Black Friday 2026: what to personalise in your store three weeks out

Three weeks out from Black Friday (Friday 27 November 2026, with Cyber Monday on Monday the 30th) it is too late to launch new A/B tests, but not too late to prepare personalisation: different messages depending on the campaign a visitor arrives from, whether they are new or returning, their device and how close they are to free shipping. You set them up now, review them calmly, and on the 27th all you have to do is watch. This is what we prepare with the stores we work with.

An A/B test compares two versions to learn which works better, and it needs weeks of normal traffic to give a reliable answer. Black Friday week is the opposite of normal traffic. People who do not know the brand arrive in a hurry, looking at price, and leave on Monday. What you learn those days does not hold for the rest of the year, and if the test ends mid-campaign you decide on odd data.

Personalisation is not trying to learn anything. It decides what each kind of visitor sees according to a rule: "if they come from the customer email, show them their code". That does make sense during a campaign, because it responds to situations you know in advance.

If you want the calendar for what to test in October, what to freeze in November and what to leave for December, we covered it in Black Friday for a new brand. Here we stick to what can still be done.

That week you will get traffic from several places at once: the email to your list, Instagram ads, perhaps Google Shopping, a creator with their own code. Each visitor arrives with a different promise in mind.

What works is for the store to carry on the conversation the ad started:

• People arriving from the customer email see a bar saying "Your early access: 20% off with the code from the email", without going back to their inbox to find it.
• People arriving from an ad for a specific product see that product featured on the homepage, not the usual carousel.
• People arriving from a creator see that creator's code already written in the notice.

All of this runs on rules about UTM parameters. Agree with whoever runs the ads that every campaign has properly written UTMs before the 20th. It is what fails most: an untagged campaign falls into "direct" and never gets its message.

A new visitor needs trust: who you are, how long shipping takes, how returns work. Someone who already knows you wants to get to the point.

For new visitors we reinforce trust signals on the product page and cart: real delivery time, a one-line returns policy, reviews. For returning visitors we cut the noise and give them a direct way into the deals section.

A word of caution: "returning" means that browser has visited the store before, not that it has bought. If you want to tell existing customers apart, the store has to tell the tool (in Bixlow that is done with identify). If you do not have that set up already, do not improvise it the week before.

On almost every store we see, most campaign traffic comes from mobile, from someone who has just watched an Instagram story. Check it on your own phone, on mobile data and not the office wifi:

• The offer bar does not push the buy button off the first screen.
• The discount code can be copied with one tap.
• Checkout does not force account creation before payment.
• The payment methods people actually use on mobile are visible.

You can serve a shorter version of the bar or the notice on mobile only. That is device-based personalisation, not a test.

Free shipping over a set amount is the clearest lever on order value for a small store, and on Black Friday many people land just below it. Prepare:

• A cart notice showing what is left: "You're €8 away from free shipping".
• Suggestions of inexpensive products that complete the order.
• If you change the threshold for the campaign, make the message identical everywhere. Nothing is more frustrating than reading "free over €39" in the header and seeing €50 in the cart.

Check delivery times too. Carriers are overloaded that week and deliveries take longer. If you will be slower, say so on the product page. A late order turns into a return or a bad review in December.

A countdown to the end of the offer works because it is true: the offer ends on Monday at 23:59. A countdown that restarts every time you reload is a lie, and customers notice.

The rules we follow:

• The countdown shows a real time, the same for everyone.
• When it hits zero, the offer actually ends.
• No "only 3 left" unless it comes from real stock.

And on pricing, remember the EU 30-day rule: if you advertise a discount, the previous price you show must be the lowest you charged in the preceding 30 days. Raising prices in early November to cut them on the 27th is not just ugly, it can get you into trouble.

• Nothing in the checkout. If a change breaks payment on the 27th, you lose the most important day of the year. Freeze the checkout from mid-November.
• Nothing shorter than the campaign. A test that starts on the 25th and is read on the 30th measures traffic that looks like no other week.
• Nothing on price. Showing different prices to people who will compare them in a group chat is a bad idea at any time, and worse that week.

If you have a test running since October, consider pausing it before the 23rd or at least analysing the campaign days separately.

1. We install the script (if it is not there yet) in the head or through Google Tag Manager and check that it loads on the homepage, product page, cart and checkout.
2. We create the audiences: "Customer email" and "Instagram" by UTM, "Mobile" by device, and "New" and "Returning".
3. We ask the agent for each personalisation in plain words, for example: "for the Customer email audience, show a bar with the code EARLY20 above the header".
4. We review each one in preview, with each campaign's URL and UTM. Nothing goes live without approval.
5. We publish them before the 20th, so they have been running for a week when the peak arrives and there is time to spot anything wrong.
6. We leave an order completed goal in place (by confirmation URL or a dataLayer event) to see afterwards how each audience responded.

On Monday the 30th at night, switch off the campaign-only ones. Put it in the calendar, because on Tuesday at nine nobody remembers.

Can I run an A/B test on Black Friday if I have huge traffic? Lots of traffic helps you reach the sample, but it does not change the fact that this traffic behaves differently from the rest of the year. What you learn may not apply in January.

When should everything be published? At least a week before the peak. For 2026, Friday 20 November is a good deadline.

Does personalisation affect SEO? A different bar or notice depending on the campaign does not change the page's main content. Google sees the normal page.

What if my campaigns have no UTMs? Add them now. Without them you cannot tell where each visitor comes from or personalise by campaign.