Every quiz has a result mode, and it decides what the shopper's answers are actually for. The three modes are genuinely different mechanisms, not presentation choices, so picking the right one at the start saves rebuilding later.

Use it for: "find the right product for me" quizzes. Skin type, riding style, gift finder, size and fit.

Each answer you write can be mapped to one or more products. When a shopper selects an answer, every product mapped to it gets one vote. At the end, votes are tallied across every answer they chose, and the product with the most votes is the result.

That means a product mapped to three different answers is more likely to win than one mapped to a single answer. This is the main lever you have over a match quiz: if a product never wins, map it to more answers; if one wins too often, map it to fewer.

Ties are broken predictably, not randomly. If two products finish level, Arbyn takes the one it met first, walking your questions in order and the answers within each question in order. So the earlier a product appears in your quiz, the better it does in a tie. If you have a house favourite you would rather win a coin-toss, map it to an early answer.

When no product wins at all, meaning nothing the shopper selected was mapped to anything, the fallback product is shown instead. The fallback also steps in when a product did win but is no longer sellable, for instance because it was unpublished or deleted since you built the quiz. In both cases the shopper is shown the fallback rather than an empty screen.

If the fallback itself is missing or unpublished, the shopper sees a plain "no product" result. It is worth setting a fallback that you know will stay published.

Before a match quiz can go live, at least one answer somewhere must map to at least one product, and every required question must have at least one mapped answer. A required question whose answers lead nowhere would let a shopper spend effort on a question that cannot influence anything, so activation is refused.

Use it for: assessments and diagnostics where the outcome is a verdict rather than a product. Skin-concern severity, experience level, readiness scores, "which tier are you" quizzes.

Each answer carries a point value. The shopper's raw score is the sum of the points on everything they selected, and that is turned into a percentage of the maximum they could have scored.

The maximum is calculated per question and summed:

Negative point values are allowed, and they are floored at zero when working out the maximum, so a shopper can never score above 100 percent by avoiding the penalty answers.

Tiers then map that percentage to a result. Each tier has a minimum percentage, a label, and a message. The shopper gets the highest tier whose minimum they have reached, and landing exactly on a threshold counts as reaching it. A tier at 60 percent catches a score of exactly 60.

One tier has to start at zero, and Arbyn refuses to activate the quiz otherwise. Without a floor tier, a low-scoring shopper would finish the quiz and land nowhere.

Each tier carries its own reward, and each must be complete in its own right. That is what makes scored quizzes powerful: a high scorer and a low scorer can receive genuinely different things.

Use it for: lead capture and promotions where the questions exist to engage and qualify rather than to decide anything. A style quiz that ends in a welcome discount for everyone.

The reward is fixed regardless of what anyone answers. Product mappings are ignored entirely in this mode, so there is no point setting them.

This is the simplest mode to keep working, because there is no way for an answer to lead somewhere broken. If your goal is captured leads rather than a tailored recommendation, it is usually the right choice, and it will not quietly deactivate itself the way a match quiz can when a product is unpublished.

A quick way to decide: if the shopper's answers should change what they are shown, use Match or Scored. If the answers are really there to hold their attention, use Offer.

Cross-sell products show on the result screen for all three modes, not only match. They also appear on the screen a shopper sees when they come back having used up their retakes.

This is worth knowing because it is a free win: even an Offer quiz, which recommends nothing on its own, can put a few products in front of someone at the moment they are most engaged.

The recommended product on the result screen carries two buttons: Add to bag, which drops it into the shopper's chat bag so they can keep browsing, and Buy now, which builds a checkout and sends them to it directly.

This matters because the result screen is the highest-intent moment the quiz produces. The shopper has just been told which product is right for them, and asking them to go and find it themselves loses people. Both buttons work whether or not the shopper has ever sent a chat message, so a quiz taken cold still converts.

The buttons appear on every product card the quiz shows: the recommended product, the fallback when it steps in, and each cross-sell. They are also on the screen a returning shopper sees once their retakes are used up, so a product they were shown last time is still one tap from the bag. In a scored or offer quiz that recommends nothing, there is simply no product card to put them on.

In the widget customizer they are visible but inert. The preview has no live storefront behind it, so there is no real variant to add or check out with. The buttons render so you can see the true layout, and they do nothing until the widget is live on your storefront.

In match mode, what happens if a shopper's answers tie? The product Arbyn met first wins, reading your questions and answers in the order you arranged them. It is deterministic, not random.

Why does one product always win my match quiz? It is almost certainly mapped to more answers than the others. Vote counts are the whole mechanism, so map it to fewer, or map its rivals to more.

Does the scored percentage mean percent of questions right? No. It is the points scored as a share of the maximum points available. A quiz with no wrong answers can still produce a spread of percentages.

Can a shopper score over 100 percent? No. Negative answers are excluded from the maximum, so the ceiling holds.

Do product mappings do anything in Offer mode? No. Offer mode ignores them and gives everyone the same reward.

Why did my live quiz go back to draft? Something it depends on has gone: a mapped product unpublished, or a reward pointing at a discount that was deleted. The quiz tells you which check failed.