Replacing Notion and Jira

The seam nobody writes down

What moves, what arrives incomplete, what stays put

Sorted by how much manual work each row costs you.

What the Notion importer really does

Taken from the shipped function, not from a roadmap.

• It lists what the integration can see

  You paste an integration secret or connect Notion once for the whole organisation. The picker shows every page shared with that integration and preselects the top-level ones.

• It carries text, structure and to-dos

  Paragraphs, three heading levels, bulleted and numbered lists, to-dos with their checked state, quotes, code, callouts, dividers and toggles all arrive. Sub-pages come along to three levels deep and become nested docs.

• It skips media and tables, and counts what it skipped

  Images, files, embeds, tables, databases, synced blocks and column layouts are dropped, and the import screen tells you how many blocks it dropped. Numbered lists arrive as bullets. Toggles arrive as bullets and stop folding.

• It has hard stops

  Twenty-five selected pages per run, sixty pages and fifteen hundred blocks total, four levels of indent, four thousand characters per block. A large workspace needs several runs, and page properties, comments and version history never come across.

A four-stage migration

Stage one takes an afternoon. Nothing after it is urgent.

1. 1

   Import the docs that are still true

   Do not import the whole Notion workspace. Pick the pages your team opened in the last quarter, run the import, and read the skipped-block count. If it is high, those pages were mostly screenshots and tables, which tells you something about how much of your documentation was actually text.

2. 2

   Rebuild the live sprint by hand

   Export the current sprint from Jira to CSV, then create the tasks in a Polaris workstream with owners and dates set as you go. Retyping thirty tickets takes about two hours and produces a backlog with no dead items in it, which a CSV import would not.

3. 3

   Run one sprint in parallel

   Keep Jira open and read-only for one full cycle. Product writes in Polaris Docs, engineering works the Polaris tasks, and nobody grooms two backlogs. If the team drifts back to Jira, you have learned something cheap.

4. 4

   Hire the worker that closes the seam

   Assign an AI worker the standing job of reading the doc and the task and flagging where they disagree. A cloud machine wakes per task, works the tool loop, and delivers the difference as a comment with a file. You close the task; the machine never does.

5. 5

   Cancel the Jira seats, not the Jira account

   Drop write seats once nobody needs them and keep the archive readable if your retention policy asks for it. Cancelling the account is the step teams regret, and it saves almost nothing compared to dropping seats.

Read next

Questions people ask

Related