Bulk Suppressing a Do-Not-Contact List in Klaviyo with AI

A client sent us a do-not-solicit list from another part of their business: roughly 1,600 contacts who needed to be removed from email marketing in Klaviyo. Not "moved to a different list." Removed. Verified. Done.

This is the kind of request that normally eats an hour or two of someone's day. Instead, we handed it to an AI agent with Klaviyo API access and it was finished in minutes, with a full accounting of every single contact. Here's how that worked, and why the interesting part isn't the speed.

Why This Is Normally Painful

On paper, suppressing a list of emails in Klaviyo sounds simple. Klaviyo even has a CSV upload for suppressions. But anyone who's actually done this for a client knows the upload is the easy 10% of the job. The real work is everything around it:

Cleaning the file you were given, because it never arrives in a tidy one-column format
Figuring out which of those emails actually exist in the Klaviyo account
Checking who's already suppressed or unsubscribed, so you know what actually changed
Verifying afterward that everyone on the list is truly unmailable

And the stakes are higher than a normal list-cleaning task. A do-not-solicit list is a compliance request. If one contact slips through and gets a promo email next week, that's not a metrics problem, that's a trust problem with the client. So you double-check. Which is why a "simple" suppression job turns into an hour or two of spreadsheet work and spot-checking profiles one at a time.

The Setup: An AI Agent With Klaviyo API Access

We run Claude Code hooked into our operations: Klaviyo accounts, email, Drive, our project management. Internally we call it Jarvis, and it more or less acts as a client brain. The piece that matters for this story is the Klaviyo API connection. The agent can look up profiles, read their suppression status, and suppress them, using the same API endpoints we already use for automated list hygiene.

So instead of opening the CSV and starting the usual slog, the instruction was basically: here's a do-not-solicit list from the client, these people need to be removed from marketing in Klaviyo, make sure they're handled properly.

What It Actually Did

The agent parsed the file, then looked up every email address against the Klaviyo account and checked each profile's current status. The breakdown on ~1,600 contacts:

~1,200 matched a profile in Klaviyo
~400 weren't in Klaviyo at all, so there was nothing to do
~1,000 of the matches were already correctly suppressed or otherwise unmailable
~200 were still emailable, and those got suppressed

That's the whole job. Every contact on the client's list accounted for, the 200 that actually needed action handled, and a clear summary we could send back to the client.

The Audit Is the Point, Not the Speed

The fast version of this task has always existed: dump the CSV into Klaviyo's suppression upload and walk away. The problem is you learn nothing. You don't know how many of those contacts were in the account, how many were already handled, or whether the upload actually changed anything.

Checking state first and only acting on the profiles that need it gets you three things a blind upload doesn't:

A verifiable record. When the client asks "are these people definitely removed?" the answer is a per-contact status report, not "we uploaded the file."
A sanity check on the data. If the numbers had come back weird, say only 50 matches out of 1,600, that's a signal something is off with the file before anything gets changed.
A minimal change. Only the 200 profiles that needed suppression were touched. Everything else stayed exactly as it was.

In the manual world, that level of rigor is exactly what makes the task take two hours, so most people skip it. When the lookup-and-verify step costs nothing, you get the rigorous version by default.

Why Suppression Is the Right Tool for Do-Not-Contact Requests

Quick note on mechanics, because this trips people up: for a do-not-solicit request you want account-level suppression, not list removal and not profile deletion.

Removing someone from a list doesn't stop them from getting emails through other lists, segments, or flows.
Deleting the profile erases the record entirely, so if that person re-enters your ecosystem through a popup or a checkout, they start getting marketing email again with no trace of the original request.
Suppression keeps the profile but blocks all marketing sends to it, permanently, unless someone deliberately unsuppresses. The do-not-contact status survives.

This matters for deliverability too. Suppressed profiles don't drag down your engagement rates the way ignored-but-mailable ones do. If your account is carrying a lot of dead weight, that's a separate problem worth fixing; we covered our approach in the automated sunset flow post, and our deliverability guide goes into why the engaged-to-unengaged ratio drives inbox placement.

The Bigger Shift

The do-not-solicit list is one example, but the pattern generalizes. Klaviyo work is full of one-off operational tasks that are too small to build a tool for and too tedious to enjoy doing by hand: reconciling a list against the account, checking a batch of profiles for a property, cross-referencing an export from another system. Each one is 30 minutes to 2 hours of careful clicking.

With an AI agent wired into the Klaviyo API, that whole category of work collapses into: describe the task, review the plan, check the summary. The agent does the lookups, the matching, and the verification that a careful human would do, without the part where a careful human loses an afternoon to it.

If you've got a Klaviyo account with operational cleanup that keeps getting pushed off, reach out. This is the kind of thing we handle for clients as part of running their email program.