What should I fix on my website before Black Friday?

November has quietly eaten December.

Statistics South Africa publishes the cleanest view of this. In 2011, the year before Black Friday appeared online here, November was 42.5% of combined November and December retail sales. By 2025 it was 45.9%. The month has shifted: the spending that used to wait for December now happens before it.

Two things in that data are worth more than the headline. The high was 2019, at 46.4%, so 2025 sits just below a peak set six years earlier. Black Friday did not keep growing; it arrived, moved the money, and settled. And Statistics South Africa says plainly that it cannot isolate the day itself from the month: November 2025 came to R117.9 billion in inflation-adjusted sales, 3.6% up on the year before, and how much of that was Black Friday is not something the data can tell you. They compare November against December because that is the honest proxy available.

So the useful conclusion is not "Black Friday is getting bigger". It is that a large share of your year now lands in one month, and the month is nearly here.

What actually breaks.

Not the discount. The plumbing around it.

A payment processor, Peach Payments, reported about 2.69 million transactions worth roughly R1.86 billion across the four days from Black Friday to Cyber Monday in 2025, up 93% in rand value on 2024. That is one processor's own slice of the market rather than the market, but the shape of it is the point: the traffic arrives compressed into days, and anything that is marginal on a normal Tuesday fails under that.

Three failures account for most of it, and all three are testable today.

The page a machine reads is not the page you see.

Before anyone buys, something has to find you. A search engine or an assistant reads your page as it leaves the server, before anything in the browser runs. If your words arrive after that, they may as well not arrive.

This is not theoretical for us. We ran our own sixteen checks on our own old homepage and got 15 out of 32. On the rebuilt site we get 32 out of 32. Same instrument, our homepage against our homepage, and we publish the result because we would rather show it than describe it. If your offer lives inside a script that runs late, the machine sending you customers in November never sees it.

The site that is fine on a Tuesday.

Nothing on your site is tested by a normal week. The traffic does not spread itself politely across the month, and a page that is merely slow on a quiet Tuesday is a different proposition when everybody arrives at once.

You cannot simulate that properly without tools, but you can ask three questions and have them answered in writing before November.

Ask whoever hosts the site what happens at ten times your normal traffic, and whether it costs you anything when it happens. Ask whether the pictures on your busiest pages are the originals straight off a camera or sized for the web, because unsized images are the commonest reason a page that was fine becomes a page that is not. And open your own site on a phone, on mobile data, away from the office, because that is the connection most of your customers will be on and it is nothing like the one you test on.

None of those is a rebuild. All three are a phone call and ten minutes, and you will know by the end of the week whether you have a problem or not.

The enquiry that arrives at nine at night.

Black Friday does not respect office hours. The question is not whether you can reply at nine at night, it is whether the enquiry still exists in the morning.

Most of the losses we find are not dramatic. A form that sends to a mailbox nobody owns. A notification that fires before the record is saved, so the record is never saved. A WhatsApp number on the site that belongs to someone who left. None of those show up in a dashboard, because a lost enquiry leaves no trace of itself.

That is the awkward part of this one. The other two failures are visible once you look. This one is invisible by design, because the only evidence of it is something that did not happen, and the only way to find it is to send yourself an enquiry and watch what the business does with it.

What to do with the three weeks you have.

In this order, because the order is the saving.

1. Read your own page the way a machine does. If the words are not in the served page, nothing else on this list matters.
2. Send yourself an enquiry from a phone, on data, at night. Then check it arrived, check who owns the mailbox, and reply to your own message.
3. Ask the three hosting questions and get the answers in writing.
4. Decide the one thing you are promising and put it where it is read first. A page offering everything at a discount reads as nothing at a discount.
5. Leave the redesign alone. Three weeks before your biggest month is the wrong moment to change the structure a search engine depends on. Fix, do not rebuild.