Forty-two split tests. Eleven accounts. Golf gear, electrolytes, oral care, skincare, streetwear, and a handful of others that share no customers and no product between them. Every one of those accounts changed the same variable in a welcome or abandon flow, and every one moved the same direction. Across the six tests I can actually show you, Revenue per Recipient (RPR) climbed from +138% to +731%.

The variable wasn't which offer to lead with. It wasn't the product shot. It was who you're writing to. Copy that speaks to who the subscriber already is beats copy that proves what the product does, and it does it across categories with nothing in common. When a result turns up that many times in accounts that share zero data, I stop filing it under coincidence.

For a while I thought the lever in a non-converting welcome flow was the offer. Too soft, too buried, not enough urgency. Then I actually watched what the feature-first subject lines were doing. They open well. Sometimes better than the identity versions. They pull clicks too. They just don't turn into orders.

A golf training brand gave me the cleanest version of this. The subject line built around tour-pro authority earned more opens than the founder-voice one, and more clicks. Then I pulled click-to-order: 11.11% on the feature subject, 36.84% on the identity one - same email body sitting underneath both. Opens flatter you. Orders keep score, and orders are the number bots can't fake, so that's the one I trust. The feature subject opened a loop the body was never built to close. Feature clarity buys you curiosity. Identity clarity buys you the decision.

Here's what that tells me about the subscriber who's been in your welcome flow and hasn't bought. They're not sitting there unconvinced that your product works. They've scrolled the proof already. They haven't bought because they haven't decided they're the kind of person who owns this kind of thing, and a feature list never answers that. Identity framing does.

Control ran a brand-origin story in the subject with a "tour-grade tools, built for results" preview. The variant ran "a message from our founder" with a preview about why the brand exists and how it helps you score lower. One is a company talking about itself. The other is a founder talking to your swing.

(Full disclosure, I do not golf. I once put a range ball through a clubhouse window and have kept a respectful distance ever since. ⛳)

Control led with "Ends Tonight: 10% Off" and restated it in the preview. The variant went with "Last chance to fix your swing" plus the offer and social validation stacked in the preview. By email nine the subscriber has seen eight of your emails and still hasn't bought. You can shout "10% off" in bigger caps, or you can name the original problem that got them onto your list.

Control leaned on the review count - thousands of five-star reviews for the energy - with a "bundle that actually works" preview. The variant opened with the felt problem in the inbox and pointed straight at the fix. By email six the subscriber has sat through five rounds of "look how many people love us." They don't need a sixth. They need to feel like it was built for someone dealing with exactly what they're dealing with.

Sits on a small handful of orders, so I hold this one loosely - but the direction matches everything else here.

Control was "a full clean in 20 seconds" over a preview about brushing smarter and faster. The variant ran "the smarter way to brush your teeth 🦷" with a preview joking about how far we've come since the 1500s. Opens came out roughly even, so this was won on the body's terms, not the open. The control hands you a complete, closed picture before you click, nothing left to find out. The variant opens a claim in the subject and answers it sideways with a bit of humour in the preview - two reasons to engage instead of one.

Thin order counts again, same caveat - but the mechanism is visible.

Control was "Together, Let's Build Your Confidence!" over "You deserve this." The variant ran "Julie, confidence starts here 🩵" over "Real women are feeling like themselves again. Without hormones." The control is aspirational in a way that could belong to any brand. The variant removes the single biggest purchase objection in nine characters - "without hormones" - before the email even opens. That isn't identity framing in the abstract; it's knowing which barrier the subscriber is carrying and taking it off the table upfront.

This one has real volume behind it - one I'd act on today.

Control asked "Did you forget something in your cart?" over "We saved your picks for you." The variant named the brand and the item as something the subscriber already owns in their head, then added forward motion and a cost to waiting: "that piece is still in your cart 👀" over "You're closer than you think. Grab it before it's gone." "Did you forget" casts the subscriber as someone who slipped up. The variant casts them as someone who's almost there.

The model I keep landing on: a welcome subscriber is asking themselves "am I the kind of person who uses this?" A feature tells them the product works. Social proof tells them other buyers are happy. Both are true and neither one answers whether they see themselves owning it. Identity framing answers the actual question. It either reflects the specific problem back at them so it reads as "built for someone like you," or it hands them the identity of the person who already owns the thing.

One honest note: a few of these wins sit on small order counts and short windows - the supplement and oral-care tests especially. The golf pair and the intimacy test carry real volume, and those are the ones I'd rebuild a flow around without blinking. The rest are worth putting into your test queue, not worth betting your whole welcome sequence on tonight. What makes the whole thing worth writing up rather than shelving is that forty-two tests across eleven unconnected accounts all leaned the same way.

Pull up your welcome flow subject lines. You're not hunting for urgency or offer clarity this time. You're hunting for identity.

If the honest answer is "we've mostly got features and a pile of social proof," you just found your test queue.

The economics here aren't abstract. Say your welcome flow reaches 10,000 new subscribers a month at a 0.5% purchase rate. A +200% RPR lift - right in line with the results here that carry real volume - doesn't ask you to go buy more traffic. It asks your existing subscribers to spend more on the same traffic, every month, and that keeps compounding in the background.

Forty-two tests across eleven accounts isn't a lucky accident of this one write-up - it's a slice of what's always running.

Right now there are 140 to 180 flows under management, over 1,800 messages tested on rotation, 12 to 15 flows in the testing queue for every client. We launch more than 100 tests a week and conclude about as many, and every result lands in a database built to find what repeats across brands that share nothing.

That's how a golf finding ends up sharpening a supplement account - because each test stands on every test before it. This isn't guesswork with better lighting. It's aimed, and it sharpens every single week.

So the ask this week is small. Pick one welcome email that opens well and doesn't sell. Rewrite the subject line to name the person instead of the product, leave the body alone, and test it. Don't touch the flow. If it moves, you've got your next twelve tests, and if it doesn't, you've lost a week and learned your list is odd, which is also worth knowing.

If you'd rather not spend the next six months finding your own forty-two, that's the job. I run the strategy on every account in the book and I'm in the accounts doing the testing, so the golf finding sharpens the supplement account and back again.

Your data stays yours. The patterns travel. I'll also tell you flat when a win is too thin to bank on, which is the bit most people skip. Come find me. ✌