Most survey tools stop at drawing charts. The interesting questions start after the chart: is this different for managers? What do the unhappy respondents say in the open comments? Is that gap real or just noise? YourOpinion’s results page is built for those follow-up questions. Every answer is a filter, every question is a segment, and a statistics engine works through the comparisons you didn’t think to make.
[Image: The results overview page showing several question cards with charts, the Recommended splits bar at the top, and the channel and partial-response controls.]
The overview reads like your survey
Results are shown in the same order people answered them: your sections, your intro text, your questions, each with a live chart underneath. There’s no dashboard to configure and no report builder to learn. A new response lands and the numbers move.
Two controls sit above everything. A channel filter shows one distribution channel at a time, so you can compare the email campaign against the QR code on the poster. A partial-responses switch decides whether people who stopped halfway count; leave it off for clean numbers, turn it on when you want every scrap of signal.
Open-text questions appear as a quote wall: the full list of written answers, newest first. Quotes follow every filter you apply, which turns out to be the fastest way to read them with context - filter to the detractors and read what they wrote.
A chart for each kind of question
Question types differ, so their charts should too. A 1-to-5 agreement scale and a pick-your-country question have nothing in common statistically, and showing both as the same bar chart wastes what the data knows. The results page matches visualisations to each question type and lets you flip between the sensible options with an arrow control on every card. Your choice is remembered per question.
- Single choice questions offer horizontal bars, columns, and a donut paired with a distribution list. Crowded pies are quietly demoted; nobody needs a nine-slice donut.
- Yes/no questions add a diverging bar that puts the two camps on either side of a centre line.
[Image: A single-choice question card in the “Pie + distribution” view - donut chart with a colour-matched distribution list beside it. Smaller frames show the same question’s other views: horizontal distribution bars and vertical columns. The 1/4 view-switcher arrows are visible in the card header.]
- Agreement scales (Likert) lead with the median, the top-2 and bottom-2 box shares, and a labelled distribution; a diverging stacked bar shows polarisation at a glance.
- Numeric scales lead with the mean marked on the scale, plus top-2 and bottom-2 box, next to the full distribution.
- Number questions show mean, total, and response count.
[Image: An agreement-scale question card in the “Scale + distribution” view - median headline with top-2 box and bottom-2 box percentages on the left, labelled per-answer distribution bars on the right. A second frame shows the same question as a single diverging stacked bar, disagreement to the left of the centre line, agreement to the right.]
- Matrix questions collapse to a mean-per-row comparison, or a diverging bar per row for agreement rows - seven rows, one readable chart.
- Ranking questions show summed scores per option.
[Image: A matrix question rendered as “Mean per row” - one labelled mean score bar per row on a shared 1-5 scale, with the response count next to each row.]
A question linked to a KPI leads with its benchmarked score in front of the health bands, so the metric and the raw distribution live on the same card.
Click an answer to filter everything
Every bar, slice and label is a button. Click “Fully remote” and the entire page - every chart, every statistic, every text quote - now describes remote workers only. Click a second answer to widen the set, click answers on other questions to narrow it further; filters on different questions combine.
Each filter appears as a chip at the top of the page. A toggle on the filtered question flips it from including to excluding, so “everyone except the two people who said No” is one click, not a select-all exercise. Chips jump back to their question; the ✕ removes them.
The whole exploration state lives in the page URL. Copy the link mid-analysis and a colleague opens exactly what you see: same filters, same split, same charts.
[Image: A distribution bar chart with one answer selected (highlighted with a selection ring), the filter chip row above the results showing two active filters - one green “including” chip and one red “excluding” chip with struck-through value - and the Including/Excluding toggle on the filtered question’s card header.]
Any question is a segment
Every single-choice and yes/no question has a Split results button. Click it and the whole report is broken down by that question’s answers: choice questions become grouped bars per segment, scale questions gain a segment-means view that lines the groups up on the same scale. A toggle on each card flips between the split view and the overall numbers, and segments too small to trust are flagged as low n rather than silently plotted.
This is ordinary cross-tabulation - the analysis that usually hides behind a top pricing tier or an afternoon of spreadsheet work - with none of the ceremony. Demographics you collected become lenses on everything else - satisfaction by tenure, obstacles by department, trust by seniority - each one click, each one click to undo.
[Image: Two split views of the same report, split by work arrangement. Left: a choice question in the “Split distribution” view - grouped bars per answer, one bar per segment, with a per-segment legend showing counts. Right: a scale question in the “Segment means” view - one mean bar per segment on the shared scale, the smallest segment flagged “low n”. The Split/Overall toggle is visible in the card header.]
Recommended splits, tested before suggested
With twenty questions there are hundreds of possible cross-tabs, and nobody runs them all by hand. The analytics engine does. It computes every metric-by-segment comparison in your data and puts each one through significance testing with multiple-comparison correction, so a gap must be both large and statistically solid to survive. Only then does AI get involved: it reads the surviving findings and writes each one up in plain language.
The result is a short Recommended splits bar above your results - “Work obstacles and ownership differ between on-site and hybrid workers” - where one click applies the split. Findings anchored to a single question appear as a card under that question’s chart, graded info, notable or critical. Expand one and you get the written analysis next to its own comparison chart. Pin the ones worth keeping, dismiss the ones you disagree with, and when new responses stop supporting a finding, it retires itself.
The order matters here: statistics decide what is real, AI only explains it. A suggestion you see has earned its place in the list. And because the analysis runs on every comparison in the data - not on the questions you thought to type into a chat box - there’s no prompt to write and no monthly quota of AI questions to spend. The work is done before you open the page.
[Image: An expanded inline insight card below a question’s chart, with the “critical” severity badge and “segment gap” label, the AI-written analysis paragraph, Pin and dismiss actions, and an embedded per-segment means comparison chart with its own view switcher. Above it, the Recommended splits bar with one suggestion card active.]
Watching answers change over time
The overview shows where answers stand today. To watch a question move, link it to a metric: the metric tile becomes that question’s timeline - the current score with a trend line behind it, coloured by the health bands you set, updating as responses arrive. One question in one long-running survey is enough; a weekly pulse with a single linked satisfaction question is already a working trend chart. The same mechanism scales up to metrics fed by several surveys when you need it.
[Image: A metric tile for a single survey question: current score, 30-day trend line changing colour where it crosses a health-band threshold, next to the question’s distribution chart on the overview.]
Individual responses and export
Two more tabs sit next to the overview. Responses is a table of individual submissions with status, duration, completion and country; add any question as a column, open any row to read the full response. Download exports the complete dataset as an XLSX file. Your data is yours; there’s no export paywall and no lock-in.
[Image: The Responses tab: table of individual submissions with ID, date, status, time spent, completion and country columns, one row opened in a side panel showing the full response.]
A fourth tab, Funnel, shows where respondents drop off before they ever reach these results. It’s a big enough subject to get its own page.
Access that respects your respondents
Written answers are often where the sensitive material lives, so text responses sit behind their own permission. You can give someone the full statistical picture - charts, splits, funnels - while the verbatims stay restricted to the people who need them.
Everything on this page - the charts, the filtering, the splits, the AI analysis, the export - is included in every workspace, on every plan. Analytics is not a pricing tier here. Start from a template, share the link, and the first response draws the first chart. If you are deciding how many responses you need before the numbers mean something, the sample size guide walks through it.