Rhythm flags likely duplicate contacts by comparing email addresses, names, and postal codes. Use this article to understand what makes a match likely, why some similar contacts do not appear, and why imports are not checked.
What Raises a Match
Rhythm scores each pair of contacts by combining a few kinds of evidence:
- A shared email address is strong evidence on its own. If two contacts share any email address, whether it is the main address or a secondary one, Rhythm scores them high enough to show as a likely pair. That same evidence can matter elsewhere in Rhythm, too. If an email address is already on a contact record, another process that tries to create a new contact or Portal credential with that address can treat it as belonging to the existing contact and refuse to create a separate one. Remove the email address from the existing contact first to clear the way.
- First and last names that sound alike also count as evidence. Rhythm uses a sound-alike comparison instead of exact spelling, so minor misspellings and phonetic variations can still match. A shared last name by itself is not enough; the first names must sound alike too. A matching first name with a different last name is never matched by name, so a surname change is not caught by name matching alone. Middle names are not used.
- Common English nicknames count as a first-name match. For example, "Bob" and "Robert," or "Ron" and "Ronald," can match by name even though they are not spelled or pronounced alike. Rhythm includes this nickname list, and your account cannot add to it. The list also appears to cover English-language names only.
- A shared or nearly shared postal code adds weight when the name or email address already suggests a match. Home, work, and alternate addresses count. A preferred address chosen from a custom type does not count, and custom address types themselves do not count. Postal codes add more weight to a name match than to an email match because a shared email address is already strong evidence.
Why a Pair Might Not Be Shown
Rhythm compares every scored pair against a minimum strength. Pairs below that minimum are discarded instead of shown to staff.
- Any shared email address always scores well above this minimum, so Rhythm does not filter out a pair that shares an address.
- A name match must have closely sound-alike names, or it must be supported by a matching or nearly matching postal code. Two names that only cross-match loosely, without postal code support, do not meet the minimum.
If a pair you expected to see is missing, the most likely reason is a weak name-only match with no postal code support.
The minimum strength and the weight for each kind of evidence are fixed the same way for every Rhythm account. There is no setting to raise or lower duplicate-detection sensitivity or set your own minimum-match percentage. Support cannot turn it up for your account because changing it would require a product change that applies everywhere, not an account-level configuration. Rhythm also compares active and inactive contacts when looking for likely duplicates, since it has no way to know whether your account wants inactive records left out. Seeing an inactive contact in a likely-duplicate result is expected and does not mean something is wrong.
Why an Imported Contact Might Not Be Checked
Contacts saved through Rhythm's data import are left out of duplicate matching entirely. This works the same way for every account, and there is no setting to turn matching on for imports. Staff also do not see a notice that an imported batch was not checked. If you import contacts and want them checked against your existing records, plan to review them separately because the import alone will not surface likely duplicates.
The risk of creating an unwanted duplicate is highest when an incoming row has no email address and does not match an existing contact by a stable identifier such as a Contact ID. Without one of those anchors, the import is more likely to create a new record instead of updating the person who is already there. Before you import, scan your spreadsheet for more than one row with a similar first and last name and resolve those likely duplicates yourself. That is exactly the case matching cannot catch for you once the file is uploaded.
After an imported contact is edited and saved the usual way, that save is checked normally, including against contacts imported with it.
How a Likely Pair Is Remembered
When Rhythm finds a likely pair, it stores that pair under both contacts, so you can see it from either contact record. A pair can have a status of likely, dismissed, or manually added by staff, and your account's likely duplicate list uses that stored status.
Each contact also keeps a running count of its likely duplicates and the strength of its best match. This lets you see how many possible duplicates a contact has without opening the full list. That count updates only when the contact itself is saved. If a duplicate is dismissed, deleted, or changed enough to stop matching, the other contact's count and best-match strength can lag until that contact is recalculated too. Because of that lag, a contact can still appear in Manage Duplicate Contacts as having a likely duplicate, but opening it shows no duplicate left to merge. This does not happen for every listed record, but it is expected when the paired contact has not been recalculated since the match stopped applying.
Rhythm calculates likely-duplicate data from your contact records rather than storing it as a separate source, so it can always be rebuilt from your contacts. If a historical export or recovery point of your account does not include this data by itself, that is not a gap. The data can be recreated from the contacts it was calculated from.
Duplicate Data Never Crosses Between Accounts
Rhythm only compares contacts within your own account's data. A contact in your account is never compared against, or revealed to, another organization's account. Nothing about how Rhythm flags a possible duplicate for you depends on or leaks into another Rhythm customer's records.
Related
For how a new organization can end up matched, or unmatched, by its web domain in a similar way, see Domain Matching and Company Profiles.