
If your store crashes during a Black Friday sale, spend the first five minutes confirming what is broken and from where, and the next ten reducing load so the server can recover. Leave the root cause for later. Serve a short maintenance page with HTTP status 503 instead of a broken page or a "200 OK" error page. Pause the paid traffic and email sends that are pushing visitors at a store that cannot serve them. Then protect the orders you already took: payments that went through while your site was failing still need to become orders.
Black Friday falls on 27 November 2026 and Cyber Monday on 30 November. That leaves ten days to write this runbook down, give each step an owner and rehearse it once. A crash at 09:00 on sale day is a bad time to find out that nobody has the hosting login or the ad account password.
Peak days pack traffic and discount logic into a few hours, and many stores ship last-minute changes the same week. A stress test tells you where the ceiling is (see our six-week stress test plan), but on the day you control only the response. These platform behaviours, as of November 2026, shape it:
Retry-After header that Googlebot may use to decide when to recrawl. The same guidance warns about sites that return 200 for downtime pages.Retry-After header if you can set one. Say what is happening, that the sale continues, and when you will update. Do not put a discount code in it; visitors will screenshot and share it.curl -I https://yourstore.example/. Maintenance plugins and platform features vary. WooCommerce's built-in "Coming soon" mode hides the store behind a landing page, and its documentation does not say which status code that page returns, so test it before you depend on it./wc-api/, /?wc-api= or your custom webhook path, payment providers will keep failing and retrying.| Risk | Owner | Check | Frequency on sale day |
|---|---|---|---|
| Store unreachable for some visitors | Developer or agency on call | External checks from at least two regions | Every minute, automated |
| Site up but orders stopped | Store manager | Orders per hour against the same hour last week or last sale | Hourly, or alert-based |
| Paid ads sending traffic to a broken page | Marketing lead | Landing pages of active campaigns load and add to cart | Every two hours, and at once during an incident |
| Payment taken, no order created | Finance or store manager | Payments in provider dashboard against orders in the store | After every incident, then end of day |
| Webhook endpoint blocked by maintenance rule | Developer | Provider dashboard shows deliveries as succeeded | After every maintenance-mode switch |
| Outage page returns 200 | Developer | curl -I on home and product URLs shows 503 |
Each time the page is enabled |
| Customers not told what is happening | Incident lead | Status page, social post and support macro updated | Every 30 minutes while the incident lasts |
Print this, fill in the blanks this week and pin it where the on-call person can see it.
BLACK FRIDAY CRASH RUNBOOK (store: ____________)
Incident lead: __________ Backup: __________
Host support contact and account ID: __________
CDN account owner: __________ Waiting room configured? Y / N
Ad accounts to pause (names, who has access): __________
Email/SMS tool to hold scheduled sends (who has access): __________
[ ] 0-5 min Confirm from mobile data + second region. Note start time.
[ ] 0-5 min Check status pages: platform / host / CDN / payments.
[ ] 5-15 min Pause ads and scheduled sends.
[ ] 5-15 min Switch off heavy features: ______, ______, ______
[ ] 5-15 min Lower waiting room threshold (if available).
[ ] 15-30 min Maintenance page ON. Verify: curl -I shows 503.
[ ] 15-30 min Webhook paths excluded from maintenance: ______
[ ] 15-30 min Customer notice posted (social, status page, support macro).
[ ] 30-60 min List payments without orders since start time.
[ ] Recovery Storefront -> checkout -> marketing traffic, 15 min watch each.
[ ] After Reconcile outage window. Resend failed webhooks.
[ ] After Write a short incident note: start, end, cause, orders affected.
The numbers below are invented to show the procedure, not taken from a real store.
A WooCommerce store selling outdoor gear sends its main Black Friday email at 08:00. By 08:06 product pages take 20 seconds to load and checkout returns 502 errors. The agency's external monitor confirms failures from two regions, and the host's status page is clear, so the problem is on the store's side.
At 08:09 the store owner pauses two shopping campaigns and holds the 10:00 reminder email. The developer turns off the AJAX product filter and the related-products widget, which run uncached database queries on every page view. At 08:15 load is falling but checkout still times out, so the developer enables a maintenance page for store pages and checks it: the first plugin they try returns 200, so they switch to a server rule that returns 503 with Retry-After: 1800 and leaves the PayPal and Stripe webhook paths open.
At 08:40 the developer brings the storefront back with the filter still off, then checkout at 08:55. Ads resume at 09:15. When the finance lead compares the payment dashboards with WooCommerce orders for 08:00 to 08:55, they find 14 Stripe payments and 3 PayPal payments. Twelve Stripe orders arrived through automatic webhook retries once the site recovered; the developer resends the other two events from the Stripe Dashboard. The three PayPal payments arrive through IPN retries within the hour. The team refunds nothing by mistake and does not have to ask customers for proof of payment.
There are limits to what a runbook can do. If the database is corrupted or the host is down at the data-centre level, the first hour is spent waiting for someone else, and your job narrows to communication and reconciliation. A waiting room reduces load but does not fix a slow query. Webhook retries recover orders only if your integration accepts events that arrive hours late. Test that with a real resend before the sale.
Yes, if the store cannot complete orders. A maintenance page that returns 503 is better for customers and search engines than a page that half-loads and fails at checkout. Keep payment webhook paths open and test the status code with curl -I.
Google's documentation says 5xx errors slow crawling temporarily and that URLs are dropped only if server errors persist. It gives no exact threshold. A short outage served as 503 is the case Google's downtime guidance describes; a week of errors is a different risk.
Find the payment in your provider's dashboard and check the webhook or IPN delivery log. If delivery failed, resend the event (Stripe allows Dashboard resends for up to 15 days) or wait for the automatic retry. Create the order by hand only if the retry fails, and record the payment ID on the order so it is not created twice. Our payment gateway backup plan covers the case where the provider itself is down.
Last reviewed: 17 November 2026
Two rows in the table above can run without a person watching. Pingvera checks the store every minute from several regions and confirms a failure across locations before alerting, which covers step 2. For WooCommerce stores, its order anomaly detector compares order flow with an hourly baseline and alerts when the site is up but orders have stopped, the failure the orders-stopped guide describes. Deploys and incidents land on one timeline, so after the sale you can see which change came before the crash.
Pingvera watches whether an online business actually works — uptime, checkout, orders, domain, SSL and server — and alerts you in Telegram, email or a webhook before a customer has to tell you.
Start freeRead next: Website Stress Test Before Black Friday: A Six-Week Plan for Online Stores · Payment Gateway Down? A Backup Plan for Online Stores Before Black Friday.
Read next: Black Friday Email Deliverability: 2026 Store Checklist · Website Stress Test Before Black Friday: Store Plan · WooCommerce 11.3: What to Test Before Black Friday · Merchant Center Availability Mismatch: Black Friday Fix · Run a free site check.