1.3Filters, sorts & linked views
The essentials in 30 seconds
- Filter = show, sort = order: the filter only keeps the rows matching a condition («Status = In progress»), the sort orders them (by due date, priority).
- Specific to each view: each view has its own filters and sorts — «My tasks for the week», «Overdue projects» on the same database.
- Linked views, zero copying: display the database elsewhere (dashboard) with its own settings — one dataset, one place, several views.
Running mini-case — Lina steers without copying in Oran
Lina's «Projects» database is growing and her team is getting lost in it: everyone copies rows into their own table, and versions diverge. Lina creates three views — «This week» (Due date within 7 days, sorted by date), «My projects» (Owner = self, grouped by Status), «Overdue» (Due date passed AND Status different from Done) — then displays the database on the «Dashboard» page via a linked view. A single source, visible everywhere: personal copies disappear.
Continued from cases 1.1-1.2 — Lina turns her database into dashboards.
1. Filtering and sorting: making data actionable
A database becomes useful when you filter and sort it. A filter only shows the rows matching a condition («Status = In progress», «Due date = this week», «Owner = me»). A sort orders the rows (by due date, by priority). Combined, they turn a mass of data into actionable dashboards: «My tasks for the week», «Overdue projects».
Lina filters «Due date within 7 days» and sorts by date: her «This week» view tells her every morning where to start.
2. Specific to each view: multiply angles, not data
Key point: filters and sorts are specific to each view. You create several views of the same database, each with its own settings, without duplicating data. Editing a task in one view edits it everywhere, because there is only one source.
Lina corrects a due date in «This week»: the «Overdue» view and the Kanban follow immediately — one dataset, a single place.
3. Linked views: the same source, displayed elsewhere
Even more powerful: linked views display a database elsewhere in Notion (another page, a dashboard) with their own filters, without duplicating data. A single source of truth, displayed wherever useful: the end of copy-paste and out-of-sync versions.
Lina inserts a linked view of «Projects» on her «Dashboard» page: her manager sees progress without ever touching the source database.
The graveyard of copies (never do this)
Moral: copying is divorcing the truth — linking is staying in sync.
🔎 Concrete example — 🔎 Concrete example — Three useful views
On the same task database.
«This week» : Due date ≤ 7 days, sorted by date
«My tasks» : Owner = me, grouped by Status
«Overdue» : Due date < today AND Status ≠ Done
→ linked view on the Dashboard page.Reading: three filtered windows on the same Lina projects.
Lesson: one dataset, a single place, several views.
Section vocabulary
- Filter
- A condition that hides rows that do not meet it (e.g. Status = In progress); specific to each view.
Simple example: The «Due date within 7 days» filter of Lina's «This week» view.
What it is not: deleting the other rows from the database.
Reflex: one clearly named filter per use.
- Sort
- The display order of rows by one or more properties (by date, priority…); specific to each view.
Simple example: The sort by due date that orders Lina's projects from most urgent to most distant.
What it is not: a filter that hides rows.
Reflex: sort by what drives action (date, priority).
- Filtered view
- A database view configured with filters/sorts for a specific use («My tasks for the week»).
Simple example: Lina's «Overdue» view (Due date passed AND Status different from Done).
What it is not: a duplicated database to «make a filtered version».
Reflex: one view per question: what this week? what is overdue?
- Linked view
- A display of an existing database on another page, with its own filters, without duplicating data.
Simple example: The «Projects» database displayed on Lina's «Dashboard» page.
What it is not: an independent copy that will diverge from the original.
Reflex: displaying elsewhere = linked view, never a copy.
- Single source of truth
- Principle: a dataset exists in only one place, displayed everywhere via views; avoids out-of-sync copies.
Simple example: Lina's projects, entered once and visible in Kanban, Calendar and dashboard.
What it is not: five personal copied tables that no longer say the same thing.
Reflex: copy spotted = linked view to create.
In practice — In practice — Building filtered and linked views
- On your task database, create a «This week» view: Due-date-within-7-days filter + sort by due date.
- Create a «My tasks» view: Owner = you filter, grouped by Status.
- Create an «Overdue» view: Due date < today AND Status ≠ Done filter.
- On a «Dashboard» page, insert a linked view of this database to display it without duplicating it.
Ready-to-say script — stopping diverging copies (30 s)
«Stop copying the table: I'll create a linked view on the Dashboard page with your filters. You'll see the same source as everyone, updated in real time — and every correction is made in a single place.»
To say to a colleague who maintains «their» copy of the table.
Copy-paste template — view plan
SOURCE DATABASE: _______________
VIEW 1 «This week»: filter _______________ + sort _______________
VIEW 2 «My tasks»: Owner = me filter, grouped by _______________
VIEW 3 «Overdue»: filter _______________ AND _______________
LINKED VIEW on page: _______________ (own filters: _______________)
COPIES TO DELETE: _______________Team that copies everything
List the personal copies, replace them one by one with linked views filtered by Owner: everyone keeps «their» vision, on the single source — like Lina in Oran.
Weekly steering
Monday ritual: open «This week» sorted by due date + «Overdue» — ten minutes to prioritise, with no new table.
Management dashboard
A «Dashboard» page with two linked views (Kanban by Status, «Overdue» list): management follows progress without touching the working database.
Genuine template reuse
Only legitimate duplication case: restarting from the structure for a distinct dataset (e.g. new year, new project) — otherwise, linked view.
✓ Best practices
- One named view per use.
- Linked views for dashboards.
- One dataset, one place, several views.
✗ Common pitfalls
- Duplicating to «filter».
- Linked view confused with copy.
- Diverging personal copies.
Key takeaways
- A filter only shows the rows matching a condition; a sort orders them.
- Filters and sorts are specific to each view: several views of one database, with no duplication.
- Linked views display a database elsewhere with their own settings.
- Result: a single source of truth, displayed everywhere — the end of out-of-sync copy-paste.
48h challenge — apply it tomorrow
Create your 3 views («This week», «My tasks», «Overdue») on your database + 1 linked view on a «Dashboard» page. Then delete one copy. 25 minutes is enough.
- I can explain filter (show) vs sort (order), specific to each view.
- I know when to duplicate (template, new dataset) vs link (same source).
- I have 3 views + 1 linked view + deleted one copy.
Frequently asked questions
If I create several views, do I duplicate my data?
No, never — and that is the whole point. A database holds the data only once; views are just windows filtered/sorted on that same data. Editing a task in one view edits it everywhere, because there is only one source. It is exactly the opposite of the «I'll copy my table to make a filtered version» reflex, which creates out-of-sync duplicates. In no-code as in databases, the golden rule is: one dataset, a single place, several views.
What is the difference between a linked view and a duplicated database?
A linked view points to the source database: the data stays unique and synchronised; you see the same information elsewhere, with filters suited to that place. A duplicated database creates an independent copy: from then on, both evolve separately and diverge. Use the linked view whenever you want to «show the same thing differently/elsewhere» (dashboard, project page). Only duplicate a database if you really want to restart from a template for a distinct new dataset.
If I create several views of the same database, do I duplicate the data?