The right replacement depends on how your people ask questions of data. Analysts who live in SQL want a code-first tool, teams that need one trusted set of metrics want a semantic layer, and business users who'd rather not write queries want a spreadsheet or chat style interface. Pick the style first, then shortlist.
What do you want people to do with your data once it's sitting in a dashboard? If the honest answer is "ask their own questions without waiting for an analyst", you're shopping for something quite different from a team that mostly writes SQL and shares charts with each other. This list covers eight business intelligence tools for anyone moving on from Preset, in no particular order, grouped by how each one expects people to work.
Where a Superset-based setup gets tested
Preset describes itself as AI-native business intelligence built on Apache Superset, covering dashboards, embedded analytics, self-service exploration and a conversational layer, with an open source core underneath. That's a broad promise, and it's why switching is rarely about one missing feature. It's usually about how questions get asked, who's allowed to ask them, and who tidies up afterwards.
So the thing to check in any replacement is two-sided. First, how does someone who isn't an analyst get an answer: by typing a question, by filtering a dashboard, by working in a spreadsheet view, or by asking a colleague to write SQL for them? Second, once a dashboard exists, how is it shared, who keeps it current when a metric's definition changes, and what happens to the old copies people saved to their desktops?
Tools in this category split on exactly those points, and also on how much you set up and host yourself versus how much the vendor runs for you. That's why a feature grid won't settle it. Run the same real task through each shortlisted tool instead: one question your sales or finance lead asked recently, answered end to end, then shared with the people who asked it.
Three working styles among Superset replacements
Code first: SQL, markdown and notebooks
In this camp, Evidence treats reports as code written in SQL and markdown, Mode brings SQL, R and Python together with charts in one workspace, and Redash focuses on connecting to lots of databases, writing queries and sharing the visualisations. They suit teams where analysts or engineers do the building and everyone else reads the output. If your data people already keep their work in version control, this style will feel natural.
Semantic layer: one set of definitions
Cube, Metabase and Holistics put a modelling layer between the raw tables and the people asking questions. You define revenue, active customers or churn once, and every chart and AI answer uses that definition. The vendors all lean on the same idea: answers you can check and trust. This style needs someone to own the definitions, but it pays off when lots of non-analysts are exploring data on their own.
Spreadsheets, chat and search for business users
Sigma and ThoughtSpot aim at people who'd never open a SQL editor. One gives business users a familiar spreadsheet-style way to work on warehouse data, with chat, SQL and Python available too; the other centres on AI agents and automated insights that turn up inside the tools people already use. Both assume your data is already in a well-run cloud warehouse, so check that before you get excited by the demos.
Preset alternatives side by side
| Tool | Approach | Good for |
|---|---|---|
| Cube | Semantic layer analytics | Shared metrics behind dashboards and embedded analytics |
| Evidence | Code-based BI | Data products written in SQL and markdown |
| Holistics | Self-service BI | Data teams handing self-service analysis to colleagues |
| Metabase | Open source analytics | Checking the query behind each answer |
| Mode | Collaborative data platform | Analysts mixing SQL, R, Python and charts |
| Redash | Query and visualise | Querying many databases and sharing the results |
| Sigma | Warehouse-native BI | Business users building on governed warehouse data |
| ThoughtSpot | Agentic analytics platform | AI agents and insights inside everyday work tools |
The eight BI tools, A to Z
Alphabetical and in no particular order. Each line reflects how the vendor describes its own product, rewritten in plain English.
Cube
Best for: Teams wanting one semantic layer to feed internal BI and analytics embedded for customers, which the vendor frames as answers people can trust.
Evidence
Best for: Analysts who'd rather write than drag and drop; the vendor describes an open source, code-based approach where reports are built with SQL and markdown.
Holistics
Best for: Data teams setting up AI-assisted self-service reporting for the wider business, with the vendor's pitch centred on dependable answers to hard questions.
Metabase
Best for: Open source fans who want each answer tied to agreed metrics and business logic, and who like being able to inspect the query underneath, per the vendor.
Mode
Best for: Analyst teams working across SQL, R and Python alongside visual analysis, all in one shared workspace as the vendor describes it.
Redash
Best for: Engineers connecting to sources such as PostgreSQL, MySQL, BigQuery or MongoDB, then querying, charting and sharing results from one place.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Sigma
Best for: Organisations whose data sits in a cloud warehouse; the vendor says any team can build apps and dashboards through chat, a spreadsheet, SQL or Python.
ThoughtSpot
Best for: Larger teams wanting AI agents, automated insights and analytics embedded wherever people work, which is how the vendor sums up its platform.
My advice on choosing a Superset replacement
Start with the people, not the platform. If most of your questions come from analysts, a code-first tool keeps them fast and keeps your reports reviewable. If most come from managers who want to explore on their own, a semantic layer or a spreadsheet-style tool will save your data team hours of repeat requests. And if you're embedding dashboards in your own product for customers, put that requirement at the top of the list, because not every tool here is built for it.
Don't underestimate the move itself. Dashboards are easy to rebuild; agreed metric definitions and the trust people have in them are not. Write down your top ten metrics and how each is calculated before you touch a new tool. For a wider view of the category, see my guide to business intelligence tools, and if staying open source matters to you, my list of open source BI tools is the next stop.
Questions to settle with your data team first
- Who asks most of the questions: analysts, managers or customers inside your product?
- Where does the data live today, and is it clean enough for business users to explore without help?
- Who'll own metric definitions once they're in the new tool, and how will changes be announced?
- Do you need to self-host, or are you happy for the vendor to run everything?
- Which three dashboards would cause the most pain if they broke during the switch?
- How will people share a chart outside the tool, and should they be allowed to?
Do you make business intelligence software? Tool makers can ask to be considered for a featured place on this list.
Want your tool on this list? Featured spots and reviews here are paid partnerships. Here's how to partner with me.
Frequently asked questions
What is a semantic layer in BI?
It's a shared set of definitions that sits between your raw data and your charts. You define a metric once, and every dashboard and AI answer uses that same definition, so two reports don't disagree.
Is open source BI a good fit for a small team?
It can be, as long as someone is comfortable hosting and updating it. If nobody on the team wants that job, a vendor-hosted option usually works out simpler day to day.
Can business users ask data questions in plain English?
Several modern BI tools now offer chat or search style questions. The answers are only as good as the definitions behind them, so set up your key metrics before you open it up.
How long does it take to move dashboards to a new BI tool?
Simple dashboards rebuild quickly. The slow part is agreeing on metric definitions, permissions and who owns each report, so plan the move around those rather than the charts.
Do I need a data warehouse before choosing a BI tool?
Not always, but many newer tools assume one. If your data still sits in app databases and spreadsheets, check which sources each tool connects to directly.
Should analysts and managers use the same BI tool?
Ideally yes, so everyone sees the same numbers. Some teams pair a code-first tool for analysts with a simpler view for managers, but that means two places to keep in step.
