Martech, explained · Data foundations · 2:47
Identity resolution: who owns the match
Match and merge, the two ways to be wrong, and who owns the rules that join your customers
The short answer
Identity resolution is two jobs: the match decides which records belong to the same person, and the merge decides which address and email survive. Nobody has the answer key, so every threshold trades over-merging for under-merging. The real choice is who owns the logic that decides two records are one person, and whether anyone can read it.
Questions this answers
What is the difference between deterministic and probabilistic matching?
Deterministic matching needs a shared identifier, like a hashed email. Probabilistic matching scores how alike two records look and calls it a match above a threshold; most real systems use rules first and scores on what's left.
Is a high match rate good?
Not on its own. A high match rate can mean the system is finding real connections, or that it is over-merging different people into one profile.
Why do teams need different matching rules?
Suppression would rather over-merge, so nobody gets the same email twice; a service desk would rather under-merge, so nobody sees a stranger's orders. One set of rules means somebody decides which team is right.
Who should own identity resolution?
You can rent the match, use the one inside your CDP, have a matching platform run it, or run your own rules on your own data. The more of the rule you own, the more you can read, and the more you have to keep.
Transcript
Every chapter as text. Click a time to play the film from there.
0:00 Same person?
Sarah Jones lives at twelve Oak Street.
Sara Jones, no h, lives at twelve Oak St.
Somebody has to decide whether that's one customer or two.
0:13 Two jobs: match and merge
That decision is identity resolution, and it's really two jobs.
The first is the match: which records belong to the same person.
The second is the merge: once they're joined, which address and which email survive.
Most of the marketing is about the match.
Most of the pain in production is in the merge.
0:36 Rules and scores
Deterministic matching needs a shared identifier, like a hashed email.
Probabilistic matching scores how alike two records look, and calls it a match above a threshold.
Most real systems do both: rules take the easy matches first, and scoring works on what's left.
0:55 Two ways to be wrong
Look, here's the awkward part: nobody has the answer key.
Lower the threshold, and two people become one.
Raise it, and one person stays two.
Every threshold you set trades one of those mistakes for the other, and no setting avoids both.
So a high match rate can mean the system is finding real connections, or that it's over-merging with enthusiasm.
1:21 Different teams, different rules
And different teams want different mistakes.
Suppression would rather over-merge, so nobody gets the same email twice.
A service desk would rather under-merge, so nobody sees a stranger's orders.
Give them one set of rules, and somebody has to decide which team is right.
1:41 Who owns the logic
The real choice is who owns the logic that decides two records are one person.
You can rent the match from an identity vendor, use the one inside your CDP, have a matching platform run it, or point your own rules at your own data.
The further right you go, the more of the rule you can read, and the more of it you have to keep.
2:01 Why agents raise the stakes
This matters more once agents act on the profiles.
If nobody can explain why two records were joined, nobody can explain why the budget moved.
2:13 Three questions to ask
Before you pick a tool, ask three things.
How many sets of rules do your teams need, and what does the second one cost?
Does the match have to happen while the customer is on the page, or is last night's answer fine?
And are you buying the match, the merge, or both?
2:32 So who owns the match?
Sarah and Sara will be joined, or kept apart, by somebody's rules.
Find out who owns the match.
Learn more at Martech Stack Builder, or build your own.
Read more
The articles this film is drawn from.