Metabase is the best starting point for most small teams — ask questions in plain language, self-host free or pay for cloud. Looker Studio is free and fine if your data already lives in Google. Preset (managed Superset) and Lightdash are the open-source-backed options with real modelling. Power BI is cheap only on the right Microsoft plan — Pro is included with E5, not with ordinary Microsoft 365 subscriptions; Tableau on visual depth at enterprise cost. Hex and Omni are the newer analyst-first tools. And if the dashboard needs to sit inside an app you built, next to the data it reports on, that is a different shape — describing it to Greta builds it. Choose on who writes the queries, because that decides everything else.
The dashboard tool is rarely the reason a small team's reporting fails. The reason is that nobody agreed what the numbers mean.
Who writes the queries?
That question sorts the category faster than any feature list. Nobody technical → Metabase's question builder or Looker Studio. One analyst → Hex, Omni, Preset, or Metabase with SQL. A data team with dbt → Lightdash or Looker. On a Microsoft E5 plan → Power BI, where Pro is included; on other Microsoft 365 plans it is a separate per-user purchase, so check before assuming it is bundled. Buying a tool designed for the next tier up is the most common expensive mistake here.
At a glance
| Tool | Who it is for | Self-host | Pricing shape |
|---|---|---|---|
| Metabase | Small teams, non-technical questions | Yes, open source | Free self-host; paid cloud tiers |
| Looker Studio | Google-centric reporting | No | Free |
| Preset | Managed Superset, analyst-led | Superset is open source | Tiered |
| Lightdash | dbt-modelled data | Yes, open source | Tiered |
| Power BI | Microsoft shops on the right plan | No | Pro included with E5; per user otherwise |
| Tableau | Visual depth, enterprise | No | Higher per user |
| Hex / Omni | Analyst-first notebooks and modelling | No | Quoted |
| Greta | Dashboards inside the app you built | Your deployment | Free tier; $20/mo ($5 welcome) |
Pricing shapes checked September 2026 — every vendor here re-tiers; verify live before budgeting.
The tools, briefly
Metabase earns its default status: someone non-technical really can answer a question without SQL, and the self-hosted version is genuinely free. Looker Studio costs nothing and connects to Google's world effortlessly; it gets awkward once your data is not in that world.
Preset gives you Superset without operating Superset, which is a real service. Lightdash is the right answer when your semantics already live in dbt — it reads your models rather than asking you to redefine them in a BI tool, which is where definition drift usually starts.
Power BI is hard to argue against on price if your Microsoft plan already includes it — Pro comes with E5, and on E3 or a small-business plan it is a separate per-user licence. Two separate things to check: authoring a report and sharing or viewing it have different licence requirements, so a team where one person builds and eight people read is eight licences to account for, not one (verify against Microsoft's current licensing documentation). Tableau remains the strongest at visual analysis and prices accordingly. Hex and Omni are built for analysts who want notebooks and modelling in the same place, and they are overkill for a team whose main need is a weekly number.
Got an idea? Build it now!
Just start with a simple prompt. No coding required — Greta.sh turns your idea into a working app in minutes.
The problem no BI tool solves
Definition drift. Finance's "active customer" excludes trials; the product dashboard's includes them; the board deck uses a third number from a spreadsheet. Every tool above will faithfully render all three. The fix is a written definition per metric, owned by a name, referenced by the dashboard — which is process, not software, and no purchase substitutes for it. The dashboard review guide is the checklist for catching drift before it reaches a board deck.
When the dashboard belongs inside the app
If the numbers you report on are generated by an app you built, a separate BI tool means a connection, a sync and a second place to maintain definitions. Building the reporting into the app — describing it to Greta — keeps the definition next to the data that produces it, and the team already has logins. Free tier is 30 credits a month (5 a day); $20/month after ($5 with the current welcome offer).
Stated plainly, because it matters: this is not a BI tool. Ad-hoc exploration, cross-source joins, scheduled distribution, drill-downs and the visualisation library that Tableau has spent twenty years on are all things you would be giving up. If people need to ask questions nobody anticipated, buy Metabase. This path fits a fixed set of numbers that belong to one app.
FAQ
What is the best free BI tool? Metabase self-hosted if you can run it, Looker Studio if you cannot and your data is in Google. Both are genuinely usable rather than crippled trials.
Metabase or Power BI? Power BI if your Microsoft plan already includes Pro — E5 does, most others do not — and someone will learn it. Price the viewers as well as the author before deciding. Metabase if non-technical people need to answer their own questions without training.
Do I need a warehouse first? Not to start — most tools connect directly to a production replica. You need one when queries start affecting the application, or when you need to join sources. Reading from the primary database is the shortcut that works until it suddenly does not.
Why do two dashboards show different numbers? Almost always different definitions, not a bug. Compare the filters and the date boundaries before suspecting the tool, and write the agreed definition down once you find the difference.
