Address matching

The 'Address matching' screen is used to specify the duplicate detection method that will apply to all of your publications.

QuickFill uses a match code that consists of, in this order, the country code, the zip code (or postal code), the " soundex" of the customer's last name, and the "soundex" of the customer's company name. By looking at these fields and at the "soundex" codes, QuickFill determines whether or not a customer being entered is a duplicate of an existing customer. As explained below, this match code can be changed.

You can control which parts of the match code are used in checking for duplicates. By default QuickFill uses the country code, zip/postal code, soundex of the last name, and soundex of the company name—two customer records will be considered possible duplicates if their match codes match. If you wish you can widen the scope of the duplicate detection "net" by ignoring the company or the company and the last name soundex code. You make this choice by selecting 'Country code and postal code only' (described below) as the minimum match requirement.

When searching for duplicates QuickFill first determines if the customer you are entering is a possible match of an existing customer based on the match code; this process is described above. If a possible match is found, a score of 100 points is assigned and the second part of the duplicate detection process is performed. In this part, QuickFill uses each of the components that you define on the 'Address matching' screen to compare the possible matches.

A score is assigned based on how well these address components match. Points are deducted from the maximum score of 100 for each component such as the first name, street number, or post office box that does not match. The 'Address matching' screen provides the ability for you to define the number of points to be deducted as well as the minimum score for a customer to be considered a near or an exact match. QuickFill's address matching is not case sensitive. For example, "BOX" and "Box" are considered the same.

You can control the points that are deducted for each component by changing the value in the field next to each of the criteria listed on the 'Address matching' tabs.

• Warning: do not select the "Country code and postal code only" option if you have large numbers of customers concentrated in a small geographic area, say for a city or regional magazine. If you have hundreds of customers all in a single zip code, QuickFill will take a very long time to perform duplicate detection for customers in that zip code.

• Note: If you have reversed the sequence of the last name and company name soundex codes by selecting the "Zip/Company/Name" option on the Lookup tab of the Preferences screen, then your choices on this screen are also reversed. Your choices are "Country code and postal code only", "Above plus soundex of company" and "Above plus soundex of last name".

• Note: bad debt customers. In order to catch customers that request a "bill-me" subscription using false aliases, special rules are used if a customer has 'B' ("bad debt") in the 'Type' field. Bad debt records will be included as possible matches if the country code and postal code match, regardless of what "minimum match requirement" field you select. Furthermore, only the street address is used during scoring. No points are deducted for mismatched names, no matter how different the names are.

Minimum match requirement...

Use these fields to control the size of the candidate pool (the number of possible matches).

Selecting 'Country code and postal code only' gives the largest pool size. That is, if a customer's country code and postal code match that of another customer, QuickFill will do a complete check of both customer records. The suspected duplicate customer will be assigned a score based on the components of their name and address that matched the other customer's and the values assigned to each of those components. If the score is greater than or equal to the 'Minimum score for a "near" match', the customer records will be displayed as possible duplicates.

Selecting 'Above plus soundex of last name' gives a smaller candidate pool size. QuickFill will only perform a complete check of possible duplicates if their country codes, postal codes, and soundex of their last names match each other. Again, a complete check is done and a score is assigned. If the score is greater than or equal to the 'Minimum score for a "near" match', the customer records will be displayed as possible duplicates.

Selecting 'Above plus soundex of company name' gives the smallest pool size. If you have a very large database with large numbers of names in densely populated areas such as Manhattan, you should select 'above plus soundex of company'. QuickFill will only perform a complete check of possible duplicates if their country codes, postal codes, and the soundex of their last names and company names match each other.

Minimum score for a "near" match

Candidates with scores equal to or greater than this value will be displayed so that you can decide whether they really are the same person. Use caution when changing this value. Set it too low and QuickFill will show you too many names. Set it too high and you will end up with multiple records in the database for the same person. For example, let's look at the following customers:

If the value for 'Minimum score for a "near" match' was set to 80, these customer records would not be considered duplicates. Based upon the default values defined on the 'Address matching' screen, the following is a description of how QuickFill would view these two customers.

Again, based on a 'Minimum score for a "near" match' of 80 and the other default values defined on the 'Address matching' screen, the customers above would not be considered duplicates because they received a total score of 75 points; the starting score of 100 minus the total points deducted (25).

Minimum score for an "exact" match

This value is only used with subscription and prospect import programs. (Click here for more details on importing prospects and importing subscriptions.) Incoming customers that match existing customers with a score equal to or greater than this value will automatically be merged. Use extreme caution when changing this value. Set it too low and you risk combining the customer records of separate persons. Remember: points are deducted for components of the customer records that do not match. The higher that a customer's score is, the more likely it is that they are a duplicate.

Values to deduct…

Last name soundex codes do not match

This value is deducted from the score if the last names do not sound alike. This value is used only if you checked the 'Country code and postal code only' radio button for the "minimum match requirement." In the other two cases the last name must sound alike for the record to be placed in the match candidate pool.

Soundex codes match, actual last names do not

This value is deducted from the score if the names sound alike but are spelled differently (e.g., "Bean" and "Bain").

Initial letters of first names do not match

This value is deducted from the score if the first name has an initial letter that doesn't match that of the customer being entered. (e.g., "Ben" and "Jerry"). Don't assign too high a value to this test since some mismatches are misleading (e.g., "Bob" and "Robert").

