Data Cleaning Exercises

Capstone · Customer operations after a platform migration

Merge the CRM and the web shop customer lists

The company moved its web shop to a new platform. Customers now exist in the CRM and in the shop with different ids, inconsistent emails and spelling variants of names.

Produce one customer list with a documented survivor rule, a review queue for records that can't be matched safely, and clean consent flags.

Checkpoints

  1. 1. Emails normalized under the declared policy.

    Normalize emails under a declared policy
  2. 2. Customers merged on normalized email; review queue.

    Resolve customer records by normalized email
  3. 3. Consent flags parsed without treating missing as no.

    Keep identifiers as text and parse yes/no flags
  4. 4. Names normalized without changing letters.

    Normalize names without changing letters

Final quality report

Write it in your private notes below (saved to your account) or in your own document. It isn't auto-graded; use it to practise the handoff interviewers ask about.

  • Merge counts: survivors, merged, needs_review.
  • The survivor rule and why the CRM is the system of record.
  • Examples of pairs sent to review and what evidence would resolve them.
  • Risks of a false merge and how the process avoids them.

Sign in to bookmark this and keep private notes.