Find and Merge Duplicate Contacts
Duplicates can appear in your data for many reasons, including mistakes by members of staff and also when you let members of the public enter data. CiviCRM has duplicate matching rules to minimise the chance of this happening, but when it does happen, it provides a set of tools for finding and merging duplicate contacts. These are accessed through the admin interface (Contacts > Find and merge duplicate contacts).
You can set up different rules for identifying suspected duplicates. When you use these rules (by clicking on Use Rule), Suspected Duplicates are shown side by side in a table.
Clicking on merge brings up a table showing details of each contact. For each detail you can choose whether to keep the original data (shown on the right) or copy the duplicate data.
Note associated tags, groups and activity data (including event attendance, contributions, etc.) will appear in addition to data already recorded in the original record, not in place of it.
Unsupervised rules vs Supervised rules
There are two categories of dedupe rules - called 'Unsupervised' and 'Supervised' (as of 4.3). Unsupervised rules should be configured with a relatively tight definition of what constitutes a match (for example... email address + first and last name for Individuals). The default Unsupervised rule is automatically used when new contacts are created (including contacts created by constituents during online contributions, etc.). They are also used when you Import contacts. You can only have one default Unsupervised rule for each contact type.
Supervised rules should be configured with a looser definition of what constitutes a 'match'. The default Supervised rule is automatically used to check for possible duplicates when contacts are added or edited via the user interface. You'll probably also want to use a Supervised rule when scanning your database for possible duplicates. Click Use Rule to scan for duplicate contacts (of that contact type) using the selected rule.
You can configure up to five fields to evaluate when searching for 'suspected' duplicate contact records of a particular contact type. For each field, set a numeric weight which determines the relative importance of a match on that field.
You can also set a length value which determines how many characters in the field should be compared.
EXAMPLE: If you set a length of 8 on 'Street Address', then '101 Delaplane Ave.' would match '101 Delaplain St.', because the first 8 characters are the same. If length is left blank, then the comparison is done on the entire field value.
Finally, you set a numerical threshold for the rule. When the total of the weights of each match is equal to or greater than this threshold - the contacts will be listed as suspected duplicates by the Find and Merge Duplicate Contacts task.
Possible Duplicate Contact Alert Emails
When a new organization contact is created via an online contribution form using the "On behalf of" functionality and CiviCRM detects a possible duplicate contact using the default Strict Organization dedupe rule, CiviCRM sends out an alert email.
The email, with a default subject line of "CiviContribute Alert: Possible Duplicate Contact Record", is sent to the Email Receipt From field of the online Contribution Page Thank-You and Receipting configuration . In 3.2.3 the related contact is also sent a copy of a message, but there is an open issue to limit the recipient to only the form's administrator as set in Email Receipt From.
The subject and body of this message can be customized in System Workflow Messages tab of the Message Templates administration, specifically CiviCRM › Administer CiviCRM › Message Templates > System Workflow Messages > Contributions - Duplicate Organization Alert.