You are here:
Tips for Improving Account Matching
Creating rules with overly broad criteria can result in poor account matching. Consider these guidelines when setting up identity resolution match rules for unifying accounts.
Limit Geographical Location
For example, a match rule with these criteria is unlikely to result in good matches.
- (Fuzzy Match - High Precision) Account Name
- (Normalized) State
- (Normalized) Country
Overly broad criteria can cause two seemingly conflicting problems.
- Too Many False Positives--In large datasets, overly broad matches with these criteria can try to evaluate every account in a particular state and country. This can lead to too many accounts being matched into a single group.
- Missed Matches--When too many source profiles match, some records can be skipped during processing. The records skipped can be different each time identity resolution runs because records are reviewed in arbitrary order. This can lead to matching records being missed.
You can reduce the number of records considered for matching by defining more specific geographic areas. For example, include city or postal code as match criteria rather than allowing matching across an entire state, province, or country. Reducing the geographical area leads to more reliable and precise matching.
Combine Multiple Match Rules
Each match rule you add to a ruleset provides an opportunity to match accounts in a different way. Allow for multiple sets of match criteria and match methods by setting up multiple match rules within a ruleset.
This set of five match rules allows for a variety of match options within a ruleset.
- Fuzzy Account Name and Normalized Address
- Fuzzy Account Name and Normalized Phone Number
- Account Identifier -- D-U-N-S Number
- Account Identifier -- External MDM Id
- Account Identifier -- Demandbase
