Why the Sharpest Analysts in the Room Stopped Paying for the Software

The data tools industry has a strange business model: charge the people who understand data the most for the privilege of using software that increasingly gets in their way. Tableau licenses run thousands of dollars per user annually. Looker sits behind enterprise contracts that require a sales call to even get pricing. Microsoft Excel, the old reliable, is now bundled into a subscription that costs money every single month for software that hasn't fundamentally changed how most people use it in fifteen years.

Some analysts looked at this landscape and made a different call. They stopped buying the tools and started building them.

The Bloat Problem Is Real

Before getting into what people are switching to, it's worth naming what they're switching away from — and why.

Modern BI (business intelligence) platforms are built for procurement committees, not analysts. They're full of collaboration features, governance layers, permission systems, and dashboards designed to look impressive in executive presentations. All of that infrastructure is genuinely useful at scale, in large organizations, with dedicated data teams.

For the individual analyst, the independent consultant, the small finance team, or the startup data person who wears six hats? It's overhead. Expensive, slow, opinionated overhead that often makes simple tasks harder than they need to be.

"Tableau is incredible if you need to deploy visualizations to a hundred stakeholders who can't write a line of code," says Rachel Domingo, a financial analyst in Austin who consults for mid-size companies. "If you're just trying to answer a question about your data, it's like using a forklift to move a cardboard box."

Enter the Lightweight Stack

Domingo's current toolkit would be unrecognizable to most corporate data teams. At its center is SQLite — a file-based database that requires no server, no installation beyond a single binary, and no configuration. For exploration and transformation, she uses DuckDB, an analytical database that can query CSV files directly and runs entirely in memory. For visualization, she uses Observable Plot, an open-source JavaScript library.

Total licensing cost: zero dollars.

"I can open a 2GB CSV file and query it in seconds with DuckDB," she says. "That's not something Excel can do without choking. And I own every piece of the stack."

Ownership matters here in a specific way. When your analysis lives in a proprietary tool's file format — a Tableau workbook, a Power BI .pbix file — it's readable only by that tool. If the tool changes its format, raises its prices, or gets discontinued, your work becomes inaccessible. Plain SQL queries in a text file will run against any SQL database that has ever existed and will ever exist.

VisiData: The Spreadsheet That Respects You

For raw data exploration, a growing number of analysts have discovered VisiData — a terminal-based tool that lets you open, filter, sort, and transform tabular data from the command line. It handles CSV, Excel, JSON, SQLite, and dozens of other formats. It's scriptable. It's fast. And it runs entirely on your machine.

Mike Tran, a data engineer in Seattle, calls it "the spreadsheet for people who are tired of spreadsheets."

"The moment you realize you can do a frequency analysis on a million-row CSV in two keystrokes, you start questioning why you ever opened Excel for that kind of work," he says. "VisiData doesn't try to help you. It just lets you work. That's what I want from a tool."

The learning curve is real — VisiData lives in the terminal and uses keyboard shortcuts that take time to internalize. But Tran argues that the investment pays off quickly. "I spent a weekend with the documentation. Now I'm faster than I've ever been with any GUI tool. The friction was front-loaded."

The Self-Hosted BI Option

For teams that need to share dashboards and reports without paying Tableau's per-seat pricing, a cluster of open-source alternatives has matured significantly in the last few years. Metabase offers a clean, approachable interface that non-technical stakeholders can actually use, and the self-hosted version is completely free. Redash is leaner and more developer-friendly. Apache Superset is more powerful and correspondingly more complex to configure.

None of them require a sales conversation. None of them send usage telemetry to a vendor. All of them can be run on a $10-a-month server or a machine sitting under someone's desk.

"We replaced a $2,400-a-year Tableau Desktop license with a self-hosted Metabase instance that cost us one afternoon of setup time," says Domingo. "The stakeholders who use it don't even know the difference. They just see the dashboard."

What You're Actually Buying With the Expensive Software

This is the uncomfortable question the enterprise software industry doesn't want analysts asking: what are you actually getting for the money?

Some of it is real. Tableau's visualization capabilities are genuinely excellent. Power BI's integration with the Microsoft ecosystem is legitimately useful if you're already deep in that world. For large teams with diverse technical skill levels, the polish and guardrails of enterprise tools have value.

But a significant portion of what you're paying for is vendor support, brand familiarity, and the comfort of having a company to blame when something goes wrong. Those are organizational values, not analytical ones.

The analysts building their own stacks have made a different calculation. They're betting that their time spent learning open-source tools is worth more than the time saved by using software that does the thinking for them. They're betting that owning their data pipeline — truly owning it, format and all — is worth more than the convenience of a polished UI.

So far, the bet is paying off.

The Spreadsheet Isn't Going Anywhere

None of this means Excel is dead or that every analyst should immediately migrate to a terminal-based workflow. For most people in most organizations, the existing tools work fine, and the switching cost isn't worth it.

But the existence of a community of data professionals who have walked away from the expensive stack — and found something better on the other side — is worth paying attention to. These aren't people who couldn't afford the software. They're people who understood it well enough to know they didn't need it.

That's a different kind of statement.

When the sharpest people in a room stop buying the premium tool and start building their own, it's usually worth asking what they know that everyone else doesn't.