How to Move a Creative Project Out of Notion: A Step-by-Step Guide

What you walk away with

• A clear diagnosis of what was actually broken, which is often not the thing you think
• Your workspace split into the four parts it really contains, each headed somewhere appropriate
• A complete export with the attachments and databases intact
• A rebuilt visual layer that preserves the arrangement, not just the assets
• An old workspace archived rather than deleted, and a rewired set of links

Read this before you export anything

The most common outcome of a well-run migration is a split, not a move.

Notion is genuinely excellent at three things: structured databases with multiple views, long-form documents, and being a team wiki. If your project relies on any of those, moving them out will make your life worse, and you will spend a month rebuilding something you already had.

What Notion is bad at is visual, spatial work. Boards where the arrangement carries meaning. Moodboards, story maps, research walls, anything where you need to see forty things at once and move them around. In Notion these become galleries of cards you open one at a time, and no amount of database configuration fixes that, because the underlying model is a list of pages, not a canvas.

So the honest question is not "should I leave Notion". It is "which half of this project was never suited to a page-based tool". Answer that first and the migration gets much smaller and much safer.

How AI fits across the six steps

Narrow but real. AI helps with step 2 (inventory) and step 6 (rewiring), and should be kept away from step 1.

In the inventory step, pointing a model at an exported workspace to classify pages into the four buckets is exactly the kind of tedious sorting it does well, and it is faster than clicking through 200 pages. In the rewiring step it can find broken references and draft the replacements.

Step 1, the diagnosis, is the one to do yourself. A model asked "should I move out of Notion" will produce a confident, balanced, generic answer. It does not know that the reason your workspace feels broken is that two people disagree about what a "project" is, which is an organisational problem that will follow you into whatever tool you pick next.

The practical constraint worth noting: after the move, whether your new tool's AI can actually see the whole board matters more than it did in Notion, because visual work is exactly where a text-only AI has the least to work with. An AI that reads a canvas including how things are arranged is doing something a chat panel reading pasted text cannot.

1. Diagnose what is actually broken

Do not skip this. Migrations that begin with "Notion is slow and I hate it" end with the same workspace in a different tool.

Write down the last five times the workspace annoyed you. Then sort each into one of these:

If most of your symptoms land in the "no" rows, the migration will not help and you should fix the process instead. If they land in the "yes" rows, the visual and spatial layer is what needs to move, and probably only that.

2. Inventory before you export

Open the workspace and list every top-level area, with one line each: what it is, who touches it, and which of the four categories it belongs to.

The four categories:

1. Visual and spatial. Moodboards, reference walls, story maps, anything where arrangement matters or where you need many items in view at once.
2. Structured records. Anything that is genuinely a table: a shot list, a contact list, a budget, an episode tracker. Rows with consistent fields.
3. Long documents. Treatments, scripts, briefs, notes, meeting records. Prose that is read top to bottom.
4. Tasks. Things with an owner and a date.

Two rules while you sort. First, be strict about category 2: many things that live in a Notion database are not really tables, they are a pile of pages someone put in a database because that was the available container. Second, anything you have not opened in six months goes on an "archive" list and does not get migrated at all. Migration is the cheapest moment you will ever have to delete things.

This is where an AI pass over the exported page titles and first paragraphs saves real time on a large workspace, though check its category calls, because it tends to over-assign to "structured records".

3. Export properly, and know what you lose

In Notion, use Settings, then the workspace export. Choose Markdown and CSV, and set the content option to include everything so attachments come with it rather than remaining as links to files you will lose access to. Export the whole workspace rather than page by page unless it is genuinely enormous.

Then set expectations, because several things do not survive and it is better to know now:

• Database views do not export. You get the underlying data as CSV. Every board, calendar, gallery, filter, and sort you configured is gone. If a view is load-bearing, screenshot it before you export so you can rebuild it deliberately.
• Internal links become local file paths. Every link between pages now points at a file on disk, which means that once the pages are elsewhere, the links are broken. This is the single largest source of post-migration damage and step 6 exists for it.
• Attachment filenames get mangled. Images and files come across, but names are often hashed and the association to their page can be loose. For a moodboard this is the difference between forty-one useful images and forty-one anonymous ones.
• Arrangement is not a thing Notion stored. This is the one people find hardest. If a gallery looked meaningful, that was the gallery's rendering, not saved positions. There is nothing in the export that says which image sat next to which.

Because of that last point, do one thing before exporting: screenshot every visual page at full length. Those screenshots are your only record of arrangement and you will use them constantly in step 5.

4. Send each category somewhere it fits

Now place the four piles, deliberately.

Long documents. Honestly, consider leaving these in Notion. It is a good document editor and there is no prize for a clean break. If you are moving them, move them to a document tool, not a canvas.

Structured records. These want a database or a spreadsheet. If Notion databases were working for you, leave them. If the records need to be visible next to the visual work, a canvas that supports tables can hold the small ones, but do not migrate a 900-row tracker onto a canvas.

Tasks. Into whatever the team already uses. Tasks are the category people most often try to bring onto a canvas, and it is reliably a mistake: a task without a due date and an assignee in a system that chases them is a wish.

Visual and spatial. This is the pile that motivated the migration and the one that benefits. It wants an infinite canvas, where arrangement is preserved, many items are visible at once, and moving something is how you think rather than an edit to a record.

