Adding a filter
Filters live above the charts on the Results tab. Every chart on the page responds at once — you are not filtering one question, you are narrowing the population every chart is drawn from.
- 1Pick a question
Any question whose answers form a fixed set — see the list below for which qualify.
- 2Pick a value
The chart grid redraws immediately. The response count above the charts tells you how many people are left.
- 3Add another to narrow further
Filters combine with AND. Age 25–34 plus West gives you people who are both, not either.
Start with the demographics you didn't have to write
Surveys started blank or with AI Author get three demographic questions appended automatically — age range, gender identity, and ZIP or postal code. They are there precisely so this page has something to slice by on day one.
They behave like any other question here: they appear in the filter list, they combine with the rest, and if you removed them when building the survey they simply won't be offered.
What you can filter on
Anything where the answers form a fixed set the platform can offer you as options:
- Single select, multi-select, dropdown — Filter by any option, including options only a handful of people chose.
- Image choice — Filter by the image they picked.
- Grid — Filter by a specific row–column pairing.
- NPS, rating, linear scale — Filter by a specific score.
Not an oversight. “Contains, starts with, exactly, or fuzzy?” is a genuine interface question and a guess would be worse than the gap. For open text, the AI Analyst clusters it into themes instead — that is the intended route to a segment.
Scoping to one item in a looped question
If a question loops — the same follow-up asked once per thing the respondent picked — a separate chip row lets you scope the charts to a single item.
Without it, a “rate ease of use” chart for a looped question pools every item into one distribution, and a genuinely good item and a genuinely bad one average into an unremarkable middle. Scoping separates them.
Scoping narrows looped answers only. An ordinary question wasn't asked per item, so it keeps its full population — your demographics and your overall NPS still describe everyone, which is almost always what you want when you are asking “what did the people who used this one think?”
Speeders, straight-liners, duplicates and attention-check failures are left out of the charts before any filter applies. A toggle brings them back. Worth clicking once on any result you intend to act on — if including them changes the picture materially, that is itself a finding.
On a survey with anonymity turned on, filtering down past the reporting floor — five people by default — hides the charts rather than showing them. Not a bug and not something to work around: at that size, someone who knows the group can work out who said what. Remove a filter to widen the segment.
Sharing what you're looking at
Filters and item scoping are written into the page address as you set them, so a filtered view is a link. Send it to a colleague and they open the same slice, not the unfiltered page with instructions attached.
Exports follow the same rule: with filters active, an export gives you the filtered slice, and the export dialog tells you how many responses that is before you download. If you want everything, clear the filters first.