TL;DR Verdict
Evidence Dev earns a 7.5 out of 10 for data teams that think in SQL and want to ship polished reports without wrestling a drag-and-drop interface. It is built for analysts and developer-adjacent folks who are comfortable in a code editor, not for business users who expect a point-and-click experience. Its biggest strength is the clean, version-controlled workflow that makes reports feel like software rather than dashboard sprawl. Its biggest weakness is that the learning curve immediately filters out anyone who does not write SQL for a living.
What Evidence Dev Actually Is
Evidence Dev is an open-source, SQL-first business intelligence tool that turns markdown files and SQL queries into polished, interactive reports published as static websites. Think of it as the reporting tool built for analysts who already live in a code editor and version-control everything.
The core idea is simple. You write a .md file with regular markdown prose. Inside that file, you drop SQL queries that pull from your data source. Evidence Dev runs those queries at build time, slots the results into chart and table components, and generates a fast, shareable static site. No proprietary drag-and-drop canvas. No chart editor buried behind five modal windows.
The company was founded in 2021 by a small team in Canada. By 2026, the product has matured well past its early experimental phase. The open-source repository has thousands of GitHub stars, the core connector library covers the data warehouses most small teams actually use (BigQuery, Snowflake, DuckDB, PostgreSQL, Redshift, MotherDuck), and Evidence Cloud, their hosted deployment service, is stable enough for production use.
The audience is specific on purpose. Evidence Dev is not trying to compete with Tableau or Power BI for enterprise drag-and-drop reporting. It targets data analysts, analytics engineers, and small technical teams who find those tools slow and bloated. If you already use dbt to transform your data, Evidence Dev sits naturally right after it in your stack. Your models become the data sources, your reports become the deliverable, and the whole thing lives in git alongside everything else.
That focus keeps the product coherent and avoids feature bloat. It also means the tool simply will not work for non-technical stakeholders who need to build their own views.
Pricing And Plans
Evidence Dev follows an open-core model. The core reporting framework is free and MIT-licensed. You can self-host for nothing, and the feature set you get is genuinely full-featured, not a stripped-down teaser.
Evidence Cloud, the managed hosting layer, is where they monetize. As of 2026 the tiers look roughly like this:
| Tier | Price | What You Get |
|---|---|---|
| Open Source | Free forever | Self-hosted, full framework, all connectors |
| Evidence Cloud Free | Free | 1 project, public reports, community support |
| Evidence Cloud Team | Around $50/month | 3 projects, private reports, team access controls, SSO |
| Evidence Cloud Business | Around $200/month | Unlimited projects, priority support, custom domains |
| Enterprise | Custom pricing | Air-gapped deployments, SLAs, dedicated support |
A few gotchas worth knowing. The free cloud tier only supports public reports. If your data should not be visible to anyone with the URL, you need at least the Team tier. Build minutes are also capped on lower tiers, so if your queries run long or you rebuild frequently, you can hit limits faster than expected.
Self-hosting on your own infrastructure is genuinely free with no seat-based catches. You need somewhere to run Node.js and a deployment pipeline to trigger builds. Platforms like Vercel, Netlify, or Cloudflare Pages work well. If you are already paying for one of those, the total cost for a self-hosted Evidence setup can be effectively zero.
The pricing is fair for what you get. Small teams on a budget should start with the open-source route. If you want Evidence Cloud to handle deployments and access control, the Team tier is a reasonable spend compared to proprietary BI tools that charge per seat.
Setup Experience
The first 30 minutes with Evidence Dev go smoothly if you have Node.js already installed and some comfort with a terminal. If you do not, the initial setup feels like a toolchain installation more than a product onboarding.
You run npx degit evidence-dev/template my-project, open the folder in VS Code, run npm install, then npm run dev. A local dev server starts, and you get a browser preview of a sample report. The whole sequence takes about five minutes when nothing goes wrong.
The VS Code extension is where things get genuinely pleasant. It adds autocomplete for Evidence components inside markdown files, which matters because the syntax for things like <LineChart> or <DataTable> is not always obvious from memory. Without the extension, you are flipping between your editor and the documentation constantly.
The documentation is good. It is well-structured, has real code examples, and does not assume you already know how the tool works. The connector setup pages are particularly clear. Connecting to BigQuery, for example, gives you exact steps for creating a service account, where to paste the JSON key, and what to set in your .env file.
Where things trip up early: the build and query caching behavior takes a session to understand. Evidence Dev caches query results and only reruns them when the query text changes. This is smart for performance but confusing when you are debugging a query and not seeing fresh data. You have to learn to force a cache clear, which is not prominently documented in the getting-started flow.
The templated routing system, where a single page file like [customer].md generates one report per customer, is powerful but not immediately obvious. Most people miss it on the first pass. Once you find it, it changes how you think about building report libraries.
