Why CRM data quality decays, and what actually holds it

CRM data is usually good on the day the system launches and unreliable within eighteen months. The decay is predictable, and it follows the same pattern in most organisations: the people entering data are not the people who need it to be accurate.

The three common causes

• Fields exist because a report was requested once, not because anyone uses the answer. Sellers learn which fields nobody checks, and those fields fill with defaults.
• Duplicate creation is easier than duplicate detection. If searching for an existing account takes longer than creating a new one, duplicates are the rational choice.
• Nobody owns the customer master record. Sales, marketing, finance and service each hold a partial version, and no process reconciles them.

Controls that pay for themselves

1. Duplicate detection rules configured against how records are actually created, including phonetic and partial matches on organisation names.
2. Required fields limited to those with a named consumer. If nobody can say who uses the value, remove the field.
3. Progressive capture: ask for information at the stage where the seller already knows it, not on record creation.
4. A visible data quality report by owner, reviewed in the same forum as pipeline. Quality improves when it is discussed where performance is discussed.

Ownership is the durable fix

Technical controls slow decay. They do not stop it. What stops it is a named owner for the customer master, an agreed definition of what makes a record complete, and a routine, monthly is usually enough, where exceptions are reviewed and corrected by the people who created them. That routine costs less than the reporting rework it prevents.