[npunt]

I tried writing one of these laundry lists of how screwy and frustrating Notion's UX is, from the information architecture (mixing page, project, account, and presentation options) to the writing experience/feel to the weird block-centricity... and I just gave up in frustration.

The conclusion I came to is if you're going to remake and extend the word processor, you must take the lessons of the last 40 years to heart. There are in-built expectations for word processors, including keyboard commands, selection & cursors, speed, permanence & permission, etc, and Notion violates many of them for no good reason. The result is an unpleasurable writing environment. That should have been the easiest thing to get right, the foundation upon which amazing new features are built.

I use Notion for a few things, mostly because existing tools have yet to move from document-centric to project-centric designs, but it is not a tool that I would foist on the average user nor one I especially enjoy using. So much potential is held back by bad UX, which I think a lot of people don't see because Notion does have good graphic design and a few really stellar capabilities.

My impression is the team has probably concluded their rapid growth is a validation of the UX, rather than other positive factors of Notion, namely being the first halfway decent project-centric editor. Hoping they do a UX refactor soon.

[jitl]

(I work at Notion)

We didn’t conclude anything as a validation of our UX - there is nothing in Notion that we consider “ideal” or even “finished”; we are painfully aware of how far we still have to go. There is just a lot of work to do for our small (but growing!) team.

[npunt]

Thanks for responding. I imagine many things are growing and changing there, and it's great to hear the team retains that kind of humility and self awareness. In our industry it's pretty common for companies to let success get to their heads and not do the kind of critical examination necessary to overcome the momentum of early missteps or treat early temporary decisions as permanent. Having run product/design for a few startups I'm extra aware of this since I've been guilty of it myself.

I think the part of Notion that most gives me this impression is the information architecture, which can be an extremely difficult thing to change once a significant number of users get to know a particular layout, and hasn't meaningfully changed since launch. Of all the writing tools I use, Notion's IA has always been the least straightforward.

Between the '...' menu and the Settings & members section, as well as several options like favorite, share, updates, and options on the sidebar, there's a lack of clear hierarchy and separation between account, workspace, page, or presentation options. This lack of clear hierarchy, combined with some product choices that apply at an unexpected level of hierarchy or permission model, means many actions exact capabilities are unclear or violate expectations of where they apply - account v workspace, workspace v page, and all users v me.

Take the top bar for example. From left to right:

1. Breadcrumbs: workspace & page

2. Share: workspace-specific (where can I share just an individual page? why is this separate from Copy Link?)

3. Updates: page-specific

4. Favorite: page-specific that either applies to the workspace or is a personal favorite (unclear, given "favorites" typically = personal, while "pin" = others see it)

5. '...' Menu: contains mostly page options with a few workspace ones (e.g. Import, possibly Export?). I even got confused about whether style options are personal (e.g my reading preference for font) or apply to the page style, given many writing tools allow this kind of style configurability as a user setting.

There's also duplication of options around the UI including Import, Favorite, Move To, New Page and Updates (how is this different than Page History?), which further leads to confusion about where to take what kind of action and clutter the interface.

The sidebar mostly makes sense as its all workspace-related. However, new users are likely to get tripped up on the difference between workspace name/icon and the root page name/icon, but this is a minor point.

Over in the Settings & Members section, you also have a mix of options that don't really meet expectations. The menu itself is launched from the sidebar, where workspace options are, but the menu also contains account settings, which is higher up in the hierarchy.

As far as product choices, one would expect notifications and connected apps to be workspace-based, as they are in other popular collaboration tools that use the same workspace patterns (like Slack/Discord), given that some workspaces I want to be watching closely or use particular accounts and others I don't. But these are instead account-based.

Logging in and out of workspaces is also in an entirely different section (the workspace picker), which is quite different from most apps where you tap your user icon and make account choices. Also in Settings & Members is Dark mode, floating on its own not even under account settings. Are there other personal presentation options? What if I have accessibility needs or want a bigger reading font?

The overall impression I have of Notion's IA is that it minimizes UI footprint to a point of lack of clarity and that its structure is buckling under a heavy feature load. This is a signal of graphic design having priority over UX. Rather than express clear hierarchies in layout, they're bundled together to save a bit of space (e.g. topbar) or to minimize the number of maintained pages (e.g. a dedicated Home / workspace page, account management page, global activity page, etc). These are the types of things that bite back quickly as you scale features and expand audiences.

For an example of a product that really nails information architecture, look at Trello. It's got a really clean UI separation between the hierarchies of cross-Board features (top nav), within-Board (sub nav), List (main canvas), and Card (modal). You always know where to find each action and what each does in their context. Coda is also halfway decent, if a bit feature-rich and ugly. I also wrote about my own experience refactoring OneSignal's IA to manage future feature growth here: https://nickpunt.com/blog/when-to-add-hierarchy/