If you have used a static site generator like Next.js or Hugo, the mental model clicks fast. If your entire reporting background is Tableau or Google Data Studio, the model feels foreign at first. Give yourself a full working day before deciding whether it suits you.
What It Does Well
SQL as the only interface
You write SQL, you get data. There is no abstraction layer between your query and the output. You can use CTEs, window functions, subqueries, and anything your database supports. This matters when your reports require complex logic. In many GUI-first tools, you sometimes have to work around the tool’s limitations to express what you actually need. In Evidence Dev, if your database can run it, your report can show it. For analysts who spend their day in a SQL editor, this feels like reporting finally caught up with how they already think.
Version control as a first-class concept
Every report is a text file. That means you get git history, pull request reviews, branching for draft reports, and rollback if something breaks. You can open a PR for a new quarterly report, have a teammate review the SQL before it goes live, and merge it like code. No other mainstream tool in the bi-tools category treats reporting workflow this naturally. This single feature makes a real difference for teams where data accuracy is non-negotiable and audit trails matter.
A mature and useful component library
Evidence Dev ships a solid library of chart types: line, bar, scatter, area, candlestick, funnel, and more. Tables support sorting, pagination, and column formatting out of the box. There are also UI components for filters, date pickers, and dropdowns that feed into your SQL via query parameters. You can build a genuinely interactive report, not just a static chart dump, using only markdown and SQL. The chart defaults look clean without requiring custom CSS work on your end.
DuckDB integration for local and embedded analytics
DuckDB works as a native connector, which means you can run queries directly against local CSV or Parquet files without a cloud warehouse. For small teams doing ad-hoc analysis, this is a significant capability. You can point Evidence Dev at an exported data file, write your SQL, and have a shareable report in minutes. For teams using MotherDuck, the cloud version of DuckDB, the same connector works with minimal configuration change. It makes Evidence Dev viable even before you have a warehouse set up.
Templated pages for scalable report libraries
The file-based routing system lets you write one report template and generate hundreds of individual report pages from it. A single [region].md file can produce separate report pages for every region in your data, each with the correct filtered data. This is the kind of feature that sounds like a nice-to-have until you are managing 50 customer-facing reports and realize you can maintain them all by editing one file. It scales far better than duplicating dashboards manually.
dbt integration that actually works
Evidence Dev connects naturally with dbt projects. You reference dbt models as your data source, which means the transformation logic stays in dbt and the presentation logic stays in Evidence. The two tools have overlapping communities and the integration is well-documented. If you are already on dbt, Evidence Dev fits the existing workflow without forcing you to reorganize your project. See also how to connect dbt to your BI tool for a broader look at this workflow pattern.
Where It Falls Short
Non-technical users cannot self-serve
If a product manager or a finance stakeholder wants to pull their own numbers, Evidence Dev does not help them. There is no drag-and-drop interface, no formula builder, and no way to explore data without writing SQL. Every report has to be built by someone who codes. This is a fundamental architectural choice, not a missing feature you can expect in the next release.
Build times get slow with many queries
Every time you deploy, Evidence Dev runs your queries, fetches the results, and generates static pages. With a few queries on a fast warehouse this is fine. With dozens of queries across multiple slow data sources, builds can stretch into minutes. There is no incremental build system that only re-fetches changed queries. The caching helps during development, but production deploys on large sites can feel sluggish compared to tools with live query execution.
Alerting and scheduled delivery are not native
Evidence Dev produces static reports. It does not send you an email when a metric crosses a threshold or Slack you a chart on Monday morning. For those workflows you need to wire in a separate tool. This is a real gap for teams that rely on proactive data delivery rather than people actively visiting a report URL. You can work around it with external schedulers and notification services, but it adds complexity to what is otherwise a simple stack.
The self-hosted path requires DevOps comfort
Self-hosting is free but not zero-effort. You need a Node.js environment, a CI/CD pipeline to trigger builds, and somewhere to serve the static output. Evidence Cloud removes that friction but adds cost. Teams without someone comfortable in the terminal will find the self-hosted route intimidating. The documentation covers it, but it is still more setup than clicking “deploy” in a SaaS product.
No real-time data support
Because Evidence Dev builds static sites at deploy time, your reports reflect data as of the last build. Live dashboards that update every few minutes are not the right use case here. If you need near-real-time visibility into operational metrics, you need a tool with live query execution. Evidence Dev is best for reporting on data that changes daily or weekly, not second-by-second.
Best Alternatives To Consider
If Evidence Dev does not fit your situation, here are four tools worth evaluating. For a broader comparison across the space, see the open-source BI tools roundup for 2026.
Metabase is the most popular open-source BI tool and the clearest alternative. It offers a visual query builder that non-technical users can actually navigate. If your team includes people who do not write SQL, Metabase is a better starting point. The free self-hosted version is genuinely capable, and the paid cloud tier starts around $500/month for teams.
