Survey Team Targeting, Segmentation & Cross-Tabs

Team targeting lets an admin send a survey to a member-tag segment instead of only a hand-picked list of user ids. Filters resolve against member tags (the org's people-based segmentation model, assigned per member under Admin → Members), so an audience is defined by the tags members carry. Selecting multiple tags targets members with any of them.

Member tags power result segmentation after collection, tags are the org's people-based segmentation model, assigned per member under Admin → Members. A tag condition matches respondents who carry that tag. Cross-tabs extend segmentation to any question answer: single/choice responses segment by the option picked, multi-select responses by whether the selection includes an option, and scale/rating responses by a numeric band (respecting each question's scale minimum and maximum, so a 1 to 7 scale is handled correctly rather than assuming a fixed 5). Free-text responses cannot be used to segment.

A cross-tabulation matrix crosses two categorical dimensions. Each axis is a member tag or a single/choice question, and every cell reports the respondent count plus a row percentage and a column percentage, with row and column totals. Multi-select questions are excluded from the matrix because a respondent can fall in several columns at once, but every single/choice and multi question still gets a per-option frequency distribution within a segment (count plus percentage of the segment; multi-select percentages may sum above 100). Anonymous surveys hide cross-tabs entirely, since tags could re-identify a respondent.

• Open a survey and go to the Send to people tab. Use Target a segment to pick one or more member tags; a live preview shows the matching count and members before you send.
• Send to segment fans out to every matching org member with the same in-app notification and email behavior as a manual send. When filters match zero members, the UI says so plainly instead of reporting a misleading success.
• Re-send to an additional segment later: it adds new recipients without removing existing recipient rows or resetting their status, and members who already completed the survey are not re-notified (the Qualtrics "send to non-respondents" distinction).
• In the Stats tab, use Cross-tabs to build two segments (Segment A and Segment B) side by side. Each segment combines member-tag and question-answer conditions with AND, shows its own response count, per-question scale averages, and a per-option frequency distribution for every single/choice and multi question. A help button on the panel opens an explainer.
• Use the Cross-tabulation matrix to cross one categorical dimension against another. Pick a Rows dimension and a Columns dimension (a member tag or a single/choice question); the matrix shows respondent counts with row and column percentages plus totals.
• Export a filtered segment as CSV from the Cross-tabs panel; the same member-tag join and CSV shape as the full export is used, narrowed to the segment's matching responses.

Assign tags to members under Admin → Members; both send targeting and result segmentation resolve against those tags. Segments are empty until members carry tags.

If a tag filter matches zero members, the send button is disabled and the UI says so plainly instead of reporting a misleading success.

• Audience resolution is tenant-scoped: only members of the admin's organization are ever returned or sent to, and tag filters join through the org's own member_tags.
• Explicit user ids are validated against the admin's org before any send; a user outside the org is rejected.
• The audience preview, segmentation, and export endpoints all require admin access and check that the survey belongs to the caller's org. Cross-tab filter input is shape-validated server-side before it is applied to a query or export, and a cross-tabulation matrix axis is rejected unless it is a known member tag or a single/choice question in that survey.
• Segment export reuses the org-scoped response join, so an admin can only ever export their own org's responses.