Guided selling on Shopify: ask the right question before you recommend

What is guided selling on Shopify?

Guided selling is a way of helping a shopper choose by asking about her need before showing products. On Shopify it usually takes the form of a product quiz, a product finder, or a chat assistant that asks a question and then recommends a few items that match the answer, each with a short reason.

It is the online version of a good shop-floor conversation. A shopper walks into a shoe store and the associate asks, “Running or walking?” before bringing anything out. That single question cuts the options down to the ones worth trying. Online, the shopper has to do that narrowing herself with menus and filters, and some give up first.

Baymard Institute's large-scale research on product lists and filtering found that sites with mediocre product list usability saw abandonment rates of 67% to 90%, against 17% to 33% on sites with a slightly better toolset for shoppers looking for the same kinds of products. Finding the right item is where many sales are won or lost, before any price or checkout question comes up.

Product quiz or conversational guided selling: which fits your store?

A product quiz suits stores where every shopper needs the same few answers, such as a skincare routine or a coffee subscription. Conversational guided selling suits stores where shoppers arrive knowing part of what they want. A quiz asks everyone every question; a conversation asks only what is still missing, and skips the question when the shopper already said it.

Quizzes have real strengths. They are predictable, easy to brand, and they collect zero-party data in a structured form. Their weakness is length: every extra step can lose people, and a shopper who already knows she wants a size 9 wide trail shoe still has to click through questions about her running style.

Conversation flips that. It starts from what the shopper typed, which often already contains the deciding detail. The risk is the opposite one: a chat assistant that asks five questions in a row is just a slower quiz. The discipline that makes it work is asking one question, at most, before recommending.

Which question should you ask first?

Ask the question whose answer removes the most wrong products. In most verticals that is a fit or compatibility attribute, not a style preference: shade and skin type in beauty, size and fit in apparel, width in footwear, metal and ring size in jewelry, dimensions in home. Style can be shown; fit has to be asked.

A useful test is to look at your return reasons and your support inbox. If “wrong size” dominates returns, size is your deciding question. If customers keep emailing to ask whether a sofa fits through a doorway, dimensions are. The deciding question is the one a shopper would regret not being asked.

Why ask only one question before recommending?

Because each question costs attention, and the shopper came for an answer, not an interview. One well-chosen question usually narrows a catalog enough to recommend one to three products. If more detail is needed, the recommendation itself can invite it: “Here are two in your size; want the lighter fabric or the warmer one?”

This keeps the shopper in control. She gets something concrete to react to after one exchange, and her reaction (“the first one, but in black”) is often more informative than another abstract question would have been. It also respects shoppers who arrive ready: if the message already says “size 8, wide, for trail running,” there is nothing left to ask.

• Skip the question when the shopper's message already contains the deciding detail.
• Offer answers as short choices where possible (dry, oily, combination, not sure).
• Always allow “not sure” and handle it with a safe default and a reason.
• Recommend after one question, then refine from the shopper's reaction.

How do you build guided selling from your catalog data?

Guided selling is only as good as the product data behind it. The question has to come from attributes your products actually have, and the answer has to map onto real variants. If shade, width or dimensions live only in photos or in a PDF, no quiz or assistant can match a shopper to them reliably.

Sizing is a clear example. Baymard Institute found that 83% of desktop apparel sites and 87% of mobile apparel sites fail to provide sufficient sizing information (published July 2022). If the size guide is incomplete for a shopper reading it herself, it is incomplete for any tool trying to answer her sizing question as well.

1. List the deciding attribute for each product type, using the table above as a starting point.
2. Make sure that attribute exists as a variant option, tag or text field, not only in an image.
3. Put size charts and dimensions on the product page as text, with units.
4. Use consistent values across products (one spelling of “wide,” one scale for shade depth).
5. Check that sold-out variants are excluded when a recommendation is made, using live inventory.

How Tycoon approaches guided selling

Tycoon, an AI sales associate for Shopify stores now in early access, is built this way. When the need is unclear, it asks one short question built from the store's own catalog, then suggests one to three products, each with a reason. It skips the question when the shopper already said what she needs.

Each vertical has its own deciding detail it asks first, for example skin type, shade or finish for beauty stores, size, fit or fabric for apparel, and size, width or activity for footwear. On install it reads the catalog, size guides and the store's own FAQ, so there is nothing to upload. Prices and stock on each product card are read live for every answer.

On the Tycoon plan, nightly learning looks at which questions helped shoppers decide in that store and tunes which question to ask first, learning only from that store. More on the product recommendations and guided selling feature.

Frequently asked questions

Sources

1. Baymard Institute: Product lists and filtering usability research · checked October 20, 2026
2. Baymard Institute: 83% of apparel sites don't provide sufficient sizing information (2022) · checked October 20, 2026