Lightdash is a dbt-native BI tool that sits close to Evidence Dev’s target audience. It generates a metrics layer from your dbt models and gives users a GUI to explore those metrics without writing SQL themselves. If you want the dbt integration of Evidence Dev but also want non-technical stakeholders to self-serve on top of it, Lightdash is worth a serious look.
Apache Superset is a powerful open-source option that covers a wider range of chart types and has a visual dashboard builder alongside SQL access. The self-hosted setup is more complex than either of the above, but the feature set is deep. Good for teams that want SQL access and a visual layer in the same tool without paying for a commercial product.
Observable Framework is the closest conceptual sibling to Evidence Dev. It uses a similar code-and-markdown paradigm for building data reports. The main difference is that Observable leans more toward JavaScript and D3-style custom visualizations, while Evidence Dev leans toward SQL and pre-built components. If you want more visualization flexibility and have JavaScript skills on the team, Observable is worth a direct comparison.
Who Should Use Evidence Dev
The solo analytics engineer
If you are a single analyst or analytics engineer supporting a small company, Evidence Dev fits well. You already write SQL all day. You want reports that look professional and can be shared with stakeholders via a URL. You do not have budget for a Tableau license and you do not want to maintain a heavier BI server with a GUI you find slow. Evidence Dev lets you build and deploy a full reporting site using the skills you already have, without adding a tool that requires relearning your entire workflow from scratch.
The data-forward startup team
If you are a small startup with at least one technical data person and stakeholders who trust that person to own reporting, Evidence Dev is a strong fit. Your needs are reporting and sharing, not deep self-service exploration by non-technical staff. You want reports that look polished enough to share with investors or board members. You want everything in git so the whole team can review and contribute. The free self-hosted option keeps costs low while you grow, and Evidence Cloud gives you a straightforward path to access controls when you need them later.
The dbt team looking for a presentation layer
If your team already runs dbt for transformations and you are looking for something to sit on top of it, Evidence Dev is purpose-built for this workflow. You reference your dbt models as sources, you write SQL against those models, and you get reports that reflect exactly the same data your transformations produce. For a detailed look at how this fits together, see connecting your dbt project to a reporting layer. No separate metric definitions to maintain in a GUI, no semantic layer negotiation, just SQL all the way through.
Frequently Asked Questions
Is Evidence Dev free to use?
The core framework is open-source and free to self-host with no usage limits. Evidence Cloud offers a free tier that supports one project with public reports. You need a paid plan starting around $50/month if you want private reports or multiple projects on their hosted service.
How steep is the learning curve?
If you know SQL and have used a markdown-based static site generator before, you can build your first real report in a few hours. If you are new to either, expect a full day of setup and experimentation before things click. The documentation is solid and the VS Code extension makes day-to-day authoring significantly easier once you have it installed.
What data sources does Evidence Dev support?
As of 2026 the connector list includes BigQuery, Snowflake, Redshift, PostgreSQL, MySQL, DuckDB, MotherDuck, SQLite, Microsoft SQL Server, and several others through community connectors. Most modern data warehouses are covered. Flat file support via DuckDB means you can also query CSVs and Parquet files without a warehouse, which is genuinely useful for early-stage teams.
Can I use Evidence Dev without a cloud warehouse?
Yes. DuckDB lets you run SQL queries directly against local files. This makes Evidence Dev usable even if you are working with exported CSVs or Parquet files from another tool. It is one of the more underrated aspects of the product for small teams that do not have a Snowflake or BigQuery setup yet.
How hard is it to migrate away from Evidence Dev if I outgrow it?
Because all your reports are plain markdown and SQL text files, migration is straightforward in principle. Your SQL queries port directly to any other tool that accepts SQL. The challenge is recreating the presentation layer, since Evidence-specific components are not universal. Budget a few days of rework if you move to a tool with a different templating system. The SQL reporting tools comparison covers where Evidence fits relative to the tools you might move to next.
Bottom Line
Evidence Dev is a focused, well-executed tool for a specific kind of user: the analyst or small data team that thinks in SQL, values version control, and wants to publish clean reports without a bloated BI platform in the way. It is not trying to be Tableau and it should not be. The open-source core is genuinely free, the cloud hosting is reasonably priced, and the development experience is pleasant once you understand the build model.
The score stays at 7.5 out of 10 because the tool’s strengths are real but its limitations are structural. Non-technical users cannot self-serve, real-time data is not a fit, and the setup requires more technical comfort than most SaaS tools demand. If those constraints match your situation rather than contradict it, Evidence Dev is one of the most coherent reporting tools available at any price point.
Want to try Evidence Dev? Start with Evidence Dev and see if it fits.