One has full name, other has matching initial

This value is deducted from the score if one record has a first name and the other has only an initial, and the initial matches the name (e.g., "J" and "John").

Initials match, first name soundex does not

This value is deducted from the score if the first names have matching initials but they do not sound alike (e.g., "Andrew" and "Adam").

Soundex codes match, actual first names do not

This value is deducted from the score if the first names sound alike but they are spelled differently (e.g., "Sydney" and "Sidney").

Both names have prefix, but they do not match

This value is deducted from the score if both names have a prefix such as "Mr." or "Mrs." but they do not match. It is important not to set this value to zero. If you do, QuickFill will think that "Mr. John Smith" and "Mrs. John Smith" are an exact match.

Both names have suffix, but they do not match

This value is deducted from the score if both names have a suffix such as "Jr." or "II" but they do not match. It is important not to set this value to zero. If you do, QuickFill will think that "John Jones Sr." and "John Jones Jr." are an exact match.

Both addresses have sex code, but they do not match

This value is deducted from the score if the sex codes in the two customer records differ. Nothing is deducted if either or both of the sex codes is blank.

Titles do not match

This value is deducted from the score if the title fields do not match, but only if neither record contains a last name. The purpose of this is to avoid assigning 100 percent match scores to records that contain identical company names and addresses but that are directed to different positions (e.g., "Sales Manager, World Wide Widgets" and "Comptroller, World Wide Widgets").

Departments do not match

This value is deducted from the score if the department fields do not match, but only if neither record contains a last name. The purpose of this is to avoid assigning 100 percent match scores to records that contain identical company names and addresses but that are directed to different departments (e.g., "Marketing Department, World Wide Widgets" and "Accounting Department, World Wide Widgets").

Only one address has a Zip+4

This value is deducted from the score if one record has only a five digit zip code and the other has a nine digit zip code. Remember that at least the first five digits of the zip code must match for the records to even be considered. This value does not apply for addresses outside the U.S.A.

Both have Zip+4, but they do not match

This value is deducted from the score if both records have nine digit zip codes but the last four digits do not match. Remember that at least the first five digits of the zip code must match for the records to even be considered. This value does not apply for addresses outside the U.S.A.

Only one address has a PO Box

This value is deducted from the score if one record has a post office box and the other does not.

Both have PO Boxes, but they do not match

This value is deducted from the score if both records have post office box numbers but they do not match.

QuickFill looks for post office box numbers on both street address lines. For example, let's say one of your customers, James Rutherford, has a street address on the first address line and a post office box on the second address line. Now you enter J. Rutherford with the same address except the post office box is on the first line and the street address is on the second line. QuickFill will recognize J. Rutherford as a possible duplicate of James Rutherford.

To be recognized as a post office box the number must be preceded by "PO", "POB", "PO BOX", "BOX", or "POST OFFICE BOX". Periods, as in "P.O.", are ignored.

When the addresses have matching post office box numbers QuickFill modifies the rules used for matching street names. It does not deduct any points if one address has a house number, street name, or direction and the other does not. This means that "100 MAIN ST, PO BOX 100" and "PO BOX 100" are considered to be identical.

Only one address has a house number

This value is deducted from the score if only one record has a house number. QuickFill looks for the house number on the first street address line only.

Both have house numbers, but they do not match

This value is deducted from the score if both records have a house number but they do not match.

Only one address has a street name

This value is deducted from the score if only one record has a street name. Generally this occurs if one record has a post office box number and the other has a street address.

Street name soundex codes do not match

This value is deducted from the score if the street names do not sound alike.

Soundex codes match, actual street names do not

This value is deducted from the score if the street names sound alike but are spelled differently.

Only one address has an apartment/unit/ suite number

This value is deducted from the score if only one record has an apartment number, unit number, suite number, or floor number. QuickFill assumes that any sequence of digits that follows the street name is an apartment, unit, suite, or floor number.

Apartment/unit/suite number does not match

This value is deducted from the score if both records have apartment, suite, unit, or floor numbers but they do not match.

Only one address has a prefix direction

This value is deducted from the score if only one record has a prefix direction. A prefix direction is one of the eight points of the compass, which appears before the street name (e.g., "100 NORTH MAIN ST").

Prefix directions do not match

This value is deducted from the score if both records have a prefix direction but they do not match. Abbreviated directions ("N" and "NORTH") are considered to match.

Only one address has a suffix direction

This value is deducted from the score if only one record has a suffix direction. A suffix direction is one of the eight points of the compass, which appears after the street name (e.g., "100 CONNECTICUT AVE NW").

Suffix directions do not match

This value is deducted from the score if both records have a suffix direction but they do not match. Abbreviated directions ("N" and "NORTH") are considered to match.

Only one address has a company name

This value is deducted from the score if only one record has a company name.

Company soundex codes do not match

This value is deducted from the score if both records have a company name but the first words of each name do not sound alike.

• If the company name begins with "THE," the second word is used instead.

Soundex codes match, actual company names do not

This value is deducted from the score if the first words of the company names sound alike but the full company names are not identical.

Label images do not match, ignoring punctuation

This value is deducted from the score if the mailing labels for the two records are not the same, ignoring differences in punctuation and spacing. By default this value is set to zero so that differences in the mailing labels have no effect on the score. If you require that exact matches must have identical label images, then set this value to 1 and set the minimum score for an exact match to 100. You should also reduce the minimum score for near matches by 1, since near matches will never have identical label images.