The three Klaviyo flows that actually earn their place

Every store I audit has a long list of Klaviyo flows, most of them switched on years ago and never looked at since. Then you open the revenue-per-recipient column and find that three flows produce almost everything.

Here they are, and more usefully, the setup mistakes that stop them working.

1. Abandoned checkout

Always the highest revenue per recipient. Someone entered payment details and stopped — they are as close to buying as a person gets.

The mistakes I find most often:

Sending too late. A four-hour delay on the first email is common and it is too long. The cart is abandoned because something interrupted them, and the window where the intent survives is short. 30 to 60 minutes performs better in every test I have run.

Discounting immediately. If the first email carries 10% off, you are training people to abandon carts. Email one is a reminder. Discount, if at all, in email three.

No exclusion for completed orders. Obvious, and I still find it broken maybe a third of the time, usually because a custom checkout does not fire the standard event. Nothing damages trust like being chased for something you already paid for.

2. Browse abandonment

Lower revenue per recipient than checkout abandonment, far higher volume, so often more total revenue. And chronically misconfigured.

The rule that makes it work: trigger on intent, not on any pageview. Someone who viewed one product for four seconds is not a lead. My usual filter is three or more views of the same product, or two products in one collection, within a session.

Suppress anyone who has viewed a product they already own. Recommending someone the thing sitting in their hallway is the fastest way to get unsubscribed.

One store was sending browse-abandonment email to customers who had bought that exact item nine days earlier. It had run for seven months. Their unsubscribe rate halved when we fixed it.

3. Post-purchase

The one most stores treat as a receipt and nothing more. It has the highest open rate of anything you will ever send — people genuinely want to know where their order is — and most of that attention is wasted.

What works, in sequence: order confirmation, then shipping notification, then a genuinely useful message timed to arrival. For a wall panel supplier that was an installation guide. Support tickets dropped noticeably. Then, and only then, a review request.

Review requests sent before delivery are the single most common flow bug I see. The timer starts at purchase, the product arrives in five days, and the request lands on day three. You are asking someone to review a box they have not opened.

Trigger on the fulfilment or delivery event, not on the order event, then add the delay.

What I usually turn off

Win-back flows targeting 12-month-lapsed customers. For considered purchases people buy once. Emailing them monthly produces spam complaints, and complaints damage deliverability for the flows that do work.

Birthday emails where the store collects birthdays from perhaps 4% of the list.

Anything sending to unengaged segments. This is the one worth being strict about. Deliverability is reputation-based: if a chunk of your list never opens anything, mailing them teaches inbox providers that your mail is unwanted, and that lands your abandoned-checkout emails in spam. Sunset anyone who has not opened in 90 days. Sending less genuinely makes the rest perform better.

Measure against a holdout

Attributed revenue overstates everything. Klaviyo credits a purchase to an email if it happened within the attribution window, whether or not the email caused it. Some of those people were going to buy anyway.

Hold back 5% of recipients as a control and compare conversion. Every flow I have measured this way came in below its attributed number. Abandoned checkout held up best — real lift, comfortably worth running. A couple of others turned out to be taking credit for purchases that were happening regardless.