Conditional Logic in Forms: Branch After the Section, Not After Every Question

The best place for conditional logic is between sections, not after every question. A form that branches at section boundaries is easier to build, easier to test, and kinder to respondents, because each person finishes the topic they started before the route changes. And route changes matter: in one web survey experiment, 75% of people started when told it took 10 minutes, against 62% when told 30.

Fewer rules, shorter paths, no stranded questions. That is the case for section branching. Below are the definitions people mix up, the research on survey length, the exact rules our own builder runs, and 6 worked examples you can copy.

Skip logic, branching and display logic are three different tools

Most guides use these terms as if they meant the same thing. They do not, and picking the wrong one is how forms end up with 40 rules nobody can read. The difference is the unit each one controls.

Skip logic jumps forward from one answer

Skip logic is the oldest and simplest form. One answer triggers a jump to a later question, the end of a block, or the end of the survey. Qualtrics documents skip logic as sending respondents to a future point, and notes it only moves forward, never back. Its own help page recommends display or branch logic once you need AND and OR conditions.

Branching routes whole groups of questions

Branching works on sections, blocks or pages. It decides which group a respondent enters next. Two people can answer the same opening question and then take entirely different paths. Google Forms works this way: its setting is called "Go to section based on answer", and Google's help page says it works only for Multiple choice and Dropdown questions.

Display logic hides or shows one item

Display logic leaves the order alone and toggles single items. Qualtrics describes display logic as showing or hiding a question based on what you already know about the respondent. Its guidance is telling: if a group of questions shares the same logic, use branch logic instead. We agree. Display logic on six questions in a row is a section that has not been named yet.

Why branching after a section beats jumping after every question

Per-question jumps feel precise. In practice they create a form where every question is a fork, and nobody can predict the full path. We think section-level branching is the better default for almost every form, for three reasons.

First, respondents finish what they started. If a jump fires mid-section, the remaining questions in that topic vanish, and the respondent loses context. Second, the logic is auditable. Each section has one exit, so you can draw the whole form on a napkin. Third, required questions stay honest, because they only apply to sections a respondent actually entered.

Shorter paths reach more people

Branching is a way to make each respondent's form shorter without deleting questions for everyone. The research on length points the same way, with caveats we cover later. In 2009, Mirta Galesic and Michael Bosnjak varied the announced length of a web survey in Public Opinion Quarterly. 75% started the 10-minute version, 65% the 20-minute version and 62% the 30-minute versionGalesic and Bosnjak, POQ 73(2), 2009. Opt-in web survey, Croatia. Figures as reported in Hanson et al., Survey Research Methods 19(2), 2025..

The same study found a second effect. Questions placed later in the questionnaire drew faster, shorter and more uniform answers than the same questions placed early. That is a quality cost, not just a volume cost. Every irrelevant question you show someone pushes the questions that matter further down, where answers get thinner.

Fewer rules mean fewer broken paths

A 20-question form with a jump on every question has, in theory, a different path for every combination of answers. A form with 4 sections and 1 exit rule each has a handful of paths you can list. We would not ship a form whose paths we cannot list. Untested paths are where respondents hit dead ends, loops or required questions they cannot answer.

How section branching works in Would You Rather

Here is exactly what our builder does, taken from the routing code rather than the marketing page. The respondent sees one question at a time in their phone browser. Sections are created with a section break, and logic is set either on a choice question or as a rule on any question.

Choice questions branch per option

On a single choice or multiple choice question, every option gets a destination. The choices are "Continue to next question", "Go to" a named section, or "Submit form". If you allow an Other option, it gets a destination too. On a multiple choice question where someone ticks several options, the first ticked option in your option order that has a branch decides the route.

Rules combine conditions with and or or

Any question can carry rules. A rule reads "If answer to" a question, then one of five checks: is answered, is not answered, equals, does not equal, or contains. You can add several conditions and join them with "and" or "or", and the question you test can be a different one from the question holding the rule. The target is a section or Submit form. Rules run top to bottom, and the first match wins.

Sections have one exit setting

Every section break has an "After this section" setting with three choices: continue to the next section, go to a named section, or submit the form. One detail trips people up. A section that a respondent reached by a branch, and that is set to continue, ends the form rather than spilling into the next section. To chain branched sections together, point "After this section" at the next one explicitly.

Two safety rails sit underneath. The builder will not let a question branch back into its own section, and the runtime ends the form if two sections point at each other in a loop. Required questions are checked against the path the respondent actually took, both in the browser and again on our server, so nobody is blocked by a question they never saw.

6 worked examples of section branching

Each example below uses only the controls described above. Copy the structure, not the wording. The table summarises them, and the notes underneath explain the one decision that matters in each.

Example 1, the polite early exit. Put the eligibility question alone in the first section. Set "No" to Submit form, which fires immediately, and point your thank-you text at people who did not qualify. Set "Yes" to the Experience section. Everyone who stays then reaches the one question worth a spoken answer, which we cover in voice screening questions for hiring.

Example 2, follow up only on bad news. A single choice question with three options is easier to branch than a 0 to 10 score. Send "Bad" to a short section with one open question, ideally voice, and send the others to a wrap-up. If you prefer a numeric score, see the NPS follow-up question for how to phrase the follow-up itself.

Example 3, two paths that rejoin. In-person and online attendees each get a branch-only section. Set both sections' "After this section" to a shared Final details section. Without that setting, each branch would end the form at its last question, which is the trap described above.

