In 1975, a student sat down at a table in a psychology lab at the University of North Carolina. On the table: a can of butter mints, a packet of slim cigars, and a glass jar holding ten chocolate chip cookies.

She'd been told this was a consumer preference study, that the university was running several of them, and that she'd be tasting things and rating them. All of that was a cover story. The three running it, Stephen Worchel, Jerry Lee and Akanbi Adewole, were social psychologists, and the only thing on that table that mattered was the jar.

A few minutes in, a second researcher knocked. He apologised for interrupting and said he'd run short, his own participants had eaten more cookies than he'd expected. The two of them swapped jars. Her jar of ten went out of the room, a jar holding two came back in, and she was asked to taste a cookie and say what she made of it.

Nothing about the cookie had changed. The women who watched their cookies get scarce rated them more desirable and more attractive than the ones whose jar stayed full, and when asked what the thing ought to sell for, they about doubled it: 71 cents a pound against 37. One measure didn't move. The study also asked them how the cookie actually tasted, and on taste there was no effect at all, in any condition. Same cookie, worth twice as much, tasting exactly the same.

That's the whole trick sitting in a 1975 table of means. Scarcity changes how badly someone wants the thing. It doesn't touch the thing. Fifty years on, that jar is most of the reason almost every abandoned-cart email ever written carries a countdown in it. Mine carried one too. Last month I finally put it to a proper test, and it lost by 30%.

This lactation-support brand's first browse-abandon email was doing fine. I wanted it better, so I reached for the most reliable lever in the whole abandon playbook: scarcity.

I A/B tested the original's calm line, "We saved your cart for you, Mama / Your 10% is still good, no pressure, just a nudge," against a variant with a pulse:

"These keep selling out, Mama / Your 10% off is still here. Don't let it slip."

Same offer, everything below the fold untouched. One line, more urgency.

The brand, for context, makes lactation snacks and supplements for new mothers worried their milk supply is slipping.

The original won. We take the L and we move on. The part that stung:

That third line is the sting: more taps for my email, fewer buyers. The clicks were curiosity, not trust.

One thing I only noticed writing this up. That calm line isn't a lucky one-off. A month before I came along, it had already seen off a different challenger at the same position, an emotional line with no offer in it, by 151.8% RPR.

The subject I was trying to beat already had a record for walloping the competition. But, what do you do? You test it anyway. 🤷

Think about who opens this brand's emails. A brand-new mother, often up half the night, often scared her supply is dropping. She isn't hunting a deal. She came for help and a quiet signal she's doing okay. "These keep selling out, don't let it slip" puts a countdown in front of someone who's already exhausted, and reads as a stranger shouting a deadline she never asked for. The calm line holds the item for her, takes the pressure off, tells her there's no rush. For this reader, reassurance is what closes.

There's real work on what that exhaustion costs her.

Sit with that against my reader. She may have actually lost a night's sleep. Whatever bandwidth she has left is already spoken for. And into that I sent a countdown, which is a second thing to worry about wearing the costume of a helpful reminder.

My scarcity line was the sales clerk who follows you to the register going "better grab it, these are flying out!"

He was hollering "last one!" at a woman holding a newborn. To say she didn’t GAF about it being the last one or flying off the shelf was an understatement.

Scarcity works. I just aimed it at the wrong person. On a deal-hunting list hovering over a full cart, "almost gone" is exactly the shove that pays. The industry decided years ago that every abandon email needs a countdown and a "selling out" line. Most of us never checked whether that holds for the actual reader.

And it's a genuine lever, which is the annoying part. Go back to that jar for a second, because I left a detail out of it. The cookies didn't just have to vanish. Worchel's team ran one version where the jar emptied by accident, the second researcher had simply grabbed the wrong jar, and another where it emptied because other people had eaten them. Two cookies left either way. The ones other people had wanted were rated higher. Scarcity wasn't doing the work on its own. A story about other people was.

What the template doesn't show you is the bill. In 2021 three researchers published a study in Psychology & Marketing asking people about things they'd missed out on during Amazon Prime Day. Of the people who missed out because the item sold out, 68% went and bought from a competitor. Of the people who missed out for some other reason, price being the common one, only 32% did. FOMO was the thing in the middle doing the work. So scarcity, aimed badly, doesn't just fail to close someone. It can hand them to somebody else.

Half an hour, one email. The version you can run yourself today:

I lose tests, often, because that's the only honest way to find the ones that pay. I win about two in three. Which is another way of saying a third of the things I am certain about turn out to be wrong, like this one, and the test is the only reason I ever find out which third. The wins pay multiples and the losses cost a fortnight. Anyone who offers to walk in and "overhaul your flows in 30 days" without testing is handing you a coin flip, and theirs isn't even weighted. The careful way is slower and loses sometimes. It also won't overhaul you into the poor house.

One lost test is one reading off a very big instrument. I manage 140 to 180 flows, test 1,800-plus messages on rotation and launch 100-plus tests a week. Every result lands in one database built to spot what repeats across accounts that share nothing, which is also how I know my own hit rate instead of guessing at it. Your numbers stay in your account. Only the shape of what works travels.

So the real choice. You can take the half hour above, fix that one email in your own flow, and keep the win. Good option, and I mean it.

Or you spend about five minutes on the audit application and I read the whole account, write up every place it's leaking revenue, and send it back inside three business days. It costs you nothing. If you take that report and fix all of it yourself, that's a fine ending and I'll be glad I pointed the way.

The plain email won by firing the pushy guy at the register and hiring the one who says "I set it aside for you, no rush." I was so sure about that scarcity line I'd have rolled it out to every client by Friday. The test is the only reason I didn't.