E-commerce
June 28, 2026
A large catalog can become a customer problem. When a store offers hundreds of references, filters, search, and collections are not always enough. The customer does not know what to choose, combines filters without results, or gives up after several minutes of browsing.
A well-designed AI assistant transforms this labyrinth into a conversation. It asks about use cases, budget, constraints, and preferences, then proposes a short selection of truly relevant products.
This guide explains how to design this assistant, where to place it, what data to connect, and how to measure if it actually helps customers make a decision.
Summary
Why do large catalogs block customers?
The more products there are, the more effort the customer has to make to compare, filter, and choose. Expert buyers manage fine, but others can quickly feel lost.
Signs of overload
Many visits to collections, but few product clicks.
Long sessions without adding to the cart.
Vague searches like "gift", "cream", or "black".
Filters that lead to zero results.
Support tickets like "I can't find the right product".
The AI assistant serves to reduce this mental load by guiding the customer to a few options, instead of asking them to understand the entire catalogue structure.
Convert over 2,000 customers on average per month with Qstomy.
The world’s 1st Shopify AI dedicated to customer conversion
Empowering 200+ e-commerce merchants
Why are filters not always enough?
Filters are useful when the customer already knows what they are looking for. They become less effective when the customer expresses a vague need: "a jacket for traveling light", "a gift for sensitive skin" or "a cable compatible with my device".
Common limitations
The customer does not know your category terms.
There are too many filters on mobile.
Some combinations yield zero results.
Customer synonyms do not match internal names.
The assistant does not replace filters. It complements them by translating a customer intent into usable criteria.
How do you compare AI assistants, filters, and search?
Each tool responds to a different context. The right journey does not replace everything with AI: it simply directs the customer to the interface best suited to their level of certainty.
Internal search
It works very well when the customer knows a name, a reference, or a brand.
Filters
They are effective for an expert customer who already knows which criteria to select.
AI Assistant
It is best when the customer describes a usage, a problem, or a constraint without knowing the right attributes.
The best experience combines all three: search for precise queries, filters to refine, and an assistant to guide vague needs.
What architecture should be planned?
A reliable catalog assistant must be connected to product data. A generic model that does not know the stock or variants risks recommending unavailable products.
Necessary components
Step 1: A widget or choices assistant page.
Step 2: A dialogue engine to collect needs.
Step 3: A catalog index with products, variants, prices, stock, and tags.
Step 4: A scoring system to rank relevant products.
Step 5: A short presentation with three to five recommendations.
Step 6: A fallback if no product truly matches.
The catalog must be synchronized regularly, especially if stocks change frequently.
How do we design the dialogue?
Dialogue should help, not interrogate. It is better to ask a few useful questions than a long form disguised as a chat.
Recommended sequence
Ask what the customer is looking for.
Understand the usage or context.
Ask for important constraints: budget, size, deadline, compatibility.
Add an optional preference: style, material, color, or brand.
Reformulate the criteria before recommending.
Offer a short selection with the reason for the choice.
A good rule of thumb: one question at a time, with quick answers when possible.
How to recommend the right products?
The recommendation must be based on explicit criteria, not just on the best-selling products.
Useful criteria
Match with the expressed usage.
Customer's budget.
In-stock availability.
Compatibility, size, material, or technical constraint.
Popularity or reviews, only as a secondary criterion.
The first response should not display twenty products. Three to five options are sufficient, with a sentence explaining why each product matches.
Where should the assistant be placed?
Placement determines adoption. The assistant must appear in places where the customer truly hesitates.
At the top of major collection pages.
In search results when no results are found.
In the menu under a "Help me choose" entry.
On a dedicated page for complex journeys.
On mobile in the form of a discreet button, never as a blocking popup.
The customer must be able to launch the assistant when they feel the need, without being interrupted on every page.
What are some examples by sector?
Questions must change depending on the type of catalog.
Fashion: occasion, size, fit, season, and budget.
Cosmetics: skin type, goal, ingredients to avoid.
Tech or spare parts: model, compatibility, and usage.
Home: dimensions, style, room, and color.
B2B: volume, business constraints, and deadline.
An effective assistant adopts the customer's language, not just the internal taxonomy of the catalog.
How do I integrate it with filters and search?
The assistant must work with existing tools. It should not create an isolated journey.
Useful integrations
When a search yields zero results, suggest describing the need.
After a recommendation, open a pre-filtered collection.
Transform conversation criteria into visible filters.
Analyze requests with no results to enrich product tags.
This loop improves both the customer experience and merchandising quality.
Which KPIs to measure?
A catalog assistant must be judged on its ability to help the customer move toward a choice.
Launch rate: share of visitors who use the assistant.
Completion rate: share of users who go all the way to the recommendations.
Add to cart after assistance.
Conversion of assisted sessions.
Zero result rate: expressed needs with no matching product.
Decrease in "I can't find it" tickets.
How does Qstomy help with large catalogs?
Qstomy can connect the assistant to Shopify products, collections, variants, and inventory to recommend available and relevant products.
Key Capabilities
Configurable helper-to-choose dialogue.
Synchronization with Shopify products and metafields.
Short recommendations with justification.
Product cards within the conversation.
Redirection to the cart or a pre-filtered collection.
Explore the AI sales agent, AI support or request a demo.
Playbook 1: top collection tag audit
On your #1 traffic collection, check completeness of usage / budget / fit tags on hero SKUs. Target > 90% before assistant launch. List 10 "I can't find" questions from 30-day tickets.
Playbook 2: 5-question funnel
Document 5-step funnel for a mega-collection. Test mobile mystery shop: find niche product in 2 min without assistant vs with. Time the delta.
Playbook 3: collection placement
Deploy "Find your product" CTA on header of poorest converting PLP collection. Quick replies for budget + usage. GA4 assist funnel events configured.
Playbook 4: zero-result search hook
Internal search 0 results → "Describe differently" chat invitation. Measure weekly zero-match rate → merchandising tags backlog (#108).
Playbook 5: 21-day A/B test
50% collection traffic with assistant CTA vs control with filters only. KPIs: PLP conversion, assist-to-ATC, revenue per session. Prove ROI on one collection before site rollout.
Useful links
A large catalog without a conversational guide externalizes navigation to the customer or support: the assistant transforms 800 SKUs into a dialogue, not a filter labyrinth.
Enzo
June 28, 2026
