Product Recommendation Quiz: How to Build One That Sells

Why product quizzes convert

Three mechanisms, and it is worth knowing which one you are relying on:

• They remove choice paralysis. Forty products in a collection is a decision most visitors postpone. Four questions turn it into three.
• They earn commitment one tap at a time. Someone who has answered three questions has invested effort, and finishing feels better than abandoning — which is why the email ask on the last screen converts far better than the same ask on the first.
• They collect declared data. Not inferred from behaviour, not guessed from a purchase — actually stated. Skin type, room size, skill level, use case: the fields your welcome email has been writing around.

That third one is the quiet reason to build a quiz even if the recommendations themselves are simple. Segment data you asked for is worth more than segment data you modelled.

Start screen, one question per step, email on the last one, result on the thank-you screen. Everything else is decoration.

Step one: define the outcomes, then the questions

Almost every failed product quiz was written question-first. Someone lists interesting things to ask, then discovers at the end that the answers do not map to anything the store actually sells.

Work backwards instead:

1. List the outcomes. Three to six product groups a shopper could legitimately end up in. Not forty products — groups. "Dry skin routine", "oily skin routine", "sensitive starter set".
2. Name the attributes that separate them. What genuinely distinguishes group A from group B? Skin type, budget, experience level, room size, frequency of use.
3. Write one question per attribute. That is your quiz. If you have seven attributes, you have too many outcomes — merge some.

Step two: write questions people enjoy answering

A quiz question is not a form field. It should be readable in one glance, answerable without thinking hard, and phrased in the shopper's language rather than your catalogue's.

Four rules for the options themselves:

• Three to five options. Two feels like a coin flip; six needs scrolling on a phone.
• Mutually exclusive. If a shopper could honestly pick two, either allow multiple selection or rewrite the options.
• Include the honest escape. "Not sure yet" is a real answer and stops people guessing — a guessed answer poisons both the recommendation and your data.
• Write them at the same length. One long option among four short ones gets picked disproportionately, and now your data is about layout instead of preference.

Each question is a choice block: single-select for a clean segment, multi-select where a shopper genuinely wants two.

Single-select gives you one clean value per shopper, which is what you want for routing. Multi-select is right for "what are you shopping for?" style questions where several answers are true — just accept that it is harder to segment on later.

Step three: deliver the recommendation

Here is the honest part. Quiz steps run in order, and there is no built-in branching — no "if they answered dry, jump to screen 5". That constraint turns out to be liberating once you know it, because there are three good ways to deliver a per-shopper result and none of them need branching.

Option A — one thank-you screen, one destination

The simplest version, and the right one for most stores. The final screen says "Based on your answers, start here" with a button to the collection that fits the dominant answer pattern. It works because the quiz has already narrowed the shopper's mindset — they arrive at that collection knowing why.

Option B — deliver it by email

The answers are stored with the submission, so you can send the tailored recommendation from your email tool. Slower, but it gives you room for three products, a photo and a reason — and it makes the first email you ever send that subscriber genuinely useful, which does more for long-term list health than any welcome discount.

Option C — redirect to the matching collection

The pixel fires a browser event when a quiz is submitted, carrying the answers. A few lines of JavaScript on your site can turn that into a per-answer destination:

window.addEventListener('chillipopup:quiz_submit', function (e) {
  // e.detail = { quizUuid: '…', data: { 'skin-type': 'Dry', 'budget': 'Under $50' } }
  if (e.detail.quizUuid !== 'YOUR_QUIZ_UUID') return;

  var routes = {
    'Dry': '/collections/hydrating',
    'Oily': '/collections/balancing',
    'Sensitive': '/collections/gentle'
  };
  var pick = routes[e.detail.data['skin-type']];
  if (pick) window.location.href = pick;
});

Forms and surveys fire their own equivalents (chillipopup:form_submit, chillipopup:survey_submit), each carrying its own id key — so match on the right one if you run several.

Where to ask for the email

On the last screen, next to the result — and show the result either way. Three reasons:

• Effort earns the ask. Someone four taps in has invested something. Someone on screen one has not.
• Gating breeds fake addresses. Hold a result hostage and you get a@a.com, which is worse than no address at all — see how to stop spam signups.
• The answers are worth more than the address. If they abandon on the email screen, what they already answered is captured as a partial response — so the segment insight survives even when the signup does not.

Frame the ask as delivery rather than a toll: "Where should we send your results and the matching products?" — with the result visible behind it.

Four quizzes you can copy

Skincare store — 4 questions

1. How does your skin feel by mid-afternoon? Tight / Shiny / Both / Comfortable
2. What's the one thing you'd fix first? Dryness / Breakouts / Dullness / Fine lines
3. How many steps will you actually do? One / Three / As many as it takes
4. Where should we send your routine? [email]

Coffee store — 3 questions

1. How do you make it at home? Espresso machine / Filter / French press / Pods
2. How do you take it? Black / With milk / Both, depends
3. Strength? Gentle / Balanced / Hits like a truck

Apparel store — 4 questions

1. Who are you shopping for? Me / A gift
2. What's the occasion? Everyday / Work / Something special
3. Which fit do you live in? Relaxed / Regular / Tailored
4. Budget to start? Under $50 / $50–120 / No limit

Software or service — 3 questions

1. What best describes you? Solo / Small team / Agency / Enterprise
2. What are you trying to fix first? [three real use cases]
3. Where should we send the plan comparison? [email]

Where to place the quiz

A quiz is an offer to help someone decide, so it belongs where deciding is hard.

Whatever you pick, cap the frequency so a browsing session is never interrupted twice, and turn off "show again after conversion" so someone who finished the quiz is not asked to retake it.

Reading the results

The quiz analytics answer three different questions, and the third one is usually the most valuable thing the campaign produces.

1. Completion rate. The share who finish once they start. Below about half, the quiz is too long or one question is doing damage.
2. Drop-off by question. The dashboard ranks where people quit, based on the last field they touched. That is your rewrite list, in order.
3. The option breakdown. How many shoppers picked each option, per question. This is merchandising research: if 70% of your traffic answers "gift", your homepage is speaking to the wrong shopper, and no amount of quiz tuning fixes that.

Completion tells you about the quiz. The option counts tell you about your customers — and that insight outlives the campaign.

The popup analytics guide covers how to read these without over-interpreting a small sample.

Seven mistakes that sink a product quiz

1. Too many questions. Past five, completion falls off a cliff and the extra data is rarely worth what it cost.
2. Questions that don't change anything. "What's your name?" mid-quiz is a form field wearing a costume.
3. Gating the result. The shopper did the work; show them the answer.
4. Recommending everything. "Here are 12 products for you" is the catalogue again. One clear pick, maybe two alternates.
5. Free-text in the middle. It breaks the tapping rhythm and returns values you cannot segment on.
6. No promise on the start screen. Say what they get and how long it takes: "3 questions, 20 seconds, your routine at the end."
7. Never reading the option counts. The quiz is a survey of your own demand. Ignoring it is the most expensive mistake on this list.

Build it in five steps

About an hour. If the pixel is not on your site yet, start with the install guide — one script serves quizzes, popups, forms and surveys alike.

Related reading

Frequently asked questions