Black Friday: the changes to make and the ones to postpone

Most peak-season advice is a list of things not to do in November. That is correct and not very useful, because the reason a change is dangerous in November is that it has no time left to fail in front of you. The same change in February has nine months to surface a problem while someone is still calm enough to fix it.

So the first sort is reversibility. A change a merchandiser can undo in the admin in thirty seconds — a price, a collection, a banner, a piece of copy — carries almost no schedule risk and can be made any week of the year. A change that needs a deploy, a vendor, or a migration carries schedule risk regardless of how small it looks, because undoing it is a project rather than a click.

The second sort is the calendar. Structural changes want the first three quarters. Not because autumn is cursed, but because the value of a structural change is largely the problems it exposes, and you want those exposed in a month where the cost of finding one is a bad afternoon rather than a bad year.

Written down, that is the whole framework. Everything below is which changes fall where.

January to March is when the structural work is cheapest, and it is also the only time of year when you still remember accurately what broke. Four things earn the slot.

Remove the scripts. Shopify’s own theme performance guidance is to identify and remove or defer app scripts that block HTML parsing before any content is rendered, and — a specific and frequently ignored instruction — to remove A/B test anti-flicker snippets when no tests are running. Both are trivially safe in February and unthinkable in November, because removing an app is the kind of change whose side effects appear on a page nobody checked.

Clear the checkout.liquid customisations, if any survive. This is not housekeeping. Shopify documents that qualifying for the advanced discount combinations — product with order, order with order — requires a store without checkout.liquid customisations. A store still carrying them has a smaller promotional vocabulary than its competitors on the one weekend that vocabulary matters.

Instrument the integrations. Every feed to an ERP, a 3PL or a warehouse has a throughput it was sized for and a failure mode nobody has watched. Finding out which is a week of unglamorous work, and it is a week you can only spend when orders are not stacking up behind it.

Move merchandising out of code. This is the highest-leverage item on the list and the least obviously peak-related. Every campaign page that is a bespoke template is a page that can only be changed by a deploy, which means it cannot be changed during a freeze. Every campaign page assembled from tested, reusable sections is a page marketing can rebuild on the day. The work of converting the first kind into the second is a first-quarter project that pays out in the fourth.

If you make one peak-specific change before September, make it this one: build the actual promotional structure you intend to run, on a real store, and try to break it.

Shopify’s discount model has three classes — product, order and shipping — and they do not all combine. Product discounts can combine with free shipping and with other product discounts on different items; order discounts can combine with free shipping and, subject to eligibility, with other order discounts and with product discounts. Multiple product discounts on the same item are a Shopify Plus capability. Multiple shipping discounts cannot combine with each other at all.

The limits are worth knowing before you plan around them rather than after. Shopify documents a maximum of 25 active automatic discounts across the store, and allows a customer to apply up to 5 product or order discount codes plus 1 shipping discount code on a single order. Application is sequential: product discounts apply to individual items first, order discounts apply to the revised subtotal, and shipping discounts apply last.

That last detail is the one that produces the angry finance conversation. A 20% site-wide product discount followed by a 10% order discount does not take 30% off the order; it takes 10% off a subtotal that is already 20% lower. Whether that is what you intended is a commercial decision, and the time to discover which one you configured is a Tuesday in September, not the Friday itself.

These are the proposals that come up every autumn. Each of them is a good idea in the wrong month.

A replatform or a theme rebuild. The entire value of a rebuild is the list of problems it exposes. Launching it into peak means discovering that list at the exact moment you have least capacity to act on it.

A new app — especially a performance app. Installing a script to make the page faster is a plan with a visible contradiction in it. Worth noting how apps actually attach themselves: theme app extensions render app embed blocks immediately before the closing </head> and </body> tags, and Shopify documents that app embed blocks are deactivated by default after install, though apps can supply a deep link that activates them. So an install is one click from adding code to the head of every page on your store, and nobody on the team necessarily knows it happened.

A checkout change. Even a well-scoped one. Checkout is the one surface where a defect converts directly into lost orders rather than into a worse experience.

A new integration. An ERP or WMS connection sized against normal volume meets peak volume for the first time in its life on the worst possible weekend. Ship it in spring and let it see a quiet quarter first.

A redesign of the product page. Not because it will not help, but because you cannot tell whether it helped: the traffic mix on Black Friday bears no resemblance to a normal week, so peak is the worst possible period from which to read a conversion result.

A freeze that stops the business from trading is a freeze nobody will honour, so it is worth being explicit about what remains open: content, merchandising and pricing. Those are the levers the business genuinely needs during peak, and they share a property — a single person can revert them in the admin without a deploy.

The practical goal of everything in the first section is to widen that category. A store where the homepage, the campaign landing pages, the navigation and the promotional messaging are all admin-editable can respond to a bad Friday morning in ten minutes. A store where any of those requires a developer has to choose between shipping during a freeze and not responding at all, and it will pick the first one at about 9am.

One thing to be realistic about: you will not be watching performance data live. Shopify notes that data in the Web Performance report can be delayed by up to 36 hours, which makes it a tool for the ten months beforehand rather than a dashboard for the day. Whatever you want to see in real time needs to come from somewhere else, and deciding where is — again — a February job.

Most stores do not lose money at peak because they failed to make a clever change. They lose it because a structural change they should have made in March was made in October instead, or because a discount configuration nobody rehearsed did something reasonable and unintended, or because the one page that needed editing at 9am on the Friday could only be edited by someone on annual leave.

None of that is solved in November. All of it is solved in the quarter you are about to start. If you have just come out of a peak and you still have an accurate list of what went wrong, that list is worth more than any amount of planning you will do in August — write it down this week, while it is still uncomfortable, and schedule the structural half of it before March.

• Shopify support and store management
• Peak season readiness for a Shopify Plus store
• Speed and Core Web Vitals