The Unlisted Sale: How to Test a New Offer Every Month Without Losing Money

It's the Tuesday after a long weekend sale and your phone is going off. Not from the campaign you announced to the whole list. From the one nobody was supposed to notice. A little line in the announcement bar, a separate landing page, a 5% offer dressed up around a date most people forgot was even a thing. And it's quietly outconverting the big banner you spent two weeks building.

That gap, the one between the loud sale and the quiet one, is where most brands are leaving money on the table.

Here's the thing. Most founders I talk to treat offers like a once-a-quarter event. Big push at Easter, big push at BFCM, then a long flat stretch of running the same evergreen creative into the same evergreen landing page and wondering why the hit rate keeps sliding. The whole year hangs on a handful of moments, and every one of those moments is a guess you only get to make four times.

I'd rather you make that guess every single week.

Why offers, not creative, are usually the thing that's stuck

When a brand comes to us flat, the instinct is always "we need better ads". And sometimes that's true. But more often the ads are fine and the offer underneath them is the problem.

To put this into perspective: if your offer genuinely makes people want the thing, your hit rate on new creative might sit around 6-7%. If the offer is weak, you're closer to 2%, which means you're burning roughly three times the spend per ad to find a winner. Same creative team, same budget, wildly different outcome, and the difference is almost entirely the offer.

So the move isn't to make more ads first. It's to find an offer people actually want, then let your existing creative ride it. The trouble is that testing offers is normally expensive and slow. You change the front-end promo, you eat a few weeks of soft numbers, you find out it didn't work. Nobody can afford to do that twelve times a year.

Unless you stop testing on evergreen.

The four-moments-a-month idea

Here's the structure I'd run, and it's the same one we set up for clients who want to test constantly without bleeding margin.

Every month, you build four marketing moments. Two prominent, two unlisted.

The two prominent ones are your loud sales. Hero banner, full email sends, the announcement bar, the lot. These are tied to moments your audience already feels: a seasonal change, a holiday, a known retail peak. You're not inventing demand here, you're showing up when it already exists.

The two unlisted ones are the quiet ones. They live in the announcement bar and on a dedicated landing page, and that's it. No hero banner. No email blast to the whole list. You are deliberately not telling everyone. And that restraint is the entire trick, because the unlisted moment is your testing lab.

Why "unlisted" is the safest place to test anything

The reason testing usually loses money is that you're running the experiment in front of your whole audience, on your most valuable real estate, during a window that matters. If it flops, you've burned a real moment.

An unlisted sale flips that. You wrap a new offer around a smaller, lower-stakes occasion, something like an Earth Day angle, a "cardio week", a back-to-school tie-in, whatever loosely fits your category. Because it's attached to a real moment, even a minor one, you won't lose money testing it. Some of the people who buy will be your core customers anyway. The moment just gives the offer a reason to exist.

I'll be honest about my bias here: I don't think you should ever test an offer cold, on a random Tuesday, with no occasion around it. "Dog food" converts worse than "dog food for German Shepherds" run during a moment when buying that thing makes sense. Context does half the work.

Run it as incremental campaigns, never an A/B split

This is the part people get wrong, and it's worth slowing down on.

When you've got a new offer to test, the temptation is to use a tool that splits your traffic, 50% to the old page, 50% to the new. I'd avoid that entirely. On a moment that matters, a split test means half your traffic goes to the version that loses, and you've handicapped yourself on a day you can't get back.

Instead, you stack. You launch a brand new campaign to a new landing page with the new offer, running on top of what's already live. You're not stealing budget from the control. You're spending incremental money to find out if the new thing works.

The maths on this is simple and it's why I like it so much. If you split-test, your downside is real, you lose on half your traffic. If you launch incrementally, your downside is close to zero. You're just adding spend to test a new angle, and if it doesn't fire, you turn it off. When the downside is basically nothing, the answer to "how many offers should I test?" becomes "as many as you can build."

The graduation rule

Testing is only worth it if winners go somewhere. So here's the rule that closes the loop.

When an unlisted offer beats your benchmark, two or three times across separate moments, not just once, you graduate it. You strip the occasion off it and make it evergreen.

That's the whole point of the system. You took "Earth Day bundle", proved the bundle and the landing page and the angle all convert, and then you quietly remove "Earth Day" and keep everything that was working. Now it's your evergreen front-end offer, and it earned that spot by winning in a low-risk container instead of by you betting on it cold and hoping.

One win isn't a winner, by the way. A single good month can be timing or luck. I want to see an offer clear the bar more than once before it earns a permanent place in the funnel.

A monthly calendar you can copy

Here's roughly how a month looks when you run this. Adjust the moments to your own category and seasonality.

• Week 1 - Prominent moment. Your big seasonal push. Hero banner, full sends, known peak. This is a revenue moment, not a test.
• Week 2 - Unlisted test. New offer wrapped around a minor occasion. Announcement bar plus a dedicated landing page. New incremental campaign, no email blast, no split test.
• Week 3 - Prominent moment. Second loud sale. Different angle or bundle from week 1 so you're not just repeating yourself.
• Week 4 - Unlisted test. Second quiet experiment. Try the offer type you're most unsure about here, since the downside is lowest.

Four moments, two of them quietly testing your way toward your next evergreen winner, every month. Over a year that's twenty-four low-risk offer tests instead of the three or four big swings most brands take.

A quick word on judging it. Don't measure these on revenue alone, because the prominent moments will always look bigger. Judge each one on whether it cleared your conversion benchmark for that offer. An unlisted test doing strong numbers on tiny spend is a graduation candidate. A prominent moment that only looked big because you emailed everyone might not be.

Where this leaves you

Most brands run a handful of offers a year and call the flat months "just how it is". This system says the flat months are where the testing should be happening, quietly, in the announcement bar, with the downside engineered out.

If you've never sat down and mapped which of your offers actually beats your benchmark, and which ones just look fine because you've never tested an alternative, that's the gap worth closing first. A Signal/Noise Audit is one way to see it laid out: where your current offers really sit, and which quiet test would tell you the most. Or just block out next month's calendar into four moments and run the exercise yourself. Either way, the question I'd start with is the same one: which of your offers have you actually proven, and which have you only ever assumed?