SHA-256 hashing
How it works
SHA-256 turns any input into a value of 64 hexadecimal characters. The same input always gives the same hash, but the hash can't be turned back into the input. The platform hashes its own users' data the same way and compares the hashes.
That only works if both sides hash exactly the same string. So the data is normalized first:
- Email: trim spaces at the start and end, make everything lowercase
- Phone: international format with country code (E.164, such as
+14155550123); some platforms want it without the plus sign - Names: lowercase, spaces and punctuation as the platform specifies
Why it matters
Conversion APIs work best with contact details, and the platforms require them hashed. A badly normalized value still produces a valid-looking hash, it just doesn't match anyone. Errors like this go unnoticed: the match rate simply drops, and with it the quality of campaign optimization.
TSO hashes contact details with SHA-256 before they go to an ad platform.
Example
The same email address produces completely different hashes depending on how it's written:
anna.berger@example.comgives70eeff44fc61…5e5aaa16Anna.Berger@example.com(with a trailing space) gives2b388bf43230…01a5f49d
Only the first one matches the hash the platform knows for Anna. A system that passes on the address exactly as it was typed into the form loses every lead where someone used capital letters or added a space.
Common mistakes
- Mistaking hashing for anonymization: a hash is a pseudonym, not anonymous data. Anyone who knows the email address can compute the hash and link it. The GDPR still applies.
- Not normalizing before hashing: capital letters, spaces or a phone number without country code produce a different hash.
- Hashing twice: a value that's already hashed and gets hashed again matches nothing.
- Hashing everything: platforms expect some fields in plain text, such as click IDs or the IP address.