Guided Selling Quiz

When the Shopper Doesn't Know What to Type#

A meaningful share of visitors arrive on your store with no clear search query in mind. They are browsing, gift-hunting, or just curious. They open the chat, see a blank input box, and bounce.

The Guided Selling quiz catches these visitors. Instead of staring at an empty prompt, they see a tap-friendly multiple-choice card -- "Who's this for?" -- and the AI walks them through a short, store-specific conversation that ends with real product recommendations.

What the Shopper Sees#

The quiz shows up as the first option when a visitor opens the widget without an obvious question. It is a single card with:

• A short progress bar (typically 2-4 steps)
• One question at a time
• 3-5 tap targets per step, plus an optional "skip" link from step 2 onward

A typical session looks like this:

1. Step 1: Who's this for? -- Myself / Someone else / Just browsing / Not sure yet
2. Step 2: What are you after? -- Everyday basics / Something special / Surprise me
3. Step 3 (generated by the AI from your catalog): Any budget in mind? -- Under $50 / Under $100 / No limit

When the AI has enough signal, it hands off to the regular chat with a natural-language query like "a thoughtful gift under $50 for someone who likes everyday basics." The shopper sees a normal AI reply with bulleted product suggestions and clickable product cards -- exactly the same response the chat would give if they had typed that sentence themselves.

What Makes It Different#

Static "shopping wizards" exist on a few storefronts -- usually hand-built by an agency with three hard-coded questions. Kwiro's quiz is different in three ways.

1. The questions adapt to YOUR catalog#

Step 1 is the same on every store -- it is a fast, no-API-cost opener. Every step after that is written live by the AI based on the shopper's previous answers and your actual product mix.

A pet store quiz might ask "Dog or cat?" and "Puppy, adult, or senior?" A jewelry store quiz might ask "Gold or silver?" and "Daily wear or special occasion?" An electronics store gets "What are you replacing?" Same engine, different questions, all driven by your catalog -- nothing for you to configure.

2. The AI knows when to stop#

The quiz is capped at 5 steps, but most sessions finish in 2-4. The model decides on each turn whether it has enough signal to make a good recommendation, or whether one more clarifying question would help. The moment it knows enough, it bails out and hands off to the chat.

This matters because the failure mode of bad quiz UX is a 7-step interrogation that tells the shopper less than they already knew. Kwiro is built to err on the side of finishing early.

3. It hands off to the real chat -- not a canned page#

Some quiz tools end with a static "results" screen and a list of links. Kwiro's quiz ends inside the chat. The handoff query goes through the full recommendation pipeline:

• Sale-aware (factors in active promotions)
• Stock-aware (never recommends out-of-stock items)
• Currency-aware (uses your store's currency)
• Memory-aware (uses what Kwiro has learned about your store from previous winning conversations)

The shopper can keep chatting from there -- "show me cheaper options," "do you have it in red?", "add the second one to my bag" -- without ever leaving the conversation.

When the Quiz Is Shown#

The quiz is offered when the widget opens fresh -- no prior conversation, no obvious question typed yet. Returning shoppers and shoppers who type a specific question go straight to the chat as before.

The quiz is also context-aware on category and product pages. If a visitor opens the widget while browsing the "outdoor" category, the AI weights its questions toward what is in that category.

Speed and Reliability#

Each step typically returns in 1-2 seconds. If the model fails to respond for any reason (network blip, rate limit, API outage), the widget falls back to a deterministic 3-step static flow so the shopper never hits a dead end.

Available on Every Plan#

Guided Selling is on by default for Free, Pro, Growth, and Scale -- no setting to toggle, no quota separate from your conversation cap. A completed quiz that ends in a chat handoff counts as one conversation, exactly the same as if the shopper had typed their question directly.

What You'll See in the Dashboard#

Quiz sessions show up in your conversations list with the full transcript -- you can see exactly which questions the AI asked, which answers the shopper picked, and what was recommended. They are flagged with a small "guided" tag so you can filter them when reviewing performance.

If you spot a quiz step that consistently confuses shoppers (everyone picks "skip" on step 2), that is a signal worth flagging. The Memory Inspector picks up on these patterns over time and the AI adjusts its question selection accordingly.