Every failure a store can have costs more on Black Friday, for the obvious reason and one less obvious one: it is also the weekend you are least able to check. Traffic is at its annual peak, discount logic is at its most complicated, apps are being installed in a hurry, and an hour offline costs several times what an hour costs in March. This is what to do in the weeks before, what to watch during, and how the timezone maths works if you trade from New Zealand or Australia.
Almost every Black Friday incident traces to a change made close to the day. The single most effective thing you can do is finish changing the store early enough that anything broken has time to surface.
Anomaly detection compares against history, and Black Friday has no history that looks like Black Friday. Expect the baseline to be exceeded rather than met — that is fine, and it is the direction that does not hurt you.
What matters is the opposite: a period with zero orders during peak is louder and more meaningful on Black Friday than on any other day of the year, because the expected volume is so far from zero. Do not mute alerts because you expect a busy weekend. Busy is exactly when silence is diagnostic.
Everything else can wait until Monday.
The timezone arithmetic catches people out every year. New Zealand is on NZDT (UTC+13) in late November, which puts it 18 hours ahead of US Pacific. Black Friday begins in New Zealand while the United States is still on Thursday, and the US peak — the small hours of Friday and Friday evening Pacific time — lands on Saturday in New Zealand.
Two practical consequences. If you sell to the US from NZ or AU, your busiest hours are overnight local time, which is precisely when nobody is watching the store: automated alerting is not a convenience, it is the only coverage you have. And if you sell domestically, your sale starts a full day before most international coverage and competitor pricing appears, so schedule the start deliberately rather than copying a US timetable.
Set your alert timezone to the market you sell to, not the one you live in, and make sure critical alerts go to a phone rather than an inbox you will read in the morning.
In the week following, clear expired sale prices and compare-at values, restock or unpublish anything that sold out, review which apps you installed in a hurry and remove the ones you no longer need, and note every incident with its detection time. That last list is what tells you whether your monitoring was fast enough, and it is the only version of the question you will be able to answer honestly next November.
What Cassian™ actually does
Zero-order alerting compares each period against a weighted historical baseline — so during peak trading, an hour without orders stands out further from expectation than at any other time of year.
Every plan, including free. Hourly analysis on Guardian and above.
Uptime checks run every 5 minutes on Protector and Guardian and every minute on Prime, with a 30-second confirming retry before an incident opens.
Protector and above.
Price changes are diffed against the previous catalog snapshot with $0 and inverted sale prices treated as critical, and bulk changes of twenty or more variants grouped into one alert.
Every plan with a connected store; scheduled daily on Protector and above.
SSL certificate and domain registration expiry are checked daily and alerted as a countdown, so neither can surprise you during a peak weekend.
Protector and above.
Critical alerts can be routed to Telegram, Slack or Discord with a per-channel severity floor, so the channel that reaches your phone carries only what is worth interrupting a peak weekend for.
Protector and above. Email on every plan.
FAQ
Free scan, results in minutes. Zero-order alerts on every plan, including the free one.
We use cookies to keep you signed in and improve your experience. See our Cookie Policy for details.