Killed 5 of their 6 flow splits

Happy Wednesday!

Last week I was in a client's Klaviyo account and I killed 5 of their 6 flow splits. The one I kept was quietly making about $50k a year on a single email. No follow-up, no depth, nothing behind it.

The audit

Here's what they had built. Six collection-specific splits on one flow, with furniture going down one path and artwork going down another. Each collection got its own email, and each split had exactly one email in it.

That's horizontal expansion. More splits, more conditionals, more segmentation. When people say "personalization," this is usually what they mean.

Horizontal vs. vertical, defined

What they didn't have is vertical. Vertical is depth. It's the second, third, and fourth email inside any given split.

The winning furniture split was doing $50k/year off one email. I want you to sit with that for a second.

One email. $50k.

That's not a success story. That's a warning sign.

If one email can do $50k, what does a second email do? A third? A fourth?

Nobody had bothered to find out. They'd built sideways instead of down.

Why this happens

The reason I see this constantly is the same reason most retention decisions get made. Someone heard "personalization is great." So they went and built collection-specific splits without ever looking at the data underneath.

There was no baseline. No catch-all to beat. No way to know if the splits were actually outperforming a simpler flow, because a simpler flow never existed.

The 5-step framework

So here's the order I build in now. Five steps, and the order is the whole point.

That's the framework. Vertical before horizontal. Depth before width.

The hidden cost nobody calculates

There's one cost most builders never calculate, and it's the one that quietly kills accounts. Every split is a flow someone has to maintain.

Someone has to QA it, update it when the product page changes, rewrite the copy when the brand refreshes, and fix it when a merge tag breaks. Six splits with one email each isn't six emails of work. It's six flows of ongoing ops.

If I'd kept those six splits in place, the team would spend hours every quarter keeping them alive for a return that barely moves the needle. Killing five of them freed up the only thing that actually compounds, which is attention on the flow that's already working.

The rule underneath

The mistake isn't personalization. Personalization is great when the data earns it. The mistake is building for personalization before you've built the thing personalization is supposed to improve.

Build out of data. Not out of "I heard personalization is good."

If your account has a bunch of splits with one email each, or if you've never actually seen a baseline to beat, that's exactly what I look at in an audit.

Or just reply to this email. I read every one.

Have a great week.

- Raymond

I'll be posting the short version of this on LinkedIn later this week. If you want the framework in feed form, follow along over there.