Product Assistant

Product Assistant is a guided question flow that helps shoppers quickly find relevant products on listing pages. It is not just a visual widget: setup quality directly affects result relevance and conversion quality.

This page focuses on setup and matching logic. Analytics interpretation is covered on a separate analytics page.

How to set up Product Assistant

Setup is done in the admin Product assistants section:

1. Open Admin -> Product assistants.
2. Click Create assistant.
3. Fill in the basics: internal name, customer-facing title, subtitle, CTA text.
4. Choose placement (category or product listing) and optionally a specific category.
5. Set max results, sort order, and enable active.
6. Add steps (questions) and answer options.
7. For each option, configure matching rules (tags, categories, attributes, brand, price, and optionally specific product IDs).
8. Save the assistant.

After saving, run a quick quality check:

• test the assistant on the storefront for relevance,
• confirm that answers clearly change the shown products,
• check Product assistant analytics after first real interactions.

Core logic you should understand first

Each answer option acts as a product filter or preference signal. Better rule design means better shopper outcomes.

Before you build steps, align on:

• the business purpose of each question (intent, budget, use case),
• clear separation between answer options,
• enough precision to be relevant, but not so strict that results collapse.

If rules are too broad, recommendations feel generic. If rules are too narrow, customers may see very few results.

Designing effective questions and options

Recommended sequence:

1. Start with the strongest decision factor (product type, purpose, category).
2. Use middle steps for refinement (price band, brand, core features).
3. Use final steps for fine-tuning, not full re-segmentation.

Practical recommendations:

• use shopper language, not internal naming,
• validate each option against real product sets,
• avoid two options that return nearly identical products.

Matching rules in practice

Each option can combine multiple rule types:

• Tags — Broad thematic grouping (for example eco, premium, gift).
• Categories — Strong catalog narrowing.
• Attributes and values — Precise product characteristics.
• Brand — Manufacturer preference.
• Score boost — Increases a product's suitability score by X percent (does not filter, but ranks matching products higher).
• Price min/max — Price range control.
• Product IDs — Direct manual product targeting.

Use rules intentionally. More rules do not automatically mean better results.

Good vs poor configuration examples

Question: “What will you use it for?”

• Good configuration: options like “Daily use”, “Professional use”, “Gift”, each mapped to distinct product logic.
• Poor configuration: vague options with near-identical product output.

Question: “What is your budget?”

• Good configuration: price ranges that match your actual assortment distribution.
• Poor configuration: ranges where medium and high options return almost the same products.

What the Product Assistant does for shoppers

The Product Assistant shows customers a short question flow and narrows products based on their answers.

Typical customer journey:

1. Customer sees the assistant on a listing page.
2. Customer starts answering questions.
3. Suggested products update while they continue.
4. Final results are shown.
5. Customer can click a product or add it to cart.

Common configuration issues

Results are not relevant

Check:

• questions are aligned with real buying decisions,
• option rules are not too generic,
• multiple options are not pointing to almost identical product sets.

Too few results are shown

Check:

• price ranges are not too restrictive,
• one option is not over-constrained by many combined rules,
• your products actually contain the tags/attributes used in rules.

Assistant is used, but conversion impact is weak

Check:

• question order (high-impact decision first),
• clarity of question and option wording,
• CTA text and progression motivation.

Privacy note

Assistant analytics may process session-level behavioral data (for example assistant usage events and product actions). Include this feature in your privacy and cookie documentation where required by local regulations.