Product data powers recommendations. A dataset should contain the products an Advice flow may recommend, plus the fields needed for matching.
For a product finder or product selector, start with the fields that change the recommendation: use case, compatibility, size, budget, stock, material or any other decision factor shoppers understand. See product finder software for the public buying-journey context.
Where to configure this
- Product data: create datasets, import feeds and review products.
- Mapping: connect feed fields to Product ID, Title, URL, Image and Price.
- Properties: review fields that can be used for matching.
- Advice flows -> Matching: connect a dataset to a flow.
Prepare the feed
Use a feed URL or upload a CSV, XML or JSON file. Keep Product IDs stable, because they are used for updates, recommendations and conversion measurement.
Recommended fields:
- Product ID, Title and URL.
- Image, Price and Sale price.
- Category, brand, use case, size, color, stock status or other matching fields.
- Review rating and review count when you want stars on product cards.
Create a dataset
- Open Product data.
- Create a dataset.
- Add the feed URL or upload a file.
- Check the detected fields.
- Save and open Mapping.
With the free plan you can import manually. Automatic sync is available after choosing a paid plan. Datasets with a feed URL sync every 6 hours by default; uploaded files stay manual.
Map fields
Map the required fields first: Product ID, Title and URL. Then map Image, Price, Sale price, Label and review fields when available.
Use clear matching fields. For example: category, brand, material, size, stock, usage or compatibility.
Add locale product data
Add a locale feed when visitors should see translated titles, links, prices or answer values. Validate the feed first; the field picker then shows the fields found in that feed and suggests the basic mappings.
Keep the Product ID field connected to the same ID as the source feed. For extra answer fields, map the source dataset field to the matching field in the locale feed. Matching keeps using the source value while visitors see the localized value.
Use properties in flows
Dataset properties can be used for:
- answer rules in Matching;
- Product filters for a whole Advice flow;
- dropdown answers with many values;
- swatches for colors or small visual choices.
After changing product data, publish the Advice flow again when visitors should use the updated recommendations.
Check after import
- Required mappings are complete.
- Active products have title, URL, image and price.
- Matching fields show expected example values.
- Review stars appear only when review rating is mapped.
- Import history shows no errors.
If a field looks empty after import, check whether the mapping points to the right feed field.