How to Track Projects in Coda Without a Second Tool
Most project chaos is one spreadsheet with three tabs and a status doc that contradicts it. Coda collapses those into one doc with three linked tables, and the views do the reporting that someone was doing by hand on Fridays.
Build time is roughly an hour for a doc that holds 40 projects. The maintenance is where the win is: one place to update, everything else recalculates.
What are the three tables you need?
Projects, tasks, and people. Projects hold the outcome and the date; tasks hold the work with an owner and a due date; people hold names and capacity. Link each with a relation column and the rollups fall out.
Do not add a fourth table for status history on day one. It sounds thorough and it will be empty after two weeks.
How do the rollups actually work?
A rollup on a project row can count related tasks and filter to overdue ones. Coda’s rollup editor lets you aggregate a related table with a condition, so “count of tasks overdue” is one field, not a formula you hand-write.
| Rollup | Sources | Use |
|---|---|---|
| Work complete | Tasks count where done | Progress bar |
| Blocked count | Tasks where blocked | Daily triage |
| Next due | Tasks min due, not done | Alert list |
| Owner load | Projects by person | Capacity check |
The next-due rollup is the field people actually use. It turns a project row from a name into an answer.
How do you make it state its own problems?
Create views, not reminders:
- Late: due date before today and status not done, sorted by date.
- Unassigned: owner blank, any status.
- Stale: task not updated in 14 days while open.
- This week: due within seven days, grouped by owner.
Share the doc at the section level so each owner opens straight to their view.
The stale view finds more problems than any status meeting. Nobody reads a stale row and defends it.
Where does AI help and where does it not?
Use the AI column to draft a one-line status summary from a task list, which is fine because the underlying facts come from rows. Do not ask it to decide priority; it will rank by wording, not by commitment.
A weekly summary column set to run on demand saves maybe 15 minutes of writing. Set to run automatically, it burns AI credits on rows nobody reads.
For teams already standardized on a coding workflow with issue trackers, moving project tracking into Coda duplicates what the tracker already does. Keep Coda for the projects that live outside a repository, such as launches, content and hiring.
What breaks at scale
Above roughly 10,000 task rows the doc slows, especially with several rollup columns. Formula-heavy docs also get hard to hand off; write a row in your people table for “who understands this doc.”
Permissions at page level mean a doc shared with a client can accidentally expose cost or hiring sections. Use separate docs for external sharing.
Quick Summary
- Three tables, two relation columns, four rollups: that is the whole tracker.
- Rollups for work complete, blocked count, next due and owner load replace manual reporting.
- Views for late, unassigned, stale and this week beat status meetings.
- AI is fine for one-line summaries, wrong for priority calls.
- Docs get slow near 10k task rows; keep client-facing sections in a separate doc.
