Fuzzy Data Matching with SQL

Chapter 7. Phone Numbers

With the death of using tax IDs for most data matching (we will talk a bit about them at the end of this chapter), phone numbers, especially mobile phone numbers, are about as close to being a publicly available unique identifier as we’ll ever be able to access. Yes, like email addresses, some people share phone numbers, but the percentage is small and, with the ongoing expansion of mobile phone use, dwindling.

We will look at various issues with phone numbers, including formatting (of course), lack of information (area code, country code), too much information (notes tacked on the end of the number), and the fact that there sure are a lot of them. Does anyone still have a pager? Should you check against it? Let’s talk about all of that!

What Makes a “Phone Number”?

By this point you should recognize the drill. If you think about a field or fields you’re trying to match on for more than a few seconds, you can immediately start to think of things that will get in the way of that. Phone numbers are no exception. Consider the following:

• [phone]

• [phone]

• 800-555-1234

• 1-800-555-1234

• 8005551234

• 18005551234

• [phone] Ext. 67 (How does your data handle extensions? How does the incoming data represent it?)

• 8005551234,,67 (Many modern mobile phones and phone systems still understand and can dial old-style modem “AT commands,” in this case to pause a few seconds after connecting to [phone] and then dial 67.)

• 555-555-1234 ...