Friction reveals truth. Find it before ecommerce peak season does

Edition 055 · 29 October 2026

Black Friday is four weeks tomorrow. Somewhere in your business there is a weak point that peak will find, and the only question is who finds it first. Customers find it by abandoning a basket. Your team finds it by arriving at 9pm to a queue no one planned for. You can find it this week, in 35 minutes, with a spreadsheet and a willingness to be a bit pessimistic.

Previously in The Calm Advantage: in The Live Stress Test Begins, I argued that peak is the best customer research you’ll get all year, and that the plan you already have is the experiment. The exercise was a Peak Log and a baseline of eight numbers from a normal week.

This is part two of six. The Calm Advantage is a weekly series that runs to 26 November, and every edition ends with a short exercise that produces one document. You can read this one on its own. The only thing it borrows from part one is a baseline, meaning eight numbers from an ordinary week: conversion, checkout completion, average order value, orders per hour, support contacts per 100 orders, page speed, payment failures and delivery contacts. If you don’t have them, an hour with your analytics and your support tool gets you rough versions, and rough is enough to start.

Learn from ecommerce peak season by looking for the weak points before it arrives

I said last week that I’d rather you used peak as research than reacted to it. This is the first piece of that. If the aim is to learn as much as possible from the way customers behave at your busiest, you need to know in advance where you expect the strain, so you can see if you were right. A prediction you wrote down on 29 October is far more use in December than a recollection of how things felt.

It also works if Black Friday isn’t your peak. Swap the date for yours and count back from it.

Why a pre-mortem beats a post-mortem

The usual way to learn where a business bends is to wait for it to break and write up the damage afterwards. A pre-mortem does that job early. The technique comes from the psychologist Gary Klein, who described it in Harvard Business Review in 2007: you imagine the project has already failed, and then work out why. You pretend it is Saturday 28 November, peak went badly, and you explain what happened.

I like it because it changes who speaks up. In a normal planning meeting, raising a risk sounds like negativity. In a pre-mortem, the failure is a given, so naming a cause is the job. People who have been quiet about a worry for months will say it in the first five minutes, and the answers come out specific: the courier cut-off, the one promotion code no one tested on mobile, the report only one analyst can run.

A weak point you’ve named has an owner and a number attached to it. One you haven’t named has neither, and it will introduce itself at the worst moment.

Five friction points to check before Black Friday

I use five friction points, because they cover the places a retail business tends to bend under load, and they’re broad enough that every team can fill them with its own detail: checkout, fulfilment, communication, confidence and clarity.

Checkout is where load turns small faults into lost orders: payment challenges that fail more often as volume rises, a decline screen that only says “payment failed”, a promotion code that behaves differently on mobile.
Fulfilment is where a promise made on the site meets a courier cut-off, a picking rate or a stock count that can’t keep it. The commonest version is a stock or order feed lagging between the platform and the warehouse system, so the site keeps selling what the warehouse no longer has.
Communication is where two parts of your own business say different things, such as delivery cut-offs worded one way on the product page and another in the basket, with customer care picking up the contacts.
Confidence is how sure the team is that the numbers are true. Trading decisions made on yesterday’s figures, or on a dashboard that lags by an hour, are guesses with a chart on them.
Clarity is who has the authority to pause a promotion at 9pm, and if they know it.

Each of those can fail for one of four reasons: design, process, system or leadership. Tagging the reason matters, because it decides who fixes it. A leadership gap doesn’t get closed by a developer, and a system gap doesn’t get closed by a reminder email. I’ve seen a team spend a fortnight on a technical fix for what was really a question of who was allowed to say no.

Start from your own evidence rather than a blank page. Read back through the Peak Log, including whatever Prime Big Deal Days on 6 and 7 October and the first October promotions threw up, and let anything that cost you minutes or money suggest the first rows. A wobble at 1.2 times normal traffic will be worse at three times, so treat it as a preview.

What the clock tells you

In early September, the Magento and Adobe Commerce flaw now called StyleSmuggler was exploited for the first time, and one server Sansec tracked went from first hit to fully compromised in about fifty minutes. I wrote about it in Where Were You on the Clock?, and the point carries over to peak: the teams that came through it well could say, in order, what they knew and when. What made that possible was an owner and a trigger decided before the event.

