Where: open your Advice flow → Matching.
Matching translates a visitor's answer into a product property. For example, I want to go hiking fits products whose category field has the value Hiking shoes.
First make sure the flow has product data, questions and a result step.
1. Choose the products for this flow
Check the connected dataset. Under Filters & Sorting, use Product filter to select only shoes from your catalog, for example.
- Include matching products: only products meeting the filter can participate.
- Exclude matching products: products meeting the filter cannot participate.
Check the product count before continuing. A shoe filter should exclude jackets without accidentally excluding every product.
2. Create a rule for an answer
- Open Questions, choose a question and open an answer.
- Add a rule and select the dataset field, such as
category. - Choose is to match an exact category name.
- Select or enter a value that actually appears in your product data, such as
Hiking shoes. - Choose the rule type and check the matching products after saving.
The customer-facing text can be friendlier than the data value. I want to go hiking can refer to the category Hiking shoes. If your dataset uses a different name for the category field, select that field. If unsure, check the field names and sample values under Product data → your dataset → Properties & Swatches.
3. Choose the rule type
| Rule type in the app | What happens? | Example |
|---|---|---|
| MATCH | Suitable products earn points; other products remain possible. | A lightweight shoe is preferred. |
| EXACT MATCH | Products that do not meet the rule are excluded. | The shoe must be waterproof. |
| BOOST | Suitable products receive extra score. | Give an already suitable product more prominence. |
Start with MATCH. Use EXACT MATCH for genuine requirements, such as compatibility. No preference can be left without a rule.
Combine several rules using Match all rules when every requirement must match, or Match any rule when one match is enough. Keep the first setup simple; adjust individual question weights later.
4. Example: three questions for hiking shoes
Use the four fictional products from Prepare a product feed. The CSV, XML and JSON examples contain the same products and attributes. Import one of these formats into a dataset and connect it to your Advice flow.
Under Filters & Sorting → Product filter, select Include matching products, field category, comparison is, value Hiking shoes. This gives every question the same starting selection of four shoes.
In the Builder, create the three questions below with one answer per question. Connect them in this order and connect the third question to a result step. Enable product recommendations under Settings → Recommendations. Then configure the rules below for each answer under Matching → Questions. Keep the default weights and do not add boosts for this example.
Question 1: Where will you mainly walk?
| Answer | Field | Comparison | Value | Rule type |
|---|---|---|---|---|
| In the forest | terrain |
is | forest |
EXACT MATCH |
| In the mountains | terrain |
is | mountains |
EXACT MATCH |
| In the city | terrain |
is | city |
EXACT MATCH |
Here, terrain determines which products can participate. After In the forest, Waterproof Forest Hiking Shoe and Breathable Forest Hiking Shoe remain. This example uses one terrain per product; check how suitability is recorded in your own catalog.
Question 2: Do your shoes need to be waterproof?
| Answer | Field | Comparison | Value | Rule type |
|---|---|---|---|---|
| Yes, that is required | waterproof |
is | yes |
EXACT MATCH |
| No preference | n/a | n/a | n/a | No rule |
No preference allows both waterproof and non-waterproof shoes to remain. Do not create a waterproof is no rule for this answer. Choosing Yes, that is required after In the forest leaves only Waterproof Forest Hiking Shoe.
Question 3: What fit do you prefer?
| Answer | Field | Comparison | Value | Rule type |
|---|---|---|---|---|
| Preferably wide | fit |
is | wide |
MATCH |
| No preference | n/a | n/a | n/a | No rule |
Here, wide is a preference: wide shoes receive additional matching points while other fits remain possible. With In the forest → No preference → Preferably wide, Breathable Forest Hiking Shoe ranks above Waterproof Forest Hiking Shoe. If a wide fit is necessary, word the answer as a requirement and use EXACT MATCH.
Check these answer paths in Test
| Selected answers | Expected advice with the example feed |
|---|---|
| In the forest → No preference → Preferably wide | Breathable Forest ranks first; Waterproof Forest remains available. |
| In the forest → Yes, that is required → Preferably wide | Only Waterproof Forest. The waterproof requirement takes priority over the preference for a wide fit. |
| In the mountains → Yes, that is required → Preferably wide | Only High Mountain Boot. |
With In the city → Yes, that is required → No preference, no product fits: the example feed has no waterproof city shoe. That reflects the product selection and is not a reason to remove the waterproof requirement.
5. Explain the advice
Add a short USP label where useful, such as Waterproof. This helps visitors see why a product fits. Not every technical comparison needs its own label.
Using answers from a dataset attribute? Also check matching for that question. An automatically populated answer list does not replace testing the recommendations.
6. Test two different situations
Open Test and select a realistic answer path. Review the first recommendations and their score explanations. Repeat with clearly different needs.
| What you see | Check first |
|---|---|
| No products | The Product filter and EXACT MATCH rules. |
| Wrong products | The field, comparison and value for the selected answer. |
| Almost the same order every time | Whether rules match different products and questions distinguish shoppers' needs. |
| Suitable products in an odd order | Weights, boosts and sorting for equal scores. |
Do not relax a necessary safety or compatibility requirement just to get results. Check whether the catalog contains a suitable product instead.
7. Configure product cards
Go to Settings → Recommendations. Enable product recommendations. New flows start with three products per page and three pages; adjust this to your advice.
Also choose whether to show review stars, advice labels and non-matching labels. Review stars require a mapped review rating.
You are ready when: the first products fit both test situations and you can explain their ranking. Check Preview and publish the flow.
Later: boosts for the whole flow
With a paid plan, use Boosts to give suitable products extra score across the entire flow. Answer matching, question weights and product filters remain available on Free. Set the conditions, boost percentage and an optional label. A boost does not remove products and should not hide errors in filters or answer rules.
Alerts after a product update
A product update can leave a published Advice flow's Product Filter without matching products. You receive an alert on any plan.
- Open View Product Filter and compare its conditions with your current product data.
- Adjust the filter and publish your changes.
- Choose Check again. The alert clears when the published selection contains products again.
Opening the alert or only saving the draft does not resolve it. This check covers the product selection, not every answer path. Manage email alerts in your account preferences.
