“A teammate asked how they managed build analyses that colleagues can interact with without running code themselves. They started explaining and realized every step ran through hex. It had become the spine of the process without a formal decision to make it so.”
When I'm asked for a weekly revenue reconciliation that finance can run thems, I want to build analyses that colleagues can interact with without running code themselves, so I can document the logic of an analysis alongside the code that produces it.
A data analyst or analytics engineer at a company with a modern data stack — dbt, Snowflake or BigQuery, and a growing demand from business stakeholders for self-service data access. They use Hex because Jupyter notebooks are hard to share and dashboards aren't flexible enough. Hex sits in the middle: code-first enough for real analysis, shareable enough that a PM can click through an interactive version without needing to run code. They build notebooks in Hex. Business people use the published apps. This is the workflow they've been trying to build for years.
To reach the point where build analyses that colleagues can interact with without running code themselves happens through hex as a matter of routine — not heroic effort. Their deeper aim: document the logic of an analysis alongside the code that produces it.
hex becomes invisible infrastructure. Build analyses that colleagues can interact with without running code themselves works without intervention. The old problem — computation time on large queries that breaks the exploration rhythm — is a memory, not a daily fight. Computation caching that persists across sessions for expensive queries removes.
The CFO has asked for a weekly revenue reconciliation that finance can run themselves without filing a data request. The analyst is building this in Hex: SQL cells that pull from the data warehouse, Python cells for the reconciliation logic, and an app layer with date range pickers and export buttons. When they publish it, finance will have a self-service tool. The analyst will stop receiving this request every Monday. They're building that future right now.
Uses Hex 3–5 days per week. Connects to Snowflake or BigQuery via Hex's data connections. Writes SQL primarily; uses Python for transformation and visualization. Has published 6–15 Hex apps that business stakeholders use as self-service tools. Uses Hex's scheduled runs for recurring analyses. Collaborates with 1–2 other analysts in shared notebooks. Has a personal notebook library organized by domain: revenue, product, marketing, operations. Uses Hex alongside dbt — Hex for exploration and sharing, dbt for production data modeling.
The proof is behavioral: build analyses that colleagues can interact with without running code themselves happens without reminders. They've customized hex beyond the defaults — templates, views, integrations — and their usage is deepening, not plateauing. When new team members join, they hand them their setup as the starting point.
The trigger is specific: version control that doesn't integrate with the Git workflow they use for everything else, combined with a high-stakes deadline. hex fails them at exactly the wrong moment. That evening, they're reading comparison posts. What makes it irreversible: they fundamentally believe the analysis and the documentation of the analysis should live in the same place, and hex just proved it doesn't share that belief.
Pairs with `amplitude-primary-user` for the product analytics vs. ad hoc analysis workflow boundary. Contrast with `data-engineer` for the analysis layer vs. data infrastructure responsibility split. Use with `mixpanel-primary-user` for the self-service analytics gap that Hex fills for SQL-native analysts.