Quiz Branching Logic
Quiz branching logic routes each respondent to a different next question or section depending on the answer they just gave.
Key takeaways
- A branch is a condition plus a target step, evaluated once the answer is submitted.
- When no condition matches, the flow must fall through to a defined default step.
- Routes of different length reach different maximum scores; normalise before comparing tiers.
- Loops and orphan steps strand respondents; every branch needs a reachable exit.
- Each split shrinks the sample per route, so per-path analytics take longer to read.
In depth
A branch is a condition paired with a target: if the answer matches, go to this step instead of the next one. The engine evaluates the conditions attached to a question after the answer is submitted, takes the first that matches, and falls through to the default next step when none does. Three targets cover almost every case: skip forward past questions that no longer apply, jump into a block written for one segment, or end the quiz and send the respondent straight to an outcome.
Every branch you add splits the population, and splits compound. Two branch points with two options each already produce four routes to write, style, translate and test. The gain is relevance, because nobody is asked about a tool they do not use; the cost is content and analytics, since each route carries a smaller share of traffic and takes longer to read reliably. Branching early on one strong segmenting attribute buys most of the benefit at a fraction of the maintenance.
The practical pattern is to branch once, near the front, on a factual attribute such as company size, role or current tooling, then keep each route roughly the same length so paths stay comparable. In a scorecard funnel the routes usually differ in which questions are scored, so express the result as a share of the maximum reachable on that path rather than as a raw total. Tagging the path taken also tells sales which version of the conversation the lead just had.
Branching does not rescue a segmentation you have not thought through. If the question that drives the split is subjective, people sort themselves inconsistently and every downstream route receives a blurred audience. Unequal path lengths quietly bias the data, because a short route produces higher completion for reasons unrelated to its content. At low traffic, per-path numbers stay too thin to act on, and a quiz with many routes can run for months before any single one has a readable sample.
Example in practice
How to measure it
Report per route, not per quiz. For each path record sessions entered, completion rate and average questions answered. A route whose completion sits far below its siblings is either too long or receiving people it was not written for. Also count how often the default fall-through fires: a rising share means real answers no longer match any branch condition you wrote.
Then compare what the routes produce. Look at the tier distribution within each path and at what those leads do afterwards, such as booked calls or replies. If two routes yield near-identical distributions and outcomes, the split is not earning its maintenance cost and can be merged. Normalise scores against each path's own maximum before putting the tiers side by side.
Common mistakes
The costly mistake is branching on a subjective question such as the respondent's biggest challenge. People interpret those options differently, so the routes fill with mixed audiences and the tailored questions behind them lose their point. Branch on something factual the person cannot misread, such as headcount, industry or which system they use today, and keep subjective questions for scoring rather than for routing.
The second is never testing what happens when someone goes back. A respondent who returns and changes the answer that drove the split may keep the questions from the old route, carry points that no longer belong to their path, or reach a step that is now unreachable. Decide explicitly whether changing that answer re-evaluates the route, then walk the back-and-change case for every branch point before launch.
Frequently asked questions
How is branching logic different from simple scoring?
Scoring assigns points to answers but shows everyone the same questions. Branching changes which questions appear next, so two respondents can take entirely different paths through the quiz.
Does branching make a quiz shorter?
Usually yes, because respondents skip questions that do not apply to their situation. This typically lifts completion since the quiz feels more relevant and less repetitive.
What is the biggest risk with branching logic?
The main risk is creating dead-end or orphan paths where a respondent has no valid next step. Map every branch end to end and test each route before publishing.
When should I use branching instead of asking everyone everything?
Branch when a meaningful share of respondents would otherwise face questions that cannot apply to them, such as CRM detail questions for someone with no CRM. If every question is answerable by everyone, branching adds maintenance without adding relevance. The test is simple: name the questions the branch lets you skip, and check that a real segment would genuinely be unable to answer them.
How many branches are too many?
Count routes, not conditions. Once the number of distinct paths exceeds what you are willing to write copy for, translate and retest after every change, you have too many. Most funnels stay readable with a single early split into two to four routes. Deeper trees are usually a sign that several separate quizzes would serve the segments better than one.
Do different branches make the scores incomparable?
They do if you compare raw totals, because a shorter route offers fewer points to collect. Convert each result into a share of the maximum reachable on that specific path, then set tier thresholds on that share. Alternatively, keep a common core of scored questions that everyone answers and treat the branch-specific questions as segment detail rather than as part of the qualification score.
What happens if someone goes back and changes an answer?
That depends on whether your platform re-evaluates the branch. Some keep the route already taken and simply update the score, others recalculate and move the respondent. Neither is wrong, but the behaviour must be deliberate, because silently keeping an obsolete route means the person answers questions written for a segment they just left, and their score reflects a path they are no longer on.
Can a branch end the quiz early?
Yes, and it is one of the more useful applications. When an answer makes someone a clear non-fit, sending them straight to a suitable outcome respects their time and keeps unqualified records out of the pipeline. Give that early exit a real result rather than a dead end, since the person still finished something and may fit a different offer later.
How do I map branching before building it?
Draw each question as a node and each branch as an arrow to its target, then trace every arrow to an outcome. Any node without an incoming arrow is unreachable, and any node without an outgoing path strands respondents. Doing this on paper takes minutes and catches structural faults that are far harder to spot once the flow exists inside an editor.