Example 4, multiple choice needs an order. When a respondent can tick several animals, only one route can win. Put the option with the longest follow-up first in your option list, because the first ticked option in that order decides. If every ticked option needs its own questions, use a series of yes or no questions instead. We discuss form design for village surveys in forms for field research.

Example 5, route by language. Ask the language question first, then send each choice to a section written in that language. Record the question prompt in your own voice for each section, so respondents who read slowly can listen instead.

Example 6, conditions across questions. Put a rule on the "Have you contacted support?" question: plan equals Team and contacted support equals Yes, go to Support experience. Everyone else continues past that section, because it is branch-only. Add an "is not answered" condition when an optional question should skip people who left it blank.

Shorter paths help, but length is not the whole story

It is tempting to read the length research as "every question costs X% of respondents". The evidence does not support that, and we would not plan a form around a single percentage. The honest summary is that length matters a little, and relevance and design matter at least as much.

A 2015 meta-analysis by Aigul Mavletova and Mick Couper pooled 39 samples of mobile web surveys. A 30-minute survey had 1.42 times the odds of breakoff of a 5-minute one, though the length effect fell just short of statistical significanceMavletova and Couper, Mobile Research Methods, Ubiquity Press, 2015, chapter 6. 14 studies, 39 samples, 34,589 starters, 4,209 breakoffs. Length effect p = 0.07.. Length still explained the largest share of the difference between studies.

The same meta-analysis found design complexity mattered clearly. Adding grids, drop-downs, sliders, images and progress bars all together raised the odds of breakoff by 1.91. That supports a one-question-at-a-time layout as much as it supports branching.

The strongest counterweight is recent. In 2025, Tim Hanson and colleagues compared a 50-minute and a 35-minute version of the European Social Survey in Austria, published in Survey Research Methods. The shorter version lifted response from 34% to 38%, and online breakoff was 12% for the longer version against 13% for the shorter oneHanson, Aizpurua, Fitzgerald and Vukovic, Survey Research Methods 19(2), 2025. Austria, 2021, push-to-web with paper follow-up, about 2,000 sampled households per condition.. Data quality showed no clear difference. Cutting 15 minutes of content bought 4 points of response.

Our reading: branching earns its place by removing questions that do not apply, not by chasing a shorter average. A question someone cannot answer is worse than a long form. It is the moment they decide the form is not for them.

Section branching vs per-question jumps

Here is how the two approaches compare when you build the same form both ways. The last row is where per-question logic genuinely wins.

That last row is a real cost. If you only need to show a single follow-up to some people, creating a one-question section feels heavy. We still think it is worth it, because the section has a name, and a name is documentation.

Where section branching is the wrong tool

Section branching has limits, some general and some specific to our builder. You should know them before you design around it.

There is no per-question display logic in Would You Rather. You show or skip whole sections, and a single conditional question needs a section of its own. The checks are also text comparisons: equals, does not equal, contains, answered and not answered. There is no greater-than or less-than, so routing on a score range means a choice question or several "or" conditions.

You cannot branch on what someone said in a voice answer. We do not transcribe recordings, so there is no text to test. You can branch on whether a voice question was answered, and on the choices around it. Routing also only moves people to sections, never back into the section they are in.

The research has limits too. The Galesic and Bosnjak figures describe announced length, not the length of a branched path, and we take them from a later paper's summary of the original. The meta-analysis length effect was not statistically significant. The Austrian experiment tested very long questionnaires, 35 and 50 minutes, so it tells you little about a 3-minute form. None of these studies tested branching directly, which is the weakest link in our argument.

Where to start with your own form

Take the form you send most often and list the respondent types it serves. If it is two or more, sketch one section per type plus a shared start and finish. Write the exit rule for each section in a single sentence. If you cannot, the section is doing two jobs.

Then walk every path in preview, once per respondent type, and count the questions each person sees. If any path shows someone a question that cannot apply to them, that is your next branch. Our post on which question types drive form drop-off covers what to cut next, and sending responses to Google Sheets shows how branched answers land in one sheet, with blank cells for sections a respondent skipped.

References

1. Galesic, M. and Bosnjak, M., Effects of Questionnaire Length on Participation and Indicators of Response Quality in a Web Survey, Public Opinion Quarterly 73(2), 2009. Used for the effect of announced length on starting and for later questions drawing shorter answers (abstract read; start rates as reported in reference 3).
2. Mavletova, A. and Couper, M. P., A Meta-Analysis of Breakoff Rates in Mobile Web Surveys, in Mobile Research Methods, Ubiquity Press, 2015. Used for breakoff odds by length and design complexity.
3. Hanson, T., Aizpurua, E., Fitzgerald, R. and Vukovic, M., Surely Shorter is Better? A Questionnaire Length Experiment in a Self-completion Survey, Survey Research Methods 19(2), 2025. Used for response and breakoff rates by length, and for its review of Galesic and Bosnjak.
4. Google, Show questions based on answers, Google Docs Editors Help, accessed 8 October 2026. Used for how Google Forms branches and which question types support it.
5. Qualtrics, Skip Logic and Display Logic, Qualtrics Support, accessed 8 October 2026. Used for the definitions of skip, display and branch logic.
6. Would You Rather, routing code and builder logic panel, read 8 October 2026. Used for how section branching, rules and required checks behave.

The weakest point in this source base: no study here tests section branching against per-question jumps directly, so the case for sections rests on length research plus design reasoning, and the Galesic and Bosnjak percentages come from a secondary summary.