Panel is where your studies live. Browse everything you've run, open a study, and explore the results from the headline number all the way down to a single panelist's reaction.
Study links keep their brand in the URL. Opening a link selects that brand and checks your access before loading its results. The header and study details show the same brand. You can copy the study address to keep this context; a brand in a link never grants access. Older links without a brand use the selected brand. If the linked brand is unavailable, try again or return to Panel. Brand Settings asks before a brand link discards unsaved edits; section changes in the same brand keep your draft.
A study link carries its brand across tabs and sessions.
Favor-count labels, response counts and percentages sit above full-width bars on narrow screens. Switch between Bar and Pie to inspect the same parsed preferences. Demographic bars also stack beneath their labels and counts, with long labels wrapping instead of squeezing the chart.
The same counts and percentages stay readable on a phone.
Open a study in Panel to see its name, brand, audience, status, and study actions together. Saved labels appear separately from the study title. A/B studies lead with a purple summary card: the verdict, option probabilities, consensus, response count, and favor count. Favor count shows how many parsed responses preferred each option; its percentages use the parsed-response total. Switch between Bar and Pie to inspect the same responses.
The headline result comes first; each analysis section has its own card.
A Pulse can finish with only part of its panel when some evaluations fail. If fewer than 80% of the respondents you asked for answered, the summary card gives no verdict: it reads Too few respondents for a verdict, says how many answered out of how many asked, and highlights neither option. The probabilities and favor count still show what the people who answered said. Run the Pulse again for a verdict.
Ten of twenty answered, below the sixteen a verdict needs, so the card names no winner.
Enabled custom persona attributes describe the respondent. Relevant answers should reflect their assigned interests and habits, including direct questions about sports fandom. Every assigned value is included before optional descriptions. Pulse segments also consider whole custom cohorts when their answers are coherent. The minimum segment size is three people or 5% of completed respondents, rounded up, whichever is larger. Segments defined by distinguishing custom attributes have no maximum panel share; other segments remain capped at 40%. Answer similarity, overlap and separation checks still apply, so matching attributes alone do not guarantee a segment. Updated size rules apply when existing results are next opened; the original answers stay unchanged. Rerun after an answer-generation behaviour change to collect new answers.
The minimum scales with completed panel size. Custom-attribute segments can exceed the 40% limit for other segments.
Segment cards put What sets them apart in a short bulleted list. Each defining attribute compares the people in the segment with its share of the full panel. The option score table compares This segment with Panel, followed by answer similarity, distance and people counts. Open Read full summary for the original narrative. Older results without attribute highlights or option scores show their summary directly. Saving and rerunning a segment keeps its original context.
Scan the defining attributes and scores, then expand the full narrative when needed.
The section navigation follows the sections available for this study. On desktop, collapse it to an icon rail when you want more room for charts, then expand it to see the names again. On a phone, scroll the section chips and tap a destination. Distribution, evidence, segments, option spotlight, details, demographics, notes, and persona interviews appear when they are available to you.
Demographic filters and persona interviews keep the same audience controls in the card layout.
In Persona interviews, custom attributes disabled for a Pulse show Disabled for this pulse in both the respondent list and the selected 1:1 interview card. Muted chips with a dashed border distinguish them from enabled attributes. Their values stay visible for reference and result analysis, but the disabled attributes and their persona seeds are omitted from evaluations and follow-up interview context. Disabled profile overrides appear only in their labelled chip. The disabled label remains even if the attribute no longer has a current value or has been removed from brand settings. The status follows the Pulse's saved settings, including inherited rerun settings.
Disabled custom attributes keep their reference values and clearly show their status in both interview views.
Each Details field has a matching icon: a calendar for Created, a tag for Brand and Label, people for Respondents and Audience, and a clock for timings. Available study actions and chart controls also pair icons with their text labels. On a phone, Details fields stack into one column to keep icons and values readable.
Metadata icons help you scan the study setup without replacing its labels.
Click a section heading to collapse or expand its card, or use Enter or Space on the keyboard. Each card keeps its chart choices and filters when closed. Selecting a destination in On this page or the phone section chips reopens it before jumping there. Cards and tiles enter as you scroll, graph bars grow from the left, and numeric metrics count up from zero. Reduced-motion mode shows the finished values and charts immediately.
Section headings toggle their cards independently; navigation opens a collapsed destination.
A result is not one number. Every panelist behind it kept a full profile, so the same study can be re-read by any of ten dimensions — and a cohort isolated on the seven targetable ones becomes the audience of your next run. Five of those dimensions are census-weighted (age, gender, race/ethnicity, region and income); education, household, political lean, urbanicity and primary device are modelled onto the panel rather than drawn from the census allocation, and the grid marks them. (Political lean, urbanicity and primary device slice the result but cannot be targeted, so they do not seed a rerun.) Persona attributes your brand declares in Brand Settings join them as further dimensions, read with your own level names. This walkthrough takes one A/B result apart end to end.
The list shows every study for your brand with its type, status, audience, and a score preview. Search by name, label, keyword, segment, or audience, and filter to find what you need. Studies still running show live progress.
Each segment leads with the option it favoured and the scores behind it. Answer similarity is how alike the group's answers are, not how many agreed; Distinctness grades how separate the group is, not how much it liked the options.
Open any panelist to read their original reaction and ask follow-up questions — "why did you react that way?" — to understand the reasoning behind a response.