What 4.3 million A/B tests show about Black Friday: what shops test, which offers they keep, and how to set up your plan before traffic is at its most expensive.
Before Black Friday, test the offer itself, not just the page. In our database, shops keep a discount with a code far more often than the same discount applied automatically, early access more often than an open sale, and a gift more often than free shipping. Start the tests before Black Friday week, so you go in with a tested offer.
01Why test before Black Friday?
Because Black Friday traffic is the most expensive of the year. The question is not how to buy more traffic, but how to get more revenue from the traffic you already pay for.
The full talk, in German (9 minutes).Q4 traffic is the most expensive of the year.
That is why the offer is the thing to optimise most: test different discounts against each other, different prices, different offers. Ideally before Black Friday, if you run a sale beforehand. Then you go into the week with an offer that is already tested.
02How do we know what works on Black Friday?
From 4.3 million A/B tests across 151,000 shops, collected since 2011. A test changes the page, so we can see it. That is where the Likelihood to Win comes from, our prediction of which test ideas are likely to win.
Prediction-Based Experimentation: our own tests.
Usually you run one test a week, one after another, and look at what competitors do on the side. Our method is called Prediction-Based Experimentation: it uses the database to predict which ideas can work for your industry or your shop. 27% of our own tests won in 2024, 56% between April and June 2026.
4.3 million A/B tests from 151,000 shops, since 2011.
03What do shops test around Black Friday?
One test in six around Black Friday is about the offer, and Black Friday week is the busiest. Shops mostly test the promotion: countdowns, early access and pop-ups, hardly the product page.
About 35,000 offer tests from 2021 to 2025.
In my opinion that is too few. More shops should test their offer. What people test looks like this: a spin-the-wheel before checkout, an early Black Friday timer, 35% off as a pop-up, a countdown on the home page, a price guarantee on the product page.
Real Black Friday tests, brands removed.Shops test the promotion. Not the shop.Most tests start just before.
04Which Black Friday offers do shops keep?
A discount with a code, early access and a free gift are kept most often, followed by an offer for a newsletter sign-up. An automatic discount, countdowns and bundles do worse, free shipping worst.
13,600 discount and offer tests, against the average test (1.0).Five things offer tests show.
Code beats automatic: a discount unlocked with a code is kept far more often than the same discount applied automatically.
Early access beats an open sale. Not much of a secret any more, but it works a lot better.
Bigger shows no edge. 10, 20 or 30% off make no difference to what gets kept. The question is whether it makes you more profit.
Free shipping is kept least, on every tool we checked.
Countdowns show no clear edge. We have seen client countdowns win, but the margin is much closer than with the code.
Same discount. A code gets kept more often.
There is psychology behind it: whoever types a code feels they got something. So we would rather recommend the code than the automatic discount.
05What do our own tests show?
That the promotion does not pay off by itself. Taking SALE badges off the listing pages won; a newsletter pop-up with 5% off lost; a free shipping bar changed nothing. So: test it first.
Three of our own tests.What the kept Black Friday version usually looks like.
The typical kept Black Friday version reads: X% off, with a code, until a date. For example 20% off your favourite item, 22% off everything for Singles Day, or a free organic gift with a code.
Everyone tests the promotion. Nobody tests the product page more.
06Which tests can you copy before November 27?
Five: give the discount a code, give subscribers early access, test a smaller discount, test a gift instead of free shipping, and run it now rather than four days before.
The big brands have figured it out: code, early access, gift.Urgency, amounts and codes.
The big brands have figured this out. We play Robin Hood a little: we look at what the big ones already found out, and what you can learn from it.
07How do you test safely on Black Friday?
With an automatic stop. On Black Friday nothing can go wrong. In Apex, Guardian stops a variant automatically as soon as it loses more revenue than the limit you set.
A code. In our database the same discount is kept far more often with a code than applied automatically, as long as the code is easy to understand and easy to see.
Of all offer types, free shipping is kept least often. A gift or a discount with a code are the better candidates for your test.
Now, not in Black Friday week. A test needs full weeks of data. Test beforehand, ship the winners and leave Black Friday week alone.
Book a call with the team and tell us about your shop. We'll talk through where it stands, where you want it to go, and how we'd get you there. If working with us isn't the best return on your money, we'd tell you rather than take it.
