Tableau Alternative: Your Data Stays, the Per-Viewer Bill Goes

Somebody at Your Company Is Rationing Access to a Number

Not for security reasons. Because the next seat costs money, and someone has to decide whether this particular colleague is worth $15 a month to let them look at a chart.

That conversation happens in every Tableau shop. It is the clearest sign that something has gone sideways, because the whole point of buying a BI tool was to get information to more people, not fewer.

Tableau is superb software. None of what follows disputes that. The argument here is about two things it asks of you in exchange — a licence that scales with your audience, and a copy of your data on someone else's hardware — and whether the reporting you actually do requires either.

Start With the Bill, Because Everyone Else Does

Ask what people dislike about Tableau and the answer comes back the same way every time. This is not a matter of taste; the list prices are published.

$75 per Creator per month. $42 per Explorer. $15 per Viewer. Annual billing. Move up to the Enterprise tiers and it is $115, $70, and $35. Put twenty-five people on it and you are looking at roughly $20,000 to $25,000 a year — before embedded analytics, before AI add-ons, before enterprise support.

It also does not hold still. Prices went up about 9% across the portfolio in 2023, and roughly 6% again on 1 August 2025.

The real problem is the shape of it rather than the size. You are billed by audience. Every person who wants to see the number is another line on the invoice, so a dashboard that catches on is a dashboard that costs more. Success is the thing you are charged for, which is why the rationing conversation exists at all.

Yes, Tableau Server Is Real — Here Is Why Nobody Runs It

Fair is fair: Tableau Server installs in your own datacentre, and unlike certain competitors' on-premises editions it is the complete product, not a stripped-down one. If you want Tableau without the cloud, it genuinely exists.

So why does almost every deployment end up on Tableau Cloud anyway?

Salesforce took away the reason to self-host. List pricing for on-premises subscription licences rose by as much as 33% — done specifically to line them up with cloud pricing. Running it yourself used to save money. That saving was deliberately removed.

Think about what is left. You supply the servers. You run the upgrades. You staff the administration. And you pay the same per-seat rate as the customer down the road who does none of that. Self-hosting Tableau is not forbidden; it is simply a worse deal than it used to be, on purpose.

Getting the Data There Is Its Own Project

Tableau Cloud has to read your data from somewhere, and if your data sits in your own database, something has to carry it.

Extracts are the fast path. A .hyper file is a snapshot of your data, rebuilt on a timetable, parked on infrastructure you do not own. Between rebuilds, everyone is looking at a number that is already out of date. Live connections skip the snapshot and pay for it in speed, which is why teams that care about responsiveness drift back to extracts.

Tableau Bridge is what links Tableau Cloud to on-premises sources. Ask anyone who has stood one up: it is finicky, a couple of days of configuration is normal, and it becomes one more service you own, patch, and get woken up about. Strip away the product name and what you have built is a conveyor belt carrying your own records off-site, so they can be sent back as pictures.

Nobody set out to build it this way. It is the unavoidable consequence of putting the analytics somewhere other than the data.

What Changes With DataPallas

DataPallas installs on your own infrastructure, beside the database it reports on. Nothing is extracted, there is no Bridge, and there is no refresh timetable. It is modern, open source and self-hosted, it speaks JDBC so it talks to whatever you already run, and it is equally at home on Linux, Windows, or macOS.

DataPallas is open source with fair commercial pricing, and what changes here is who that pricing is built around: the people who build and administer reports, rather than the people who read them. Viewers are not a licensed tier at all.

Nothing Gets Copied Anywhere

Every query is answered by your own database, asked by a machine sitting on your own network. Since no third party ever receives the records, the entire supporting cast disappears: no Bridge to babysit, no extract schedule to tune, no snapshot going stale in the background.

For anyone working under GDPR, HIPAA, financial regulation, or a national data-residency rule, this usually decides the project. "The data never left" is a much shorter security review than explaining a replication pipeline.

Your Audience Stops Being a Line Item

Give the number to whoever needs it. Roll a dashboard out to the entire company and let the useful ones spread as far as they like — none of that changes what you pay, because readers are not what the licence counts.

