Cascading Filters
Let one filter narrow the choices in another, control the direction, and understand what viewers see.
Cascading narrows the choices in one filter based on selections in others. Pick United States in Country and the City list shows only US cities. It never changes which cards a selection filters; that is card mapping.
Cascading is on by default for regular dashboard filters. Most dashboards need no setup.
The three modes
Open a filter's settings, then set Narrow available options by on the Basic tab.
| Mode | Effect | Use when |
|---|---|---|
| All applicable filters (default) | Every related filter in scope on the current sheet narrows this list. Unrelated filters are ignored. | Related filters should narrow each other in both directions. |
| Selected filters | Only the filters you check narrow this list. | You want a fixed parent, or a one-way flow. |
| None | Other selections never narrow this list. | The list should always show every choice. |
With All applicable filters, Country narrows City and City narrows Country. Filter order in the toolbar does not create a parent-child relationship.
Make a cascade one-way
To have Country narrow City without City narrowing Country:
Set Country to None
Open Country's settings and choose None under Narrow available options by. Choose Accept.
Point City at Country
Open City's settings, choose Selected filters, and check Country. Choose Accept.
Both filters still apply their selections to their configured cards. Only the choice lists changed.
Selected filters picks which selections to consider. It cannot connect data that is not related; see When narrowing needs a relationship.
What viewers see
| Situation | The menu shows |
|---|---|
| Another selection narrowed the list | Narrowed by Country above the options. |
| No choices remain | No options match the current filters. Change the related selection to see other choices. |
| A selected value no longer matches | A note that the value has no matches under the current filters. The value stays selected until the viewer changes it. |
| A selected dependency cannot be applied | The reason, and for supported cases a Show options without Country action. |
Selections are never cleared for the viewer. If Paris is selected and Country changes to United States, Paris stays selected and keeps filtering its cards until the viewer replaces or clears it. A value missing from the visible page of choices is not treated as invalid; menus search and load more on demand.
Show options without Country browses the broader list for that dependency only, labels the result, and offers Undo. It does not change the selection, the cards, or the saved dashboard, and it never bypasses permissions or covers a database failure.
When narrowing needs a relationship
Semaphor narrows only along relationships it can prove.
Connect filters through a fact table
Two filters can only narrow each other along a relationship Semaphor can prove. Sometimes that relationship runs through a third dataset, and then you decide which one. This section walks through the setting Connect filters through, found on the Basic tab of a semantic filter under Narrow available options by.
A worked example
A retail domain models three datasets:
| Dataset | Kind | Columns | Related to |
|---|---|---|---|
| Stores | dimension | store_id, store_name, region | Sales |
| Categories | dimension | category_id, category_name | Sales |
| Sales | fact | store_id, category_id, amount | Stores, Categories |
Stores and Categories are not related to each other. A store does not have a category; only its sales rows do. The dashboard has a Store filter whose choices come from Stores and a Category filter whose choices come from Categories.
The viewer opens Category
Semaphor lists every category. No other selection is involved yet.
The viewer picks Store 12
Category is set to All applicable filters, so Semaphor checks whether the Store selection can narrow the category list. It looks for a relationship from Categories to Stores. There is no direct one; the only path runs Categories to Sales to Stores.
Semaphor stops rather than guess
Going through a fact table changes the question. "Categories related to Store 12" could mean the categories the store has sold (through Sales) or the categories it stocks (through Inventory, if that fact existed). Those are different lists. Semaphor leaves the category list unnarrowed and, in the filter settings, shows the Store filter as skipped with the reason that a connecting dataset is needed.
The author sets Connect filters through to Sales
Open Category's settings and choose Sales under Connect filters through. Choose Accept.
The list now narrows
Picking Store 12 narrows Category to the categories that appear in Sales rows for that store. The list means "what Store 12 has sold", and the author chose that meaning.
What the choices mean
| Choice | Effect |
|---|---|
| Directly related datasets only (default) | Only filters on the same dataset, or on a dataset directly related to it, narrow this list. Filters that connect only through a fact table are skipped and reported in the settings. |
| A dataset name | Filters that connect only through a fact table are linked through that dataset. The list offers the datasets in the same domain that this dashboard's filters already use. |
The setting appears only for filters whose choices come from a semantic dataset, because relationships are modeled in the domain. A filter on a direct database table has no relationship information; it narrows only by filters on the same table.
None and Connect filters through are independent. None ignores other filters' selections. A connecting dataset still limits the choices to values present in it: a Category filter connected through Sales offers only categories that have ever been sold, even with None. Choose Directly related datasets only if the list should show every category.
Custom SQL filters
A filter whose choices come from custom SQL is narrowed by its own query. The modes above do not apply; use the filter helpers in the SQL to include or omit other filters. See Filters in SQL.