Matching answers to products

#How matching works

In a quiz with Products that fit results, each answer can carry rules. A rule says what the answer does to a group of products: push them up, push them down, keep only them, or rule them out. When a shopper finishes, Teacup Quiz combines the rules of every answer they picked and shows the products that fit best.

To set an answer’s rules, click Choose products under it in the question’s settings (Edit products once it has some), or hover the answer on the canvas and click the same button there. The dialog is named after the answer, e.g. When shoppers pick “Oily”, and asks “Which products should this answer bring up?”

Quizzes with A personal result use points instead: see Result pages.

#What a rule does

Each rule is a card in the dialog. Its first field says what the rule does:

How several Only recommend rules combine

• On one answer: a product must fit all of that answer’s Only rules.
• Several answers picked in one question: a product must fit any one of them (Black or Navy). Turn on Products must fit every answer picked in the question’s More options to require all of them instead.
• Different questions: a product must fit the Only rules of every question.

#How much

Recommend more and Recommend less also say how much, in the field next to them:

A quiz drafted by AI can use other values; they show as Custom (1.5). The chips under an answer don’t show how much: open the answer to see it.

#What a rule points at

Under what the rule does, pick which products, then the value:

• Tags, product types and brands ignore capitals and spaces at either end: “Oily”, “oily” and “ oily ” are the same.
• A price range includes From but not Up to: $25 to $50 covers $25.00 up to $49.99, and a $50 product falls in the next band. A product’s price is its lowest variant price.
• A price range needs at least one amount, with From below Up to. Otherwise it says Set at least one of the two, with “From” below “Up to”.

The plain sentence

At the bottom of each rule card, a sentence says what it does and how many products it reaches, such as Products tagged “oily” rank higher (12 products)., Products priced $25–$50 are the only ones recommended (8 products). or Products from “Acme” are never recommended (no products yet). “No products yet” usually means a typo.

The count includes every active product, even sold-out ones and ones outside the quiz’s products. Check every answer warns when a rule reaches no product the quiz can actually recommend.

Adding and removing rules

• Add products (or Add more products) adds a rule card: Recommend more, Normal, With the tag and your store’s most used tag. Change it from there.
• The trash icon on a card removes that rule.
• Done closes the dialog. Changes apply as you make them, and Undo works here too.

More options, at the bottom of the dialog, holds Why these products fit (below) and Then, to send shoppers who pick the answer to another question: see Branching. It’s closed at first. While it is, a line under it shows what’s set there, such as “Good for: sensitive skin · Then: question 3”.

#Why these products fit (Good for)

Once an answer has a Recommend more rule, More options in its dialog shows Why these products fit: “Shown as ‘Good for …’ under the products this answer brings up. Empty: the answer itself.” An empty field uses the answer’s label in lower case. Below, this is called the answer’s Good for text.

• Skincare: the answer “Oily” with “oily skin”. A cleanser it brings up says Good for oily skin, breakouts.
• Coffee: the answer “French press” with “French press brewing”.

The same words do more:

• The first question with a Short label on the results page names the results after its answer’s Good for text: “Your routine for oily skin” or “Your matches for French press brewing”. This needs the Good for text typed in; the answer label alone isn’t used here.
• The same words go to Klaviyo with the email, as routine_for, for subject lines like “Your routine for oily skin”.
• With “Good for” turned on in Integrations, your products get these phrases in a product field AI shopping agents and your theme can read. See “Good for” and AI shopping agents.

#How ranking works, step by step

For each shopper, Teacup Quiz works through these steps:

1. It starts with the products the quiz can recommend: active, on the Online Store, in stock (unless Recommend sold-out products is on), inside Products in this quiz, not under Left out of this quiz, and without a tag listed under Never recommend products tagged.
2. It removes every product a Never recommend rule on the shopper’s path points at.
3. It keeps only the products that fit every Only recommend rule on the path, combined as described above.
4. It scores the rest. Each Recommend more rule adds its points to the products it points at; each Recommend less rule subtracts its points. Rules from every picked answer count, so a product that fits three answers collects three times.
5. Best matches: when the path has at least one Recommend more rule, only products scoring above 0 are shown, highest first, up to How many products. Equal scores keep your store order, which is by product ID, so older products usually come first.
6. If the path has no Recommend more, Recommend less or Only recommend rule (a Never recommend rule is fine), the shopper picked no preference. With a collection under If nothing fits their answers, they see its products first, topped up with your first products in store order when it has too few they can be shown. Without one, they see your first products in store order. Their Never rules still hold. When they get none of the collection’s products, Check every answer adds a note about these shoppers.
7. If the path has Recommend less or Only rules but no Recommend more rule, it keeps its normal matching: the products those rules allow, in score order (Recommend less pushes some down), then store order.
8. If the path has Recommend more rules but no product scores above 0, and Only rules are in play, the products that fit the Only rules are shown, best first.
9. Otherwise, the backup from If nothing fits their answers shows: your “no match” message, or the best products from the collection you picked, still obeying every Only and Never rule.
10. On your store, each picked product is checked live. Anything that’s sold out, unpublished or deleted by then is dropped, and the next best takes its place.

In a routine, steps 1 to 4 are the same; then each routine step takes its best-scoring products among the products it points at, even when they scored 0, and a product used in one step isn’t used again. See Results.

#Example: a skincare store

A shopper picks Oily, Breakouts and Redness, Yes, and Under $40. With How many products at 3:

#Example: a coffee shop

A shopper who picks French press, Chocolatey and Fruity gets coarse-ground coffees first (2 points each, plus 1 for each flavor they fit), then other chocolatey or fruity coffees (1 point). A shopper who picks Pods and Decaf only gets decaf capsules and nothing else. If you sell no decaf capsules, that path ends with no products, and Check every answer names both answers so you can fix it, for example by dropping the Only rule on Decaf only and using Recommend more instead.

#Common patterns

A budget question

Give each price band an Only recommend rule with In a price range: “Under $25” (Up to 25), “$25 to $50” (From 25, Up to 50), “$50 and up” (From 50). Add “No set budget” without a rule. Give the question the short label “Budget” so the results say “Budget $25 to $50”. Quizzes created as Find the perfect gift get this question, built from your own prices, when the store has enough priced products. In a gift routine it uses Recommend more instead, so a budget never leaves a step of the routine empty.

Allergies and diets

• When your tags say what a product is suitable for (vegan, gluten-free): a several-answers question with an Only recommend rule on each answer, and Products must fit every answer picked on.
• When your tags say what a product contains (contains-nuts): Never recommend on the answer “Nut allergy”. Never rules hold on every path and in the backup.

Never recommend samples, gift cards and add-ons

For products that should never be a result anywhere in the quiz, add their tag under Results → More options → Never recommend products tagged, instead of a rule on every answer. Gift cards aren’t left out automatically: tag them, for example gift-card, and add that tag there.

One animal, then the details

A pet store asks “Who’s it for?” first: Dog → Only recommend the collection Dog; Cat → Only recommend the collection Cat. Later questions use Recommend more for size, age and taste, so every path stays with the right animal.