Hashed suppression: honour an objection without keeping the data
When someone asks you to delete their data and stop marketing to them, you need a way to recognise them if their details come back on a list you buy later. Hashed suppression lets you do that without holding their name, email or address in readable form.
In short
- To keep honouring an objection to direct marketing, a company has to be able to spot the person again. The usual answer is a suppression list, but that means keeping the very details the person asked you to delete.
- Hashed suppression keeps a one-way fingerprint (a keyed hash) of each identifier instead. New records are hashed the same way and blocked if they match.
- This is the method guardID recommends to every company we send requests to. It's privacy by design: you hold far less data, and the person stays off your lists.
The problem with deleting everything
Data brokers and marketing companies buy, merge and re-import lists all the time. If you delete someone's record completely, nothing stops the same person arriving again in next month's list, and you're back to processing data they've already objected to.
The ICO's answer is a suppression list: rather than deleting every trace, you keep just enough information to make sure the person's objection is respected in future. Under UK GDPR Article 21(3), once someone objects to direct marketing you must stop processing their data for that purpose, so a working suppression list is how you prove you're doing that.
The catch is that a conventional suppression list is still a list of names, email addresses, phone numbers and home addresses. It's exactly the data the person wanted gone, sitting in one more system that can leak.
How hashed suppression works
- Erase the record. Act on the erasure request (Article 17) and the objection (Article 21) as normal, and delete the person's data from your marketing and resale systems.
- Keep a fingerprint, not the details. Before you delete, normalise each identifier you'd match on (email, phone number, name and postcode) and store only a keyed hash of each one, such as HMAC-SHA-256. The readable values are deleted.
- Check every new record at the door. When you acquire or import data, hash its identifiers the same way. If a hash matches your suppression list, drop the record before it enters your database.
The hash can't be turned back into the original details, but the same input always produces the same output, so you can still recognise the person. You stop them being re-added without knowing who they are.
Implementing it well
- Use a keyed hash, not a plain one. Email addresses and phone numbers are easy to guess, so a plain SHA-256 of an email can be reversed by hashing a list of known addresses. A keyed hash (HMAC) with a secret key stored separately from the list prevents that.
- Normalise before hashing. Trim spaces, lower-case email addresses, convert phone numbers to one format (such as +44), and upper-case postcodes without spaces. Otherwise "Jo@Example.com" and "jo@example.com" won't match.
- Hash each identifier separately. Store one hash per email, phone number and name-plus-postcode combination, so a record that shares any one of them is caught.
- Keep only what you need. A hash plus the date of the request and, if you use one, an expiry date is enough. Don't keep the reason or any other detail.
- Remember it's still personal data. A keyed hash is pseudonymised, not anonymous, so keep it secure and mention suppression in your privacy notice. It's simply far less data, and far less useful to anyone who steals it.
In outline, the check looks like this:
key = secret kept outside the suppression database normalise(email) = trim + lower-case hash = HMAC-SHA-256(key, normalise(email)) on request: store hash, delete the record on import: if HMAC-SHA-256(key, normalise(new email)) is in the list, skip the record
Responding to a guardID request
guardID sends erasure requests and objections to direct marketing on behalf of its members, under a signed Letter of Authority. When you've acted on one, it helps us to know which of these applies:
- Deleted: you've erased the data and don't keep a suppression record.
- Suppressed: you've erased the data and keep a suppression record (hashed, we hope) so the person isn't re-added.
- Suppressed until a date: your suppression has an expiry. Tell us the date and we'll renew the request before it runs out.
If you'd like to talk about putting hashed suppression in place, or about a more direct way to receive guardID requests, email hello@guardid.app. An example of our Letter of Authority is on our legal page.
Frequently asked questions
Your readable details are deleted. With hashed suppression, the company keeps only a one-way fingerprint of them, which it can use to recognise and block your details if they turn up again, but not to contact you or sell your data.
Yes. Under UK GDPR, hashed identifiers are pseudonymised data, which is still personal data. That's why we recommend a keyed hash, stored securely, and kept for suppression only. It's a big reduction in what a company holds, not a way around the law.
A normal suppression list works, and the ICO accepts it. But it's a readable copy of the details people asked you to delete. If it leaks, it's a breach. A hashed list does the same job with data that's useless to anyone without the key.
No. The law requires them to honour your objection and to protect the data they keep, not to use a particular method. Hashed suppression is the approach guardID recommends because it meets both duties while keeping as little data as possible.
Sources
- UK GDPR Article 21: right to object (legislation.gov.uk)
- UK GDPR Article 25: data protection by design and by default (legislation.gov.uk)
- Direct marketing guidance, including suppression lists (Information Commissioner's Office)
- Anonymisation and pseudonymisation guidance (Information Commissioner's Office)