Black Friday 2026 lands on Friday 27 November, with Cyber Monday on the 30th. If you are reading this in October, your real deadline is your code freeze, which for most stores is one to two weeks before the event.
This is the checklist we run with e-commerce clients every autumn. It is ordered by what tends to cost the most revenue when it fails.
1. Measure Core Web Vitals with promotions switched on
Your store probably passes Core Web Vitals in September. The real question is whether it still passes with the countdown banner, the spin-to-win popup, three new tracking pixels and a 4 MB hero video that marketing has queued up for launch day.
Google's thresholds are assessed at the 75th percentile of real users, on mobile:
| Metric | Good | What breaks it on promo days |
|---|---|---|
| LCP (Largest Contentful Paint) | ≤ 2.5 s | Unoptimised campaign hero images, render-blocking promo scripts |
| INP (Interaction to Next Paint) | ≤ 200 ms | Tag managers, popups, heavy filtering on collection pages |
| CLS (Cumulative Layout Shift) | ≤ 0.1 | Banners injected above the fold after load |
Build the promo state on a staging theme or preview branch now and test it in PageSpeed Insights and a throttled mobile profile. Our INP guide covers the interaction fixes in detail, and web.dev's Core Web Vitals overview explains the measurement.
Quick wins that consistently pay off:
- Reserve space for every banner and badge, so injected content cannot shift the layout
- Serve campaign images in AVIF or WebP at the exact rendered size, with
fetchpriority="high"on the hero only - Load chat widgets, reviews and popups after interaction or idle, never in the
<head> - Audit the tag manager: every pixel you do not need on 27 November is free INP headroom
2. Load test the dynamic path, not the homepage
CDN-cached pages almost never fall over. What falls over is everything the CDN cannot cache: cart, checkout, inventory, search, account and the APIs behind them.
Script realistic journeys with k6 or a similar tool:
- Land on a collection → filter → product page
- Add to cart → apply a discount code
- Start checkout → shipping rates → payment (against a sandbox)
Target 1.5–2× your projected peak for the normal curve and 2–3× for flash spikes such as the 9 a.m. email drop. Watch p95 latency and error rate per step, not the averages.
On hosted platforms like Shopify, the core checkout scales for you. Your risk sits in apps, custom storefront APIs, middleware and third-party integrations. Load test those directly, and read their rate limits before your customers find them.
3. Know your failure modes before they happen
The usual suspects, in our experience:
- Payment provider rate limits or 3-D Secure timeouts under load
- Inventory contention: overselling or locked rows on a hero SKU
- Coupon and discount services that recalculate the whole cart on every keystroke
- Search and filtering services hitting query quotas
- ERP, OMS or warehouse sync jobs that cannot keep up with order volume
- Email and SMS providers throttling transactional messages
For each one, write down the symptom, the dashboard that shows it, who owns it and the fallback. If you are on a custom or headless build, add your own API routes and caching layer to the list.
4. Build kill switches, then rehearse them
Every promotion and non-essential feature should be switchable without a deploy:
- Popups, spin-to-win, product recommendations, live chat
- Expensive personalisation or AI search
- Non-critical third-party scripts
Use feature flags or theme settings, and rehearse them on staging. A kill switch nobody has tested is a hope, not a plan.
5. Freeze code, not fixes
A sensible calendar for 2026:
| Date | Milestone |
|---|---|
| By 30 Oct | Promo builds on staging, load-test round one |
| By 9 Nov | Fixes shipped, load-test round two, runbooks written |
| 13 Nov | Code freeze for risky changes |
| 23–30 Nov | War room hours staffed, hotfix path only |
Keep a hotfix lane with mandatory review and a one-click rollback. Our maintenance team runs exactly this lane for clients during peak.
6. Plan for AI shopping agents, too
A growing share of product discovery now comes through AI assistants and agentic checkout flows. Make sure your product feed, structured data and inventory accuracy hold up under the same load. It is cheap insurance, and it is the foundation for agentic commerce.
7. Staff the war room
During the peak window, have one person who can make decisions and one who watches the dashboards:
- Real-time conversion rate, checkout error rate and payment success rate
- Synthetic checks on the golden path every minute
- A shared channel with your platform, payment and app vendors' escalation contacts
The short version
Black Friday failures rarely come from the parts you built this month. They come from the parts nobody load tested. Measure with promos on, test the dynamic path at 2× peak, write the kill switches, then freeze.
Six weeks is enough time if you start this week. Book a readiness review and our e-commerce engineers will run the audit and load test in a single sprint.
Frequently asked questions
When is Black Friday 2026?
Black Friday 2026 is Friday 27 November, and Cyber Monday is 30 November. Most teams freeze code one to two weeks earlier, so your effective deadline is mid-November.
How much load should we test for Black Friday?
Test at 1.5–2× your projected peak for normal traffic patterns and 2–3× for flash-sale spikes such as email drops or influencer posts. Base the projection on last year's peak sessions per minute plus your growth rate.
What usually breaks first on Black Friday?
Checkout and payment calls, inventory reservation under contention, discount and coupon services, search, and third-party scripts. The storefront CDN rarely fails; the dynamic dependencies behind it do.
Should we deploy during Black Friday week?
Only behind feature flags and with a tested rollback. Freeze risky changes, keep a hotfix path open and make sure every promotion can be switched off without a deploy.