The Friction Watch List applies that discipline to peak. For each friction point you write down the likely failure, the number that would show it first, the threshold at which someone should look, and the person who looks. When 9pm on Black Friday arrives, no one has to debate if something counts as a problem, because the sheet has already said.

Picture 7:15pm on Black Friday

A customer care adviser posts in the channel: “where’s my order” contacts are climbing, and they’re all from people who ordered after lunch. The product page and the basket both say next-day delivery. The promotion email says order by midnight for next-day. The courier collected at 4pm, so everything placed since then goes out on Monday.

Nothing was broken. Marketing wrote an offer that ends at midnight, the site kept running the delivery promise it was built with, and the warehouse hit its cut-off. Three teams did their jobs and the customer was still promised something the business couldn’t deliver. Now someone has to choose between changing the banner, amending the email and calling the courier, and at 7:15pm on Black Friday it isn’t obvious whose decision that is.

That one is ordinary: fulfilment, communication and clarity bending at once, and all three are on your list. The question that finds it is “where do two teams make promises that depend on a third?” Ask it at every handover and you’ll have a good start on your list within the hour. The pre-mortem exists so you find these on a calm Thursday, when changing a banner is a five-minute job and no one is waiting for you to decide.

What a good row looks like

A weak row reads “checkout might struggle”. It can’t be wrong, so it can’t teach you anything. A strong row reads like this:

Checkout – payment authentication failures can climb as volume rises, mostly on mobile, and the cause may be the card issuer or the payment provider rather than your site. Early warning: authentication failures above 4% of attempts for 30 minutes (an example figure; set yours from your own payment data and provider). Gap: system. Owner: payments lead, covered by the head of development.

Every part of that can be checked on the night. The number is one your provider already reports, the threshold says when to act, the gap tells you which team gets the fix, and there’s a named person. If a row doesn’t have all four, it isn’t finished, and it will be the first one people skip when they’re busy.

The rows you can’t put a number against deserve a second look too. If the early-warning signal for communication is “customers seem confused”, you don’t have a signal yet. Next week’s edition is about building them, so for now write “no signal yet” and leave it visible. A gap you can see is better than a row that looks done.

Don’t do this ecommerce peak season exercise alone

I’d run this as a 35-minute conversation with three or four people rather than a solo task, if you can. Include one person from trading, one from tech or platform, and one from customer care or fulfilment. They each see a different half of a weak point, and the row on delivery promises only appears when the person from the warehouse is in the room. If you can’t get them together, send the five headings round and ask each person to add the one row they’d put money on. You’ll get decent coverage in an afternoon.

One caution on the numbers. The thresholds on a first watch list are guesses, and that’s fine. You’re writing down what you’d consider “look now” so that in December you can see if you were too jumpy or too relaxed, and adjust. A threshold you can be wrong about is better than none.

This weeks exercise: the Friction Watch List (35 minutes)

You’ll need your Peak Log and baseline from last week.

Take the five friction points: checkout, fulfilment, communication, confidence and clarity.
For each one, write the likeliest way it fails under peak load. One sentence.
Tag the gap: design, process, system or leadership.
Write the early-warning number, using your baseline where you can, and the threshold that means “look now”. If checkout completion is normally 64%, “below 58% for 30 minutes” is a threshold.
Name one owner by role and one person who covers when they’re off.
Stop at five rows, or ten at most. A watch list that takes more than a minute to read won’t get watched.

Pin the finished list somewhere the team already looks, and add its date to the top of your Peak Log. Next week it becomes the starting point for listening to your customers.

This week’s worksheet is chapter 2 of the playbook, so if you’d like the full set with a worked example to compare against, pre-register here.

If you’d rather not run the debrief alone in January, JH runs a Post-Peak Recovery Workshop for retail teams in Q1 2027.

The Calm Advantage, all six parts

Part 1, 22 October: The Live Stress Test Begins
Part 2, 29 October: Friction Reveals Truth (this edition)
Part 3, 5 November: Your Customers Are Already Teaching You (coming)
Part 4, 12 November: The Hidden Cost of Heroics (coming)
Part 5, 19 November: The Debrief Discipline (coming)
Part 6, 26 November: From Observation to Clarity (coming)

Next week

Next Thursday: Your Customers Are Already Teaching You, on where customer signals get caught during peak, who reads them, and what normal looks like by traffic source, including AI referrals. Bring your watch list.

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