Preparing an Online Store for Black Friday: A Six-Week Plan

Imagine that on an ordinary day your store processes two hundred orders and runs flawlessly. But on the day of the sale, at 10:00, ten times more people arrive — and within the first half hour everything happens at once: the catalog takes eight seconds to open, the payment gateway returns errors, the warehouse shows yesterday’s stock, and a promo code works twice on one order. This is exactly where preparing an online store for Black Friday begins.

Preparing an online store for Black Friday is not a banner on the home page or a last-minute discount. On the day of a big sale the load rises sharply on everything at once: the website, catalog, payment, warehouse, CRM, promo codes and other integrations.

And that is exactly when the weak spots show up that can stay unnoticed on an ordinary day. The site starts to work more slowly, stock is updated with a delay, and an error at one step affects the entire customer journey.

That is why before Black Friday it is important to check how the whole order journey works. Not only site speed, but also product search, the cart, payment, stock updates and data transfer to the CRM or accounting system.

When is the best time to start preparing?

The closer to the sale, the less time remains for serious changes. If testing reveals a problem with speed, integrations or performance, it must not only be fixed but also checked again.

That is why it is best to start about six weeks ahead. This gives you time to go through the main risk points without rushing, fix the weak spots and make sure the changes really work.

Below is a week-by-week preparation plan: what to check in advance, what to leave closer to the start and what to watch for during the sale itself.

Six weeks out: calculating what we are preparing for

Preparation starts not with technical work but with numbers. Take the analytics of the previous sale and of an ordinary week and calculate how many times traffic, the number of orders and server load grew in the peak hour. If last year there were 1,500 simultaneous users at the peak, plan for at least a double margin.

Immediately fix three things: the target number of simultaneous users, the maximum number of orders per minute and the list of pages that will take the main hit. Usually these are the home page, two or three promotional categories, the promotion page and the cart. These are tested and optimized first, not the whole website at once.

Five weeks out: load testing

Load testing is the only way to learn the limit of the system before shoppers find it. You should test not the home page but the full scenario: viewing a category, applying a filter, opening a product page, adding to the cart, placing an order with payment in test mode.

It is useful to run three different tests. The first is a gradual increase in load to find the point where response time starts to grow. The second is a spike test, a sharp jump within a minute, because that is exactly how the audience behaves at 10:00 after a mailing. The third is a long test of two to three hours that reveals memory leaks and queue overflow.

The result of a test is not “the site held up” but concrete numbers: at how many users the catalog response time exceeds two seconds, where 5xx errors appear and which component fails first. Most often it is the database, search or an external integration, not the server itself.

Four weeks out: Core Web Vitals and speed

Under load a site is always slower than on a calm day, so the speed margin has to be created in advance. Google evaluates pages by three Core Web Vitals metrics: LCP — how fast the main content appears, INP — response to user actions, CLS — visual stability.

The biggest effect before a sale comes from simple things: compression and modern image formats, lazy loading of banners below the fold, reducing JavaScript and an honest review of third-party scripts. Before Black Friday a site often accumulates five or six advertising pixels and chats, and each of them adds delay exactly when every second costs money.

Check the promotional pages separately. They are often assembled in a rush with heavy banners and timers, which is why the most expensive page of the campaign turns out to be the slowest.

Three weeks out: scaling and a degradation plan

Next comes infrastructure. Scaling on the day of the sale must be predictable, not manual. Decide in advance which components can be added horizontally and which will remain a bottleneck.

The basic set looks like this:

• caching of the catalog and promotion pages with rules for invalidation after price changes;
• a CDN for images and static files;
• moving heavy operations into queues: calculating recommendations, sending emails, exporting to marketplaces;
• rate limiting for bots and parsers;
• a separate capacity reserve for the database and search.

Just as important is a degradation plan: what exactly the store switches off if the load exceeds the estimate. Personal recommendations, live chat, the product view counter and synchronization with external services can be temporarily disabled. The cart, payment and order confirmation — never. Agree on this before the start, not at two o’clock in the morning on Friday.

