To use the Recommendations tool, it is necessary to first upload data on which recommendations can be based. Transaction data can be uploaded and saved as a dataset. Each dataset contains multiple transactions with different products/items at each line.

It is also possible to push this data via API. We recommend this method for real-time integrations.

Data should be uploaded as a CSV file. The CSV should contain transaction, item, and time column names in the first row. You can use any of the unique identifiers of your products or documents as an item in data set, which field is used will be set in the configuration section. An SKU of the produce or the document’s ID automatically generated by AddSearch would be a natural choice. Headers in the CSV should be lowercase.

transaction,item,time
1,229,2023-04-04T06:34+02:00
1,126,2023-04-04T06:34+02:00
2,229,2023-04-04T07:31+02:00
2,256,2023-04-04T07:31+02:00
3,117,2023-04-04T08:34+02:00

Please note that CSV files may not exceed 1MB.