Reporting That Is Not a Dashboard

Most Tableau evaluations quietly assume BI means dashboards. Look at what your organisation actually sends out each month and a lot of it is not.

• Documents that have to print correctly — invoices, account statements, payslips, regulatory filings. You design them in ordinary HTML and CSS: real page breaks, headers and footers that repeat on every sheet, and a preview that redraws the moment you save. XSL-FO handles exacting typographic work, DOCX templates with merge fields cover letters, and Excel output carries live formulas. Anyone who has tried to get a properly paginated statement out of Tableau knows this was never its job.
• Splitting one report into hundreds and sending them — divided by customer, employee, branch, or any other key, delivered over email, FTP/SFTP, cloud storage, or a portal, on a schedule, and checked for correctness before anything reaches a recipient.
• A portal people log into — where each person finds their own documents and only theirs, with alerts when something new lands, your branding on it, and payment built in if you are sending bills.

If a good share of what you ask Tableau to do boils down to "send this to these people every month," you have been paying visualisation prices for a mail run.

Charts You Can Put Inside Your Own Application

Interactive dashboards, and data tables, charts, and pivot tables that you add to your own product with one script tag — React, Vue, Angular, or ordinary HTML.

This is the scenario where per-seat pricing really bites. Analytics inside software you sell means every one of your customers is a viewer, and with Tableau each of those viewers is a line on your invoice. Here they are not counted at all — your licence covers the team building the components, however many end users see them.

One Definition of Every Metric

Cubes hold your dimensions, measures, joins, segments, and hierarchies in one place. Reports, dashboards, AI questions, and embedded components all resolve against that same definition, so the revenue figure on a slide matches the revenue figure in a script. AI writes the first version straight from your live schema.

Built in a UI, Kept as Files

Nothing here asks you to edit configuration by hand. Reports are set up on screen — source, layout, parameters, schedule — and AI sits beside any field that wants code, doing nothing at all until you ask it to.

The difference is what lands on disk afterwards. The Cube DSL, the queries, the templates — all plain text, all readable, all committable. A .twbx is a sealed package you need Tableau to look inside; these are files your team reviews in a pull request, compares between versions, and restores when a change turns out badly. Convenience and version control stop being a choice.

Speed Without Parking a Copy Elsewhere

If analytical queries would genuinely strain your production database, CDC replication streams changes into DuckDB or ClickHouse within seconds — on hardware you own, continuously rather than on a refresh timetable. All the performance of a snapshot, with the snapshot still inside your own walls.

Ask Questions Without Uploading Anything

Chat2DB turns ordinary English into SQL, and the AI crew helps across reporting, ETL, portals, and web work. Route a question through a cube and the model reasons over definitions you have already reviewed rather than guessing at column names — and the AI provider is your choice, not a decision baked into the platform.

The Honest Bit

At visual analysis, Tableau is the best there is. The drag-and-drop flow, the range of chart types, the way a good analyst can chase an idea through a dataset without ever stopping to write code — that took twenty years to refine and nothing here matches it. If you employ people who spend their days genuinely exploring data, Tableau is doing exactly what it was built for and doing it beautifully.

The question is how many of your Tableau seats belong to those people. In most companies, the majority are attached to a fixed set of reports and dashboards, consumed by colleagues who want the figure rather than the exploration. Those seats are paying exploration-tool rates, and the data is taking a trip to a vendor's cloud to make it happen.

Where to Start

1. Open your licence list and sort it by type. How many Creators are genuinely exploring, and how many seats exist so that someone can read a monthly figure?
2. Add up the annual total — then add the people who were quietly never given access.
3. Put DataPallas next to your database and rebuild one dashboard straight against live data. No extract, no Bridge.
4. Now build something Tableau was never comfortable with: a statement run split by customer and emailed out, or a report that has to paginate properly.
5. Judge it on three numbers — how many people you would actually have to licence versus how many seats you hold today, whether any data left the building, and how much of your monthly reporting one tool now handles.