A preference centre is a page where a recipient chooses what they want from you: which streams, how often, sometimes which channel. It exists because the alternative to a choice is usually a departure, and a partial subscriber is worth more than an unsubscribed one.
It is also a place where good intentions produce non-compliant designs, so it is worth being precise about what it may and may not do.
What is it allowed to replace?
Nothing. It is allowed to sit alongside.
The FTC's compliance guide permits the menu and attaches a condition to it in the same sentence:
You may create a menu to allow a recipient to opt out of certain types of messages, but you must include the option to stop all marketing messages from you.
That is the rule the design has to satisfy. A page offering "monthly instead of weekly" and "product news only" and no way out is not a preference centre with a gap; it is a non-compliant unsubscribe path.
Under European rules the position is, if anything, firmer. Article 21(3) of Regulation (EU) 2016/679 gives the right to object to direct marketing no balancing test:
Where the data subject objects to processing for direct marketing purposes, the personal data shall no longer be processed for such purposes.
There is no version of that where a granular preference substitutes for the objection.
What must it not require?
A login, a fee, or a reason.
The US rule sets the outer limit on friction:
You can’t charge a fee, require the recipient to give you any personally identifying information beyond an email address, or make the recipient take any step other than sending a reply email or visiting a single page on an Internet website as a condition for honoring an opt-out request.
Three designs fail that. A preference centre behind account authentication requires more than visiting a page. One that asks "why are you leaving?" before it will save the change requires an extra step. One that only lets an existing customer manage preferences leaves everyone else with no path at all.
The safe pattern is a page reachable from a signed link in the message, which identifies the recipient without asking them to prove anything, and which saves the change on the click rather than after a survey.
What should it offer?
Fewer choices than teams want to build, and one of them has to be the door.
Three axes cover most of the value:
- Which streams. Product announcements, the newsletter, the offers. This is the axis that actually rescues subscribers, because most people leaving want less of one thing rather than none of everything.
- How often. A frequency choice converts a volume complaint into a setting.
- Stop everything. Prominent, not buried under the granular options.
Google's own subscription guidance points at the first of those as a practice rather than a compliance matter:
Let recipients review the individual mailing lists they’re subscribed to. Let them unsubscribe from lists individually, or all lists at once.
Note the phrasing of the second sentence, which is the same requirement the FTC states, arriving from a different direction.
What to leave out: a channel switcher that quietly moves someone from email to SMS. Consent is channel-specific in most regimes, and an email preference is not permission to text. What GDPR requires and the TCPA come at that from opposite ends and agree on it.
How does this fit with a suppression list?
They are separate records, and keeping them separate is what makes either one legible.
On Bird the distinction is explicit. A stated preference is the recipient telling you what they want, and it lives on the Preferences surface; a suppression is a deliverability fact such as a hard bounce, and it lives on the suppression list. Unsubscribe links describes what happens when someone opts out:
marketing sends to that address are refused, transactional sends still go through, and the record shows who opted out and when. An opt-out can also be recorded for all messages rather than marketing only, which is the recipient asking for silence outright, and that blocks transactional mail too. Categories covers why the category on the send is what makes any of this take effect.
Reversing an opt-out is deliberately not symmetric with recording one. It needs the moment the recipient consented, and the opt-in only takes effect if that moment is later than the opt-out it reverses. That is the shape a preference centre should write into: a statement with a time, not a flag being flipped.
Does it reduce unsubscribes?
It moves some of them, which is the honest claim.
A preference centre converts a share of full departures into partial ones, and that is worth building. It does not address the reason people wanted to leave, which is usually frequency or relevance rather than the absence of a settings page. What is unsubscribe rate covers reading the number, including why driving it to zero is not the goal.