The Ecomm Analyst

Most ecommerce brands I look at have somewhere between 8 and 15 flows running in Klaviyo. Three of them produce 80% of the revenue. The rest are ambient noise that someone built once and nobody has audited since.

Here are the three that consistently earn their place, with the caveats that have caught me out.

The welcome series is the obvious one. The standard advice is three to five emails over a week. The version that actually performs in my experience is two emails sent close together, then a hard stop. The first email delivers the promised incentive (usually a code) within minutes of signup. The second, sent 48 hours later, recovers anyone who didn’t use the code. After that, you’re just badgering someone who isn’t interested.

Abandoned cart comes second. Timing matters more than copy here. The first send needs to land within an hour of abandonment. After 24 hours your hit rate falls off a cliff. I run this as a two-touch flow at 45 minutes and 22 hours. Anything beyond that I’d rather catch in browse abandonment.

Post-purchase is the one most brands underuse. Not the “thanks for ordering” email, which is fine but not where the money is. The flow that runs 14 to 30 days after delivery, depending on the product’s natural reorder cycle, asking for a review and gently suggesting a complementary product. For consumables this is straightforward. For durables you have to be more creative, often pivoting to UGC or referral asks.

I look at flow performance in ThoughtMetric, not in Klaviyo. Klaviyo’s reporting is built to make Klaviyo look good. Every email that gets opened within a 5-day window before a purchase gets credit, which means your welcome series will appear to drive purchases that would have happened anyway. Looking at the same flows in a blended attribution view usually brings the numbers down by 30 to 50%.

That sounds bad. It isn’t, really. The flows are still working. They’re just not the only thing working. Once you accept that, you can stop trying to optimize for the wrong number and start asking the right questions, which are about lift, not credit.

The flows I see most often that I would turn off tomorrow are birthday emails (negligible incremental revenue, painful to maintain), VIP segment flows that are functionally identical to the regular flows, and any flow that requires more than one branching condition. If you can’t explain a flow’s logic in one sentence, you don’t actually know what it’s doing.

Three flows, well-tended, will outperform fifteen flows on autopilot. Every time.