Black Friday 2026
The Black Friday monitoring checklist
Twelve things to have in place before peak weekend, written so they hold whatever you monitor with. Each one says what to do, then how to do it in Checkly.
The prompt is the whole checklist, written for Claude Code, Cursor, or Codex. It reads your real plan limits before it writes anything and stops at the deploy preview for your sign-off.
The Black Friday window runs November 23 to 30, 2026.
Some of the teams that monitor with Checkly
“Before, we sometimes noticed issues too late, meaning that we were made aware through customers. As a customer-first business, this was frustrating and far from ideal. The number of unknown issues after deployments is now close to zero.”
Tobias DeekensPrincipal Engineer, Frontend Architecture
Twelve things to do before Black Friday
None of this takes a quarter. Most of it takes an afternoon, and it is the difference between hearing about an outage from your monitoring and hearing about it from a customer.
Check the whole buy, not one URL
Browse, search, product page, cart, checkout, confirmation, in a real browser. A 200 on the homepage says nothing about whether anyone can pay.
app.checklyhq.comRecovered · Checkout flow passing in all locationsNameTypeLast resultsAVAAVGP95Browse and search1 minute agoACME StorestorefrontPlaywright check100 %2.1s3.4sAdd to cart1 minute agoACME StorecartapiAPI check99.9 %180ms240msCheckout flow1 minute agoACME StorerevenuecriticalPlaywright check100 %3.1s4.8sStore URL Monitor2 minutes agoSSL 28 daysstorefrontURL monitor100 %245ms347msOrder confirmation2 minutes agoACME StorerevenuePlaywright check100 %2.2s3.0sshop.example/checkoutCheckoutTrail Runner 2$129.00Shipping$8.00Total$137.00Card•••• •••• •••• 424212 / 28Pay nowSomething went wrong. Please try again.In Checkly- Playwright Check Suites or browser checks, down to a one minute frequency
- A synthetic shopper with a test SKU and a test payment method, so test orders can be filtered out of your real ones
Run from where you sell, at peak frequency, and budget for it
A checkout that works from one region and fails from another is a failure you want to find first. Peak frequency across more locations costs more, so plan the spend.
Checkout flowevery 5 min

13 passing1 failing · ap-southeast-2
✗ expect(order-confirmation).toBeVisible()
Same check, every five minutes, from multiple regions. Most runs pass everywhere. The one that does not is Sydney hanging on a confirmation that never renders — and you would never see it from a single location. In Checkly- Parallel scheduling to run the same check from every location at once
- Add temporary campaign checks under their own tag or group, then remove them when the window closes
Plan: All 21 public locations are available on Team and above.
Monitor the APIs behind the storefront with response time limits
Cart, inventory, pricing, search, payment. Under load these rarely fail outright. They get slow, and the storefront times out waiting for them.
Cart APIresponse time · last 24 runsPassingdegraded 3 sfailed 10 s12:00peak traffic arrivesnowCart API failedus-east-1 · eu-west-1Response time 380 ms over the 10 s limit. Status was 200 on every run.
Every run here returns 200. Without a response time limit the check passes all the way through the slide, and the first thing you hear about it is an abandoned cart. In Checkly- Separate degraded and failed response time thresholds on every check monitoring your APIs, not just a status assertion
Load and monitor your third parties
Payment buttons, CDN, consent banners, tag managers. Peak weekend puts them under the same load it puts you under, and when one of them gives way it takes checkout with it.
