New import

Load a file of rows to segment. Several rows per customer is expected.

File

CSV, TSV, TXT or XLSX ยท up to 200 MB
Cancel

What happens next

  1. 1
    Profile Every column is scanned: type, missing values, spread, common values.
  2. 2
    Suggest Each column gets a proposed role (ID, timestamp, feature, outcome), type, roll-up and transform, with the reason shown.
  3. 3
    Confirm You adjust anything that's wrong and save. Every save is kept as a version.

Tip: one row per event (page view, cart add, payment) with a customer reference and a timestamp gives the most to work with. Outcome columns such as purchased or fraud can be in the same file.