It is important for customer-centric organizations to have as accurate and complete customer records as possible. However, it is fairly common for organizations, especially those with very large number of customers, to have duplicate records for the same person. Duplicate customer records occur when an agent or the customers themselves open a new account for an existing customer. Duplicate records impact an organization’s bottom line in a number of ways:
- Wasted marketing dollars (e.g., duplicate mailings sent to the same customer)
- Ineffective marketing due to information being incomplete and spread across multiple records
- Inefficient and frustrating customer service. When contacting customer support to follow up on an issue, customers expect customer service representatives to be able to find their account quickly and previous interactions to be captured in a single account.
- Abuse if an organization can’t detect that a given customer has already benefited from a sales promotion or rebate.
Why basic entity matching isn’t enough
Detecting duplicate records is hardly ever just a matter of detecting identical or near-identical records. Customer records typically consist of several fields such as name, mailing address, date of birth, place of birth, mother’s maiden name, phone number, and email address. Duplicate records may differ in multiple ways. For instance, the name values may differ in spelling, word ordering or the use of nicknames; the phone number and email address may have conflicting values, the mailing addresses may match only partially, the dates may come in different formats. In addition, international corporations (e.g., large hotel chains) may have records in different languages or scripts (e.g., English, Arabic, Chinese, Cyrillic).
So how can an organization detect duplicate customer records accurately or avoid them in the first place? And how can it do so efficiently even when the customer database contains tens of millions of records?
Managing customer records with advanced entity matching software
Advanced entity matching or identity resolution software enables organizations to detect duplicates in their customer records accurately and efficiently. Records are indexed and each field is assigned a weight. For instance, the name field is typically given a higher weight than a phone number field because phone numbers are more likely to change and may be shared by multiple people (e.g., family members). Given a query to check for an existing or new customer, the system searches for possible matches and combines the evidence from the multiple fields relative to their weight into a similarity score.
NetOwl EntityMatcher leverages probabilistic machine learning and is especially designed for Big Data environments. It performs identity resolution against tens of millions of entity records with sub-second response time. It has been trained on large-scale, real-world, multi-ethnicity data and supports matching across languages. It allows parameter tuning and additions of custom rules and dictionaries.
Identity resolution software can help protect your organization from fraud, abuse, and waste. For more information on how entity matching and identity resolution software can improve your customer record management, contact NetOwl today.
