2 Klaviyo prompts for Claude before you suppress anyone

Klaviyo's own definition of a profile worth suppressing has six zeros in it.

No opens, clicks, site activity, product views, checkouts started, orders placed, and every one of those measured over all time, not over a rolling window. Also has to be 180 days old and have received five emails in the last 72 weeks to qualify.

Almost nobody cleans a list that strictly. The common version is some form of not having opened anything in 90 days, and Klaviyo treats that as completely different.

That group gets excluded from campaign sends. It doesn't get suppressed.

The six-zero group is the one that goes through one last sunset flow and only then gets suppressed if it doesn't respond.

The distinction matters because suppression barely reverses.

A suppressed profile stays suppressed even if that person starts a checkout or places an order later. The only ways back are resubscribing or a manual unsuppression.

Klaviyo is direct about the cost of getting this backwards - suppress too early, and you lose revenue from people who would have come back on their own.

Not everyone who's stopped buying is the same problem, though.

A VIP who's gone quiet is different than someone who ordered once eight months ago, and the right response to each isn't the same size.

I am sharing 2 Claude prompts to find both groups.

1/ The VIP segment deep-dive

Your VIP segment already generates more revenue than its size would suggest.

And what usually gets skipped is the churn flag sitting inside that same segment, Klaviyo's own prediction of who in that group is about to stop buying.

Churn risk on a Klaviyo profile is a defined number - the predicted probability that a customer won't place another order in the next 90 days, banded low under 33%, medium from 33 to 66, high above that.

Klaviyo only starts calculating it once an account clears a bar - at least 500 customers with a completed order, 180 days of order history, and orders within the last 30 days, with some customers having placed three or more.

Below that bar, the field sits empty rather than showing zero. A newer account might not see this section populate at all. That's a data-maturity gap.

→ Full segment definition, exactly as it's built
→ Campaign performance for VIPs against the rest of the list
→ Orders and revenue this segment drives, and what share of total list revenue that is
→ How many are active in the last 30 and 7 days
→ How many are flagged for predicted churn

The response to that prompt: VIPs account for 24% of total revenue, coming from 1% of the list, which tracks with how concentrated the VIP segment usually is.

In that same account, 94 of the 1,420 VIPs were flagged for predicted churn.

That number comes from the same signals behind the churn-risk score: recent browsing, opens, and order timing that already looks like the pattern of customers who stopped buying before. 94 is short enough to work through in an afternoon, and a personal call outperforms a discount code on this list because the discount was not the reason these particular customers bought from you in the first place.

The rest of that section of the library covers unused segments and duplicate segment definitions, list hygiene rather than who to reach out to.

This is the one prompt in it that answers who deserves a call.

2/ The lapsed customer report

94 people is a list one person can call through.

Everyone else who's stopped buying is a much bigger group, and the easy mistake is treating all of them the same - one flow, one discount code, sent to the whole pile regardless of how long it's been or what they used to spend.

→ Lapsed customers broken into recency tiers, 30 to 60 days out to a year or more
→ Average prior spend for each tier
→ Which tier is the best target, and why
→ A revenue number attached to recovering part of it

This pattern tends to hold across most accounts running this prompt.

The 1-year-plus group is the biggest by far, 18,200 people, but also the cheapest at $140 average prior spend, and it's usually where a win-back budget defaults to simply because it's the largest number on the screen.

The report pointed somewhere else instead: 90 to 180 days lapsed, 4,800 people at $240 average prior spend, with $230,000 in revenue at stake if a fifth of them come back.

It's old enough that a normal flow has already stopped reaching those customers, but recent enough that they still remember buying from you.

The year-plus group has usually drifted too far for a win-back to change anything, no matter how large that segment looks sitting at the top of the report.

A year later, the question stops being who deserves outreach at all.

And becomes who gets one last sunset flow before suppression, a deliverability decision, and it runs on a different prompt entirely.

That one, covering sunset candidates, suppression list health, and where your SPF, DKIM, and DMARC stand - I shared a few weeks back in the Claude operating calendar, if you want the rest of the picture. Check out here →

All 27 Klaviyo prompts covering reporting, flows, segments, profiles, campaigns, saved Cowork workflows - you can paste and run. Download for free →

- Ankit