The Live Stress Test Begins

Edition 054 · 22 October 2026

This is the time of year I call The Calm Advantage. The best brands aren’t reading for last-minute tips to bolt onto the site, and they aren’t looking to pivot away from plans that are already being executed. They’re focused on learning as much as they can from the best data treasure trove of the year, which is their customers at peak.

It applies just as much if you don’t think of Black Friday as a peak at all. So buckle up, and let’s go through it.

Black Friday is five weeks away. Amazon’s Prime Big Deal Days ran on 6 and 7 October, shoppers have been comparing prices since. Your October promotions are already producing the first real data on how your site, and your team and your suppliers behave under load. Some teams will treat that as noise on the way to November. I’d treat it as the warm-up, and as your chance to set up what you’ll capture when peak arrives.

This is the first of six Thursday editions on this theme.

They run every week until 26 November, the day before Black Friday, and each one ends with an exercise that takes 20 to 45 minutes and produces one document.

Each edition stands on its own, so you can start anywhere, but they build in order, and I’ve written them to be read that way. By the finale you’ll have a complete peak kit, and Peak to Plan on 3, 10 and 17 December picks up from there to turn what peak teaches you into the plan for 2027.

If you read last year’s edition of the same name, The Live Stress Test Begins, you’ll recognise some of the thinking.

This year, the order is different: we prepare to capture first, and reflect afterwards.

Peak is the best research you’ll get all year

Think about what a peak week is: as a source of data. You get more visitors than at any other time of year, many of them arriving with intent, and with less patience for anything that gets in their way. A quiet month can’t show you how they behave under pressure, and you can’t commission it. A peak week shows you:

what people search for and can’t find
which products, offers and journeys convert best, so you know what to scale
where they stop in the basket, and what they did just before
what they ask support, in their own words
which payments fail, and for which customers
how they react to a delivery promise you can only just keep

Each of those is customer evidence sitting in a system you already own. The trouble is that peak is also the week your team has the least attention to read it.

The temptation is to react. A banner gets changed at 4pm because the site felt slow, a promotion is swapped because a competitor moved, delivery wording is rewritten on a hunch. Each one is a guess made on a thin slice of data, and each one muddies the picture, because you can no longer say which change moved which number.

I’d treat the plan you already have as the experiment. Leave it alone unless something is clearly broken, decide what you’re watching before the week starts, and write down what you see. That gets you ready to capture the right things over the next six weeks, so that in December you’re learning from a proper record rather than a vague memory of a stressful week.

I wrote about the checkout no one checked after launch for that reason: a problem you can’t compare against anything gets found late.

What I mean by calm

A calm team can afford to wait for evidence. A reactive team acts on the first number that looks wrong, usually before it knows what the number means, and the fix often becomes the next problem. Calm comes from decisions made in advance: what you’re watching, what counts as a problem, who looks, and who is allowed to act.

None of that needs a new tool. It needs a few documents and an hour of someone’s attention. That’s what the six weeks build.

The framework underneath is last year’s, C.A.L.M., which stands for Capture, Analyse, Learn and Measure.

Weeks one to three are Capture: what normal looks like, where things will bend, and what customers are telling you.

Weeks four and five are Analyse and Learn: who decides what when the evidence comes in, and a debrief designed before peak so that it happens.

Week six is Measure: the diary and the one-page plan that turn what you learned into next year’s plan.

If Black Friday isn’t your peak

Plenty of businesses don’t trade this way. Your year might turn on Mother’s Day, back to school, a January sale, a trade show or a financial year-end. The exercises work from whichever date you count back from, so swap 27 November for yours. If your peak is later, build the kit now and let this autumn’s trading be the rehearsal. If it has just finished, start with the log while the detail is fresh.

The playbook

Everything you build over the six weeks will also be collected into The Calm Advantage Playbook 2026: the six worksheets, a worked example following one fictional retailer through all of them, and a one-page plan template for the weeks after peak. It’s free, and it lands in the inbox of everyone who pre-registers in the week of 30 November.

Pre-register here.

