All topics
22 / 55
Analyticsintermediate

Filters, Prompts & Presentation Variables

Distinguish column filters, dashboard prompts, and presentation variables, and explain how they interact to create dynamic, interactive dashboards.

Three mechanisms make a dashboard interactive, and the question is how a user's choice reaches the queries.

A dashboard prompt is like a thermostat dial in a building's lobby: one dial (the prompt) broadcasts a single target temperature (the presentation variable) that every vent (each analysis) in the building reads and adjusts to, instead of someone manually resetting every single vent by hand.

Key Concepts

1
    filter                 built into an analysis; always applies
    dashboard prompt       a control on the dashboard; the user chooses
    presentation variable  carries the chosen value to wherever it
                           is referenced
2
How a prompt reaches an analysis, by default. A prompt on Region affects every analysis on the page that includes Region as a filter set to "is prompted".
3
    analysis filter:  Region is prompted
    -> the prompt's value is applied
    -> no prompt value chosen: no restriction on Region
4
"Is prompted" is the wiring. An analysis without it ignores the prompt entirely, which is the usual reason a prompt "does not work".
5
Presentation variables are for everything else. A prompt can set a variable instead of filtering directly, and the variable is then referenced anywhere:
6
    prompt sets:  pv_region
    used in a filter:   Region = @{pv_region}
    used in a title:    Sales for @{pv_region}
    used in a formula:  CASE WHEN Region = '@{pv_region}' THEN ...
7
That is what lets a user's choice appear in a report title, or drive a calculation rather than a filter.
8
Always give a default. @{pv_region}{All Regions} supplies a value when nothing has been chosen — the part in braces. Without it, the first render has an unresolved variable and either errors or filters on nothing.
@{pv_region}{All Regions}
9
The variable types, which get confused.
    presentation variable  set by a prompt, scoped to the dashboard
                           session
    session variable       set at login, per user, from the RPD
    repository variable    one value for everyone, refreshed on a
                           schedule
10
A presentation variable is the user's current choice. A session variable is who they are. Using one where the other belongs is a common bug — particularly using a presentation variable for security, which the user can change.
11
Security never goes in a prompt. A prompt is a user control, so any filter driven by one is a filter the user can alter. Row-level security belongs in the RPD, applied by the server.
12
Prompt performance. A prompt listing values runs its own query, so a dropdown over a million-row dimension is slow every time the page loads. Constrain it, or use a search-style prompt rather than a full list.
13
What the interviewer is probing.1. "What is the difference between a filter and a prompt?" Probing: the distinction. Stalls: "Nothing." Moves up: a filter constrains the query; a prompt collects a value from the user that a filter or variable then uses.
14
2. "What is a presentation variable for?" Probing: the mechanism. Stalls: "Filtering." Moves up: carrying a prompted value into column formulas, titles and conditions — not only into the filter.
15
3. "Why is the is-prompted filter operator useful?" Probing: the practical idiom. Stalls: "It is not." *Moves up:* it lets one analysis respond to a dashboard prompt without hard-coding which column, so the same analysis works on several dashboards.
16
4. "What is the performance trap with prompts?" Probing: experience. Stalls: "None." *Moves up:* a prompt listing all values of a large column runs an expensive distinct query on every page load — constrain or limit it.
Use cases:promptspresentation-variablesdashboard-interactivity