For that last category specifically, the thing worth checking in whatever tool you choose is whether its AI can read the board itself. Post-migration, the visual layer is where you have the least text for a chat-based assistant to work with, so an AI that only reads what you paste into a sidebar is close to useless on exactly the material you moved. Storyflow was built around the opposite behaviour: the AI reads the whole current board, including the arrangement, and you can pull in up to one Tactic and three documents by mentioning them. It is free to start with unlimited boards, images, and collaborators, and paid plans begin at 7.99 dollars a month annually, priced per account rather than per seat, which matters when you are inviting a whole team into a workspace you are still rebuilding.

5. Rebuild the visual layer first, and rebuild it by hand

Start with the visual pile, because it is the reason you moved and because finishing it gives you a working system fastest.

Rebuild it manually. This sounds like bad advice and it is the right advice. Bulk-importing forty images produces a grid in upload order, which is arrangement noise, and you will spend longer undoing it than you would have spent placing them. Open the screenshot from step 3, place the images to match, and use the pass to prune. On every moodboard I have rebuilt, roughly a third of the images turned out to be redundant once I was forced to look at each one.

Order of work inside this step:

1. Recreate the page structure as named regions on the canvas, before adding content.
2. Place the visual material, working from the screenshots, densest board first.
3. Add the small structured pieces that need to sit beside the visuals.
4. Only then link out to the documents and tasks that stayed elsewhere.

Do one board completely rather than all boards partially. A single finished board tells you whether the new arrangement works before you have committed a week to it.

6. Rewire the links, then archive

The export broke every internal link. Find them and fix them, because a project with dead cross-references decays quickly.

Work through the moved material and replace each broken reference with either a link to the new location or, where the target stayed in Notion, a link to the live Notion page. Mixed systems are fine as long as the links between them are real.

Then archive rather than delete. Keep the old workspace read-only for at least a full project cycle. You will need it, usually for something small and specific that nobody thought to migrate, and the cost of keeping it is nothing compared to discovering in month two that a client's original brief only existed there.

Finally, write one short page in the new system that says where each of the four categories now lives. Every migration I have seen without this note ended with half the team still working in the old place, which is the worst of both.

Common mistakes I see

• Migrating everything. The correct outcome is usually a split. Moving your documents and databases out of Notion because you needed a canvas for moodboards is how a two-day job becomes a month.
• Skipping the diagnosis. If the problem is that nobody updates the workspace, a new tool changes nothing except which tool nobody updates.
• Not screenshotting before export. Arrangement is not in the export. Once it is gone, it is gone, and for a moodboard the arrangement was most of the value.
• Bulk-importing images. Upload order is not meaning. Place them by hand and prune while you do.
• Deleting the old workspace. Archive it. Something small and important is in there.
• Rebuilding every board halfway. Finish one. It tells you whether the plan is right while changing course is still cheap.
• Leaving the links broken. A system full of dead references stops being trusted, and once it is not trusted people stop maintaining it.

How this went for a documentary rebuild

The workspace was three years old and held a feature documentary: a research pile, an interview tracker, a treatment, a moodboard, and a task database. Six contributors, most inactive.

The diagnosis was blunt. Four of the five annoyances were about the moodboard and the research wall, both of which were galleries you had to open card by card. One was about nobody updating the interview tracker, which was a process problem and stayed a process problem.

So the split was: moodboard and research wall moved to a canvas, the treatment stayed in Notion because Notion is a perfectly good document editor, the interview tracker stayed a database, and tasks moved into the tool the producer already lived in. Roughly a third of the workspace moved.

Export took ten minutes. Screenshots took forty, and were the highest-value forty minutes of the whole migration. Rebuilding the moodboard by hand took most of a day and removed 38 of 104 images as redundant. The research wall, rebuilt as clustered regions rather than a flat gallery, immediately surfaced two themes that had been invisible when they were rows.

The part that took longest was rewiring links, about three hours, almost all of it because the treatment referenced research pages by internal link and every one of them had to be repointed.

The tools you will actually use

• Notion, still, for the documents and the databases. Leaving is not the goal; putting each thing where it belongs is. Notion remains one of the best document and wiki tools available.
• A canvas for the visual and spatial pile. This is the whole reason for the migration. Storyflow is built for exactly this material and its AI reads the full board, which matters most for visual work; Milanote is a strong alternative with a similar creative-boards focus; Miro is the better pick if your primary need is live multiplayer workshops rather than a persistent project workspace.
• Your existing task tool for tasks. Do not invent a new one during a migration.
• A spreadsheet for any large table that does not need to be in a fancy container. Plenty of Notion databases are spreadsheets that were promoted above their station.

You are ready

The migration that works is smaller than the one you were planning. Diagnose first, because half the reasons people leave a tool are process problems that travel with them. Split the workspace into its four real parts and move only the ones that are genuinely in the wrong kind of container, which for creative work is nearly always the visual and spatial layer.

Screenshot before you export. Rebuild arrangement by hand and prune while you do. Repair the links. Archive rather than delete, and write down where everything now lives.

Do that and you end up with the visual work on a surface that suits it, the documents and databases still where they were always fine, and a project you can see the shape of again.

Author

Justkay is a documentary filmmaker and the founder of Storyflow. He has moved several multi-year documentary workspaces between tools, and built Storyflow after concluding that the visual half of a creative project had never had a good home in a page-based app.