A Bookmark System That Survives: Fewer Categories, Clearer Rules
Why seven categories is the practical ceiling, how to categorise by moment of use rather than topic, and using click counts to decide what to delete.
Collections grow by default: saving costs nothing, finding costs more every month. There is no shortage of tidy-up methodologies; most die within a quarter because they require you to get it right every single time.
The rules below are designed to hold without discipline — the structure does the work.
Seven categories is the ceiling
People can scan five to seven categories at a glance. Past that, choosing a category becomes a decision of its own, so you pick one at random and the system stops meaning anything.
A workable starting set:
- Daily (pinned): the five to eight you open every morning
- Tools: converters, lookups, generators
- Docs: specs, manuals, internal wiki
- Reading: long articles, blogs, columns you have not finished
- Assets: design references, palettes, icon sets
- Snippets: prompts, commands, templates
- Uncategorised: a temporary inbox, emptied monthly
If your work has clear project boundaries, "project" is the one dimension allowed to grow — archive it when the project ends.
Categorise by moment of use, not by topic
This is the mistake almost everyone makes.
Topic categories (frontend, backend, design, ops) sound right, but when you search you are not thinking "this is a frontend thing," you are thinking "I need to write a SQL query right now." So the better axis is the situation:
| Weaker | Stronger |
|---|---|
| Frontend / Backend / Design | When coding / When designing / When writing |
| Study material / Work material | To read / Frequently checked / Archived |
| Domestic / International sites | Daily / Occasionally |
The test is simple: what is the first word in your head when you open it? That word is the category name.
Pin sparingly — five at most
Pinning means "bypass the ordering rules, always first." It is effective, which is exactly why it gets abused: pin ten and you have pinned nothing.
Rule: no more than five pinned items, each clicked at least weekly. Anything untouched for two weeks gets unpinned and returns to natural click ordering.
Same for categories. Usually only "Daily" deserves a pin; everything else can rely on manual order.
Use click counts to decide what to delete
Knowing what to remove is the real difficulty. Click count is an objective signal:
- 0 clicks in three months, and not reference material (docs, specs) → delete
- 1–2 clicks, with a more-used alternative in the same category → delete
- High clicks but expired content (pricing pages, campaign pages) → delete, or replace with a stable entry point
Reference material is different: a spec you consult twice a year may be critical each time. Ask "if I needed this now, could I find it again within a minute?" If not, keep it.
Naming matters more than categorising
In three months you will not remember why you saved something, but the title will be right there. Three rules:
- Start with the purpose. "Expense system" beats "Acme OA"; "Regex — strip HTML tags" beats "Stack Overflow."
- Include the qualifier. "Figma plugin — batch rename layers" beats "a plugin."
- Use a consistent prefix per project or person, so one search returns the whole set.
Tags work for lateral links, but do not run more than three tag vocabularies or you will forget which one you used.
The ten-minute monthly routine
Do it on a fixed day, in this order:
- Open "Uncategorised" and file or delete each item. This is where most of the value comes from.
- Look at the bottom ten by click count and cut what you no longer need.
- Check whether the pinned items still deserve to be pinned.
- Export a JSON backup.
- If a category exceeds thirty items, split it or move half into "Reading" as pending.
Step 4 is what makes the routine possible: backup makes deletion safe, and only safety lets you delete.
Batch operations remove most of the tedium
Re-categorising one card at a time is the slow part. Use selection mode:
- Enter batch mode,
Shift + clickto select a run, move them all at once; - Pin a batch of entries for one project;
- Delete a batch of expired campaign pages;
- Export a selection as a project handover list.
Deletion comes with a five-second undo, which keeps the cost of a mistake low — and that is the precondition for cleaning anything at all.
FAQ
Topic or situation — which should categories use?
Situation. Topic categories force a classification step before you can search; situation categories map directly to what you are doing.
Is it normal to save things and never open them?
Completely. That is the default outcome of saving. The difference is whether you have a cleaning routine: with one, your collection is what you currently use; without one, it is a graveyard of things you meant to read.
Should I tag bookmarks?
Only if you genuinely search across tags later. Tags cost more to maintain than categories, and two systems usually means both fail. Text bookmarks benefit most (sql, prompt), because their content is otherwise hard to search.
The goal is not a tidy-looking collection. It is that when you need something, you do not have to reconstruct your past reasoning to find it.
Keep reading
Open MyNav
No sign-up. Open it and start.