1. Introduction
Sensefuel analyzes customer behavior in real time and aggregates this data over time in order to identify purchasing habits, such as products that are frequently bought together.
The goal of Sensefuel is to automatically recommend relevant products that are commonly purchased together based on this learning.
When deploying Sensefuel, it is possible to bootstrap (initialize) the recommendation algorithms using historical sales data from your physical stores or online channels.
If you are already a Sensefuel customer, this process allows you to further enrich the system’s learning and improve recommendation accuracy.
2. File Format
File name: sales.csv
Format: CSV file containing a header row and one row per sale transaction.
Header:sales_id | product_ids | date
Field separator: |
2.1 Field Description
| Field name | Description | Type |
|---|---|---|
| sales_id | Unique identifier for the sale transaction. | String |
| product_ids | Identifiers of the products sold within the transaction. The product identifier depends on whether product variants are used: • With variants: provide the grouping identifier corresponding to the item_group_id attribute from the Sensefuel catalog feed. • Without variants: provide the product identifier corresponding to the id attribute from the Sensefuel catalog feed. | String containing product identifiers separated by ; (no spaces before or after the separator) |
| date | Sale date and time. | UTC date in ISO 8601 format, using the pattern YYYY-MM-DDThh:mm:ssZ |
2.2 Example
sales_id|product_ids|date 12345678|111;4487;54|2025-02-18T15:52:21Z 89876421|28945;98451|2025-01-31T10:22:48Z