Tuft & Paw More Than Doubled Their Save Rate with Skio's Cancel Flow A/B Testing

The Cat Brand That Sweats the Details

Tuft + Paw has been designing cat gear with behaviorists since 2016, on the theory that a litter box can work for the cat without wrecking your living room. The subscription side of the business runs on litter and food, so retention carries real revenue.

Hadley Warren joined in May 2026 as their first retention manager. Agencies used to run email and SMS, the director of ecom covered the rest, and eventually there was enough retention work to hire for it. She was in the cancel flow within weeks.

Reading "Other" by Hand

Two cancel flows, one per product line, each with a stack of treatments sitting behind every reason a subscriber can pick. Improving them meant reading free-text "other" answers one at a time, hunting for a complaint nobody had built a treatment for, building one, and hoping.

"Slightly less efficient, as you can imagine."

Both flows opened with a splash screen asking people to pause or skip before they'd said a word about why they were there. Hadley thought it was in the way. Someone who ordered too much litter isn't looking for a question, they're looking for the skip button. But you don't yank a screen out of a live cancel flow on a hunch.

Cut the Splash Screen

Skio offered early access to Cancel Flow A/B Testing in late June. Hadley had a test live by July 9.

The feature copies your live flow, lets you rewrite it however you want, splits traffic between the two versions, and puts both save rates side by side.

"I was really into trying it. It's the best way to get some clean, good data."

She started at the front door and deleted the screen. Then she left it alone for a month.

The 127% Lift

The food flow's save rate more than doubled.

"I was surprised by that initial lift. We were like, wow, this is a big jump."

Better still, it landed on a smaller SKU the team is trying to grow. Since the test came back on, 232 subscribers have gone through the flow.

Cutting the screen "took people immediately to treatments for their problems," and for a litter and food subscriber, that problem is usually timing. Ordered too much, traveling next month, cat's still working through the last bag. A skip solves all three.

Hadley can now find out what works instead of guessing at it.

Past the Save Rate

Next test: what to offer someone whose cat won't touch the product, where the answer might be a swap to a different litter or one of three new foods.

The rest of the picture takes longer. Two new food SKUs launched into that same flow this summer, so Hadley is tracking August subscription growth alongside churn and revenue per subscriber, with lifetime value further out. The open question is what happens around the edges of the save rate. A flow that answers the question a subscriber actually walked in with should mean fewer of them writing in to ask how to pause or cancel, and less time spent untangling a subscription that could have fixed itself. Save rate was just the first thing testing was able to answer.

Hadley was nervous about touching the cancel flow at all before that first test. She now has a queue of them lined up and no interest in guessing her way through any of it.