Are you reading numbers or changing them?
This is the split that decides the shortlist, and it is invisible in a feature comparison. A BI tool shows you what happened. A planning platform lets you write numbers back — enter a budget, run a scenario, submit a forecast for approval — and that requires a fundamentally different architecture.
Jedox and Board International are planning platforms. Jedox's in-memory OLAP engine supports write-back with workflow and approval tracking, so a budget cycle runs inside the system rather than through forty spreadsheets emailed to a controller. Board International unifies BI, planning and predictive analytics in one model, with scenario simulation across functions — change a sales forecast and see the effect on supply chain and headcount.
Lightdash, Cluvio, Toucan Toco and Luzmo are read-only analytics. They are better at exploration and visualisation and cannot accept a number back.
The expensive mistake is buying a dashboard tool for a budgeting problem. Finance then keeps budgeting in Excel, the dashboard reports last month's actuals against a plan that lives somewhere else, and the two disagree by March.
Why does Excel keep winning, and which tool accepts that?
Because finance teams are genuinely fast in Excel, and every BI rollout that tries to replace it discovers that people export to Excel anyway — at which point the governed model has been abandoned and nobody notices for two quarters.
Jedox takes the opposite position: its Excel add-in is good enough that finance keeps working in Excel, with the cells connected to the central model and writing back into it. The spreadsheet becomes an interface to the system rather than an escape from it, which is the only arrangement that has ever survived contact with a finance department.
Board International also integrates with Office for the same reason.
The lesson generalises. Any BI rollout should ask where the data currently gets analysed and either meet people there or accept that the new tool becomes a reporting layer nobody uses for decisions. Fighting Excel head-on has a consistent historical record.
Where do metric definitions live, and why does that decide everything?
Because two teams reporting different revenue numbers is the failure mode that destroys trust in analytics permanently, and it is almost always a definitions problem rather than a data problem.
Conventional BI redefines metrics inside the BI tool, so "active customer" is written once in the data warehouse and again in the dashboard — and the two drift the first time somebody changes one. Lightdash removes the duplication by reading metrics directly from your dbt project: defined once, version-controlled, reviewed in a pull request, and identical everywhere they appear.
That is a genuinely better arrangement, and it comes with a hard prerequisite Lightdash states plainly: it only makes sense if you already model in dbt. For a team without that, Lightdash solves a problem they do not yet have.
Cluvio is the honest counterpoint at the other end — SQL-based, fast to set up, and explicitly without a semantic layer, so definitions can drift. For a startup that is an acceptable trade for getting a dashboard the same afternoon; for an organisation with several teams it is where the trust problem starts.
Does your data have to move to be analysed?
No, and for European organisations that matters more than it usually does. Most cloud BI works by extracting data into the vendor's platform, which means a copy of your customer, financial and operational data now lives somewhere else and needs its own transfer analysis, retention policy and access review.
Pyramid Analytics queries source systems directly rather than extracting, so the analysis reaches the data instead of the data reaching the analysis — no second copy, no synchronisation lag, and no additional processor holding your records. It offers on-premise, private cloud and EU cloud deployment.
Lightdash self-hosted goes further: the whole BI layer runs on your infrastructure, MIT-licensed, with the warehouse it queries also yours. There is no vendor in the path at all.
Jedox and Board International both support on-premise deployment, and Toucan Toco offers self-hosting on request for enterprise. Cluvio and Luzmo are cloud-only, though both host in the EU.
Who is the dashboard actually for?
An analyst wants to explore: slice, filter, pivot and chase a hypothesis. A field engineer wants one number and what to do about it. Building for the first and giving it to the second produces a dashboard nobody opens twice.
Toucan Toco is built explicitly for the second audience, using guided narrative dashboards rather than exploration tools — contextual annotations explain what a number means and why it moved, and the design is mobile-first rather than a desktop layout that reflows. For frontline and field teams that is the difference between analytics being used and being installed.
The cost is stated openly: Toucan Toco is not suited to analyst-led exploration, and each dashboard requires design effort rather than being generated from a dataset. You are paying for editorial work, not just for software.
Lightdash sits opposite, with self-service exploration for non-SQL users on top of governed dbt metrics. Cluvio requires SQL, so it serves analysts and no one else — which is fine if analysts are the audience and fatal if they are not.