Checkout flownetwork · eu-west-1RUNNINGGET /checkoutshop.exampleapp.csscdn.shop.examplecheckout.jscdn.shop.examplev3/pay.js3rd partypay.provider.examplebanner.js3rd partyconsent.vendor.exampleinter.woff23rd partyfonts.vendor.examplegtm.js3rd partytags.vendor.examplePOST /api/cartshop.example0s2.5s5s7.5s10sAlert sent · tags.vendor.example never answered. The run timed out at 10s with the page still waiting on it.In Checkly- Do not block third-party requests in your checks, even though it makes them faster
- Read the network waterfall, console errors, and traces from the failed run to see which one went down
Assert on promo content, not just status codes
The campaign banner is live, the prices are the sale prices, the discount code applies. And the failure copy, "out of stock" or an error state, never appears where it should not.
shop.example/checkoutBlack Friday · 30% off everything with code BF30 or CYBER30ACME/CheckoutDiscount codeBF30ApplyCode BF30 isn’t valid for these items.Subtotal$629.00Discount—ShippingFreeTotal$629.00Assertion failed · every request returned 200 and the banner still promises 30% off with CYBER30. The total came back $629.00, not $440.30.In Checkly- Keyword assertions on the rendered page, positive and negative
- A check that applies the campaign discount code and asserts the total that comes back
Cut false alarms before they page someone
Peak weekend is when a single transient timeout turns into a 3 a.m. phone call about nothing. Decide now what counts as real.
RetriesSelect the retry strategy to determine how failed checks are retried.
NoneSingleFixedLinearExponentialMax retries3Base backoff (exponential)5secondsThe retry intervals for your check(s) will be 5 seconds, 25 seconds, and 2 minutes 5 seconds.
Max total retry duration600secondsAlways retry from the same location the check failed in.Retries after a failed runlast retry at 2 minutes 35 secondsrun fails60s120s180sThe strategy is the decision. None pages somebody on the first failed run; exponential gives a flaky minute two and a half minutes to sort itself out before anyone’s phone goes off, and still alerts if it does not. In Checkly- Retry a failed run from a different location before escalating
- Run-based or time-based escalation, plus a location threshold so one flaky region does not page anyone
- Playwright Check Suites use Playwright retries rather than Checkly retries
Make sure your bot protection lets your monitoring through
Before peak, teams tighten bot rules and rate limits. Checks that get blocked or challenged turn into false alarms at the worst possible moment.
GET /checkoutacme.shop · bot protectionUnidentified automationchallengedpython-requests/2.31.0 · 203.0.113.47
Verifying you are human…Bot protection403no allowlist rule matched · the checkout never renders
Checkly checkallowedCheckly, https://www.checklyhq.com · 54.151.x.x
acme.shop30% offBlack Friday · 30% offCart · 2 items · $440.30200 OKallowlisted by user agent and static IP · no interstitial to clear
One rule, two outcomes. A check that is not allowlisted gets the challenge, has nothing to solve it with, and and fails on a challenge rather than on anything wrong with the site — an outage that never happened. Re-test after every rule change: the allowlist is the only thing separating these two lanes. In Checkly- Allowlist by IP, by user agent, or by verified bot status, and test again after every rule change
- Static IP list, available per region; the Checkly user agent for browser checks; a custom header for API checks; Cloudflare verified bot status
Watch what expires or drifts: SSL and DNS
A certificate that expires on the Saturday, or a record that changed during the pre-peak infrastructure work, takes the whole storefront with it.
Connect with TLS
Use SSL monitors to verify the TLS handshake of a hostname.
Hostname *: 443Monitor DNS resolution
Verify that a domain resolves correctly and returns the expected DNS records.
Domain *ADNS resolverPortProtocolOne monitor per record the storefront depends on. A record that changed during the pre-peak infrastructure work fails here, not at the checkout. In Checkly- An SSL monitor for expiry, chain, and hostname
- SSL expiry alerts on your existing API and browser checks
- A DNS monitor on the records the storefront depends on
Cover background jobs with heartbeats
Order export, inventory sync, product feeds. Nothing on the storefront looks wrong when these stop, right up until the warehouse has no orders.
app.checklyhq.comOrder exportat Nov 27, 06:00ACME StoreorderscriticalHeartbeat monitor15minInventory sync2 minutes agoACME StorestockHeartbeat monitor5minCart recovery11 minutes agoACME StorerevenueHeartbeat monitor30minProduct feed24 minutes agoACME StorecatalogHeartbeat monitor1hPayment sync41 minutes agoACME StorefinanceHeartbeat monitor1hFraud screening6 minutes agoACME StoreriskHeartbeat monitor10minThe storefront is green throughout. The order export stopped pinging half an hour ago, and the only thing that noticed was the heartbeat. In Checkly- A heartbeat monitor per job, with the job pinging Checkly when it finishes
Plan: Available on all plans.