Two weeks out: CRM, payments and stock

At the peak it is not only the site that breaks. The most expensive failures happen at the junction of systems: the order is created but did not reach the CRM; the payment went through but the status did not update; a product is sold twice because stock is synchronized once an hour.

Check each integration separately: how the system behaves when an external service responds slowly or is unavailable. An order must be saved even when the CRM does not respond, and reprocessing must not create a duplicate. For payments, run test scenarios: successful payment, bank decline, pressing the button twice, refund.

Stock during a sale should be updated more often than usual, and the product should be reserved at the moment of checkout, not after payment. Otherwise the most popular item will be sold several times over, and instead of profit you will get a wave of returns. If the catalog and attributes are not yet in order, start with the structure of the product catalog: during a promotion, price and availability must come from a single source of truth.

One week out: UX, promo codes and final checks

When the technical part is ready, the shopper’s journey remains. On the day of the sale it must be short and clear: visible savings in hryvnias and percent, an honest timer, clear delivery and payment terms, a minimum of fields in checkout and guest checkout without mandatory registration.

Filters are prepared separately: a promotion usually adds a “discounted” attribute, and it instantly creates new address combinations. To avoid thousands of empty pages in search after the campaign, apply the rules from our article on faceted navigation.

Promo codes are a separate source of losses if they are not described by rules. Before the start, fix whether several codes can be combined, whether the discount applies to already marked-down products, whether it applies to delivery, how many times one customer can use a code and when it expires. Be sure to check that a discount cannot make the order total negative and that a code does not work twice when pressed again.

Day X: launch checklist

On the day of the sale the team works from a list, not from inspiration. The minimum set of checks and roles looks like this.

One more rule: on the day of the sale nothing is deployed to the live site except critical fixes. All changes must be ready and checked at least two days in advance.

Analytics: what to count before, during and after

Without configured analytics a sale turns into guesswork. Before the start, make sure all key e-commerce events work: product view, add to cart, begin checkout, payment, refund. Check them on a real order, not in preview mode.

During the campaign, look at conversion by hour, average order value, the share of orders with a promo code, drop-offs at the payment step and page speed under load. After the campaign, count something different: margin taking discounts and delivery into account, the return rate, the cost of acquiring a customer by channel and, most importantly, how many of the new customers came back during the following three months.

It is this last number that distinguishes a successful Black Friday from a sale that simply sold goods cheaper.

After the sale: keeping what you gained

The first week after the campaign is the time to tidy up. Remove promotional banners and timers, restore regular prices, close temporary promotion pages correctly: if a page is no longer needed, it must return a 404 or a 301 to the relevant category rather than stay empty in the index.

Then collect the technical conclusions while they are fresh: which component turned out to be the bottleneck, which integrations failed, what had to be switched off. This list becomes the basis of the development plan for the year — and at the same time the answer to the question of whether the platform has outgrown your business. We described the signs that it has in the article “When the online store outgrew the CMS”.

Frequently asked questions

How much capacity margin should I plan for?

Aim for double last year’s peak. If there is no history, take the maximum daily load and multiply it by at least five, then refine the figure with load testing.

Can I launch a new design before the sale?

Better not. Major changes should be completed at least a month before the campaign so that you have time to gather data on user behavior and fix errors. If a redesign is still planned, follow the plan for a redesign without losing SEO.

What should I do if the site goes down anyway?

Switch on a waiting page prepared in advance with an honest message, turn off secondary features and keep saving orders even if some services are unavailable. The worst scenario is a blank page and silence in communication.

Black Friday does not create problems — it shows the ones that already exist in the system. The GL.ua team can run load testing, optimize speed, check integrations with the CRM and payments and prepare an action plan for the campaign day. You can start with a technical audit or an SEO audit of your current store — more about our services on the online store development page.