You don’t need it to follow the series. Each exercise stands on its own, and the playbook is the set in one place, with the example to check your own against.

What the six weeks add up to

Each piece uses the one before it, so the order matters:

22 October, The Live Stress Test Begins: the Peak Log and a baseline, so you know what normal looks like.
29 October, Friction Reveals Truth: a Friction Watch List built from the baseline, for the places checkout, fulfilment, communication, confidence and clarity are likeliest to bend.
5 November, Your Customers Are Already Teaching You: a Customer Signal Plan covering where customer signals get caught and who reads them, including where your traffic is coming from.
12 November, The Hidden Cost of Heroics: a roles and decision card, so the person on call at 9pm on Black Friday knows what they can decide.
19 November, The Debrief Discipline: a debrief designed before peak, with the date pencilled.
26 November, From Observation to Clarity: the calendar that books the debrief, and the empty one-page plan you’ll fill in afterwards.

A baseline doesn’t need to be perfect

The exercise below asks for eight numbers from a normal week. The likeliest reason for skipping it is that the data isn’t clean: analytics set up differently on mobile and desktop, a tracking change in the summer, an order system that counts partial refunds its own way.

I’d still do it. Pick the cleanest recent week you can find, write down where each number came from, and note anything you don’t trust. A baseline with a footnote is far more useful than none, because the footnote tells the next person how much weight to put on it.

Choose the week carefully: recent, no promotion, no incident, no bank holiday. If every recent week had something in it, take the nearest to normal and say so. Black Friday itself will run differently from any normal week, so use the baseline for direction and ask what last year’s Black Friday looked like for scale. Where you can, note the weekday and the hour as well as the number.

The second thing is ownership. The log needs one named owner, because a shared document with no owner turns into a thread that no one reads by the second week. The owner doesn’t write every entry. They make sure entries get written, keep the format tidy, and have it open on the day.

Picture 9pm on Black Friday

It’s Friday 27 November, a little after nine. Trading has been strong all day, and then someone in the channel says conversion looks low. Someone else says checkout feels slow. A third person says it was like this last year, wasn’t it?

You can feel what happens next. Three people hold three opinions, and none of them can be checked. Is 1.6% low? Low against what: yesterday, which had a promotion, a quiet Tuesday in October, or last Black Friday, on a different version of the site? Is checkout slow for everyone, or only on mobile? Has anything been changed that might explain it, or did someone alter the delivery banner at five and forget to say?

Run that evening again with two things in place.

There’s a baseline saying a normal week converts at 2.0% and checkout completes at 64%. There’s a log saying the delivery banner changed at 4:40pm.

The conversation takes two minutes: conversion is 20% down on baseline, it started around five, here is what changed, put it back and watch the number. No one argues about how things feel, because the numbers and the notes are already written down.

If you take one thing from this series, I’d want it to be the gap between those two evenings. This week’s exercise builds it, and it takes half an hour.

This week’s exercise: the Peak Log and baseline (30 minutes)

You’re creating one document that holds everything you learn from now until the debrief, and fixing the numbers that say what normal looks like.

Create the document and name one owner. Everything from the next five weeks goes into it, so don’t start a second one.
Set an entry format: date and time, what happened, where, who noticed, and what it cost in minutes, money or mood.
Pick a normal week. It should be recent, with no promotion, no incident and no bank holiday.
Fill in six site numbers for that week: sessions, conversion rate, checkout completion (checkout started to order placed), mobile checkout speed (75th percentile Largest Contentful Paint), customer contacts per 100 orders, and the share of orders dispatched inside the promised window.
Fill in two team numbers for that week: messages sent between 6pm and 7am in your trading, ops and dev channels, and out-of-hours escalations.
Write the first entry. Anything from the last fortnight counts, including the Prime Big Deal Days effect.

That’s 30 minutes, and it’s the document every later exercise writes into. If you ran a log last year, pull it out now and use its entry format.

This week’s worksheet is chapter 1 of the playbook, and you can pre-register for the whole set here.

This post was originally written for CEO & Founder, Jamie Huskisson’s LinkedIn newsletter, Breakthrough Commerce.