How To Create The Golden Master Record
Even after names, addresses, phone numbers, and emails are cleansed and verified, it is not easy to assemble all of the data for a single customer. This is because the records rarely match exactly.
Personal data changes. Customers move. She has multiple contact points. For example, she may provide a work phone number at one touchpoint and a mobile number at another. She may have multiple addresses (e.g., primary and vacation homes), multiple phone numbers (e.g., fixed landline and mobile), and multiple email addresses (e.g., personal and work). How do you handle all of this disparate information?
This is an especially large problem for omnichannel marketers because shoppers often enter different information into different systems and at different times. The result is that the same customer can appear multiple times in a database. Not only is it necessary to deduplicate the customer database, but, it is critical to match customer records to build a complete picture of your customer.
The answer is to build the “Golden Master Record.” A Golden Master Record is the single source of truth about your customers in your CRM, Single Customer View, or Master Data Management application. It contains everything we know about the customer or links to everything we know about the customer, and it is always kept up-to-date because it gets information from every data silo. With links back to the source data, nothing is ever lost.
Of course, the process is more complicated than that. How do you do this when the information disagrees with what was there before? How do you bring in additional information without forgetting the old information? How do you handle human errors?
We use a technique known as “cascading” to confidently build a complete Golden Master even when some attributes disagree and to fill in incomplete information. It works by examining all of the information that we have about the customer. With a big database, the cascading calculation becomes rather complex. However, the more information we have, the more confident we become.
Examine the table below to understand how cascading works for a small number of records.
Cascading: A Key Technique
Cascading uses information across multiple records to improve the accuracy of a match. Consider the following example:
- “Record 1” contains a name, postal address, and phone number.
- “Record 2” contains a name, postal address, phone number, and email.
- “Record 3” contains a name, email, and a different phone number.
Record 1 and Record 2 match because the name, address and phone number match.
Record 1 and Record 3 considered in isolation would not match because the phone number differs and there is no other information in common for comparison.
However, Record 2 and Record 3 would be a likely match because the name and email match, even though the phone number is different.
Since Record 1 and Record 2 match, and since Record 2 and Record 3 match, cascading allows us to consider Record 1 and Record 3 a match.
For omnichannel marketers, cascading done on a larger scale becomes an ideal way to reconcile information across the data silos.
The table below shows how cascading can be used to merge records in which the data does not match exactly. Each record is for Christiane Gellesch, but the information does not match. If you only compared two records in pairs, it would be impossible to know which record is correct. However, comparing multiple records allows you to build a complete picture of the customer and to create a Golden Master Record.
Attribute | Record 1 | Record 2 | Record 3 | Record 4 | Golden Master for Single Customer View |
---|---|---|---|---|---|
First Name | Christiane | Christiane | Chris | Christiane | Christiane |
Last Name | Gellesch | Gellesch | Gellesch | Gellesch | Gellesch |
Building | Avon House | Avon House | Avon House | ||
Number | 23 | 43 | 23 | 23 | |
Street | David Place | David Place | David Place | David Place | |
District | St. Helier | St. Helier | St. Helier | St. Helier | St. Helier |
City | Jersey | Jersey | Jersey | Jersey | |
Postcode | JE2 4TE | JE2 2TE | JE2 4TE | JE2 4TE | JE2 4TE |
Country | Channel Islands | Channel Islands | Channel Island | United Kingdom | Channel Island |
Country Code | +44 | +33 | +44 | +44 | |
Local Area Code | 0 | 0 | 0 | 0 | |
Area Code | 7509 | 7509 | 7509 | 7509 | |
Local Number | 312345 | 312345 | 312345 | 312345 | |
christiane @yahoo.co.uk | christiane @yahoo.co.uk | christiane @yahoo.co.uk | christiane @yahoo.co.uk |
Of course, your data will contain more information and more records about a customer.
Significant local knowledge regarding names, addresses, and other contact information is required. Matching tolerances must be established individually for each attribute. In some densely populated locations, such as Hong Kong, sub-premises detail is important. In some countries, the family name is actually the first name. Therefore, it is important to look for name swaps. The list goes on.