Route alerts to whoever is on call over the weekend, and test it
The rota for peak weekend is not the rota from October. Send a real alert through the real channel and confirm a human receives it.
mobileCheckly Alerts+1 814 250 8623incoming callSave it as a contact and it rings through Do Not DisturbEscalation · Checkout flowRun-based · alert after 2 failed runsReminders · every 5 minutes until it recoversAlert notification log02:51Slack#peak-opsDelivered02:51SMSon-callDelivered02:52Phone callon-callDelivered02:57Phone callreminder 1 of 3DeliveredIn Checkly- Slack, PagerDuty, Opsgenie, and incident.io, plus SMS and phone call channels
- Reminders that keep firing until the alert resolves
- Rocky AI adds a root cause and a user impact summary to the alert, so whoever wakes up starts with context
Plan: incident.io and phone calls are Team and above. SMS is capped at 100 a month on Starter and 200 on Team.
Get the status page ready
When checkout is down, support gets buried. A status page that updates itself absorbs most of that, as long as it is wired up before the weekend.
ACME StatusContact supportSubscribeStatusActivityAll systems operationalCheckout99.99%uptime90 days agoTodayPayments100%uptime90 days agoTodayComponents a customer recognises, not check names. Without tag-based automation rules this page keeps saying all systems operational while checkout is down. In Checkly- Components for checkout, payments, and shipping
- Tag-based automation rules, or the page will not change when a check fails
- Subscriber notifications, so customers do not have to keep reloading it
Plan: Automation rules are Team and above.
Review your monitoring as code before the freeze
Monitoring deploys separately from the app, so a code freeze on the app does not freeze your checks. Know what is deployed before the window opens.
OpenAdd Black Friday monitoring4 files changed+160−3peak/black-friday-monitoring
__checks__
__checks__/black-friday.check-group.tsnew file+48−0
import { CheckGroupV2, Frequency, RetryStrategyBuilder,} from 'checkly/constructs'import { peakOps, onCall } from './alert-channels'// One group for the whole campaign, so it can be muted,// retagged or deleted in a single move when it is over.export const blackFriday = new CheckGroupV2('black-friday', { name: 'Black Friday 2026', frequency: Frequency.EVERY_1M, locations: [ 'us-east-1', 'us-west-1', 'eu-central-1', 'ap-southeast-2', ], tags: ['black-friday', 'revenue-path'], alertChannels: [peakOps, onCall], // One transient timeout is not an incident. Retry from a // different region before anyone's phone goes off. retryStrategy: RetryStrategyBuilder.fixedStrategy({ maxRetries: 2, baseBackoffSeconds: 30, sameRegion: false, }), browserChecks: { testMatch: '*.spec.ts' },})checkly deploy --preview
Create 1 check groupCreate 1 checkCreate 1 alert channelUpdate 1 alert channelDelete 0 checksMonitoring deploys separately from the app, so a freeze on the app does not freeze your checks. Open any file to see what the branch actually changes — the plan underneath says what deploying it would do, before anything does. In Checkly- The CLI v10 deploy plan lists creates, updates, and deletes, and flags checks that were edited in the web app
- Run `checkly test` against staging or a preview deployment
- Maintenance windows pause checks or silence alerts by tag for planned work
Plan: Maintenance windows are Team and above.
After the weekend
Peak weekend is the best performance data you will get all year. Pull it before it ages out.
- Dashboards and reporting give you average, P95, and P99, with CSV export
- On Hobby and Starter, raw results expire after 7 days, so export before you lose them
Walk into Black Friday already covered
Write your first check today and you will know about the next checkout failure before a customer tells you. Talk to sales to learn about this year’s Black Friday offers.

