Skip to content
All work
Data Quality· Enterprise sales organisation· 2023· ~10 weeks

CRM Data Quality Framework

'The reports are wrong' was always a data problem, never a reporting one. Built the framework to prove it and fix it.

  • Salesforce
  • Power BI
  • Power Query
  • Python (pandas)
  • Duplicate accounts
    −60%
    on the core book of business
  • Key-field completeness
    +28pp
    on industry, segment, country
  • DQI
    built + adopted
    reviewed weekly by the team

Business problem

Every reporting issue the commercial team raised ended up in the same place: duplicates, empty fields, inconsistent picklists. Leadership was losing trust in the numbers but kept aiming at the dashboards.

Data available

  • Salesforce Account, Contact, Opportunity — full snapshot
  • Third-party enrichment sample for validation
  • Field usage statistics per team and region

Solution implemented

  1. 1Profiled the CRM with reproducible SQL + pandas — quantified duplicates, completeness and conformity per field, per team.
  2. 2Defined a Data Quality Index (DQI) weighted by the impact each field has on revenue workflows — not by how 'full' it looks.
  3. 3Built dedup and standardisation rules (industry, country, phone) partly automated in Salesforce, partly in a clean-up playbook for sales ops.
  4. 4Published the DQI in Power BI and made it the opening slide of the weekly ops meeting — with owners per failing dimension.

Impact

Duplicate accounts
−60%
on the core book of business
Key-field completeness
+28pp
on industry, segment, country
DQI
built + adopted
reviewed weekly by the team

What I learned

The report is never the problem. It is always the field upstream — and you only fix it once there is a scoreboard.
Previous case study
Power BI Executive Cockpit over Salesforce
Next case study
Process Mining on the Salesforce Sales Funnel with Fluxicon Disco