three.

A rough guide, not a rule

Microsoft business, under fifty people needing reports: start with Power BI. Lots of operational data, explorers who ask "why?", or a big reading audience: look hard at Qlik. Salesforce at the centre, or analysts who present to the board: consider Tableau. Marketing and web reporting: Data Studio first. Own database, technical staff, tight budget: Metabase. Then test the front-runner on your own data before signing.

Protect yourself from the choice

Whichever tool you pick, keep the most valuable part, your clean, agreed business data and definitions, below the BI tool rather than inside it. Put it in a proper database or warehouse, with the logic for "revenue" or "margin" defined once. Then the BI tool becomes a presentation layer you could replace, not a vault your numbers are locked inside. It is the same principle as avoiding vendor lock-in anywhere else.

Run a proof of concept, briefly

  1. Pick two tools that fit your answers above. Not five.
  2. Use your real data from two or three sources, warts and all.
  3. Build three real reports from your ranked list of business questions.
  4. Put them in front of real users for a week or two, including a sceptical manager.
  5. Price the full picture at your expected scale in three years.

The right tool is the one your people use to make a decision in that fortnight, at a cost that still makes sense in three years. Then the hard part starts, which, as we keep saying, is the business telling the tool what it needs to see.