Home › Blog › Digital Infrastructure
Digital Infrastructure

Black Friday Monitoring Singapore: The Six Numbers to Check Every Morning of Peak Week

Black Friday Monitoring Singapore: The Six Numbers to Check Every Morning of Peak Week

Check six numbers every morning of peak week: paid orders against placed orders, the age of your oldest unshipped order, stock variance on your top twenty products, customer messages older than four hours, parcels actually handed to the courier against parcels marked shipped, and failed payments plus refunds. Read them at the same time each day, give each number one named owner, and agree in advance what action each one triggers. Black Friday falls on 27 November 2026 and Cyber Monday on 30 November. Your systems are already built and frozen. From here the work is watching, and watching only pays if it is the same six numbers, at the same hour, by the same people.

Why watch numbers at all when the systems are already fixed?

Because peak week faults are quiet. A checkout that fails for one card type does not throw an error anybody sees. A stock sync that stalls at 2am keeps showing green. A courier pickup that collects forty parcels instead of sixty leaves twenty sitting in the corner looking exactly like the ones that went. Nothing alarms. The business only finds out two days later through complaints, by which time the refunds, the goodwill vouchers and the overtime have all been spent.

A daily number closes that gap. It turns a silent fault into a visible one within twenty-four hours instead of seventy-two. For a small merchant doing a few hundred peak-week orders, one day of early detection is the difference between fixing ten affected customers and apologising to eighty.

Which six numbers actually matter?

Not every metric deserves a daily read. These six each answer a different question that only matters during peak week.

  1. Paid orders against placed orders. The gap is your checkout failure rate. A gap that widens from one morning to the next means the payment path is struggling under load, not that demand fell.
  2. Age of your oldest unshipped paid order. One number, in hours. It tells you whether the warehouse is keeping up or quietly building a backlog that will blow through your delivery promise.
  3. Stock variance on your top twenty products. System quantity against a physical spot count. Twenty items, not the whole catalogue. This is how overselling gets caught on day one rather than at the refund stage.
  4. Customer messages older than four hours. Across every channel you actually answer, counted as one figure. During peak week a four-hour-old question is already a cancellation risk.
  5. Parcels handed to the courier against parcels marked shipped. These two should match daily. When they do not, you have shipment notifications going out for parcels still sitting on your floor, which is the single fastest way to lose a customer's trust.
  6. Failed payments and refunds issued. Both counts, side by side with yesterday. A jump in either is your earliest sign of a gateway problem or a fraud pattern working through your orders.

Who reads them, and when?

One named person per number, and one fifteen-minute call at a fixed time. Most Singapore retailers and online sellers find 9am works, after the overnight orders have landed and before the day's picking starts. Nobody presents. Each owner says their number and whether it moved the wrong way. That is the whole meeting.

The naming matters more than the tooling. A number nobody owns is a number nobody checked. If the same person owns four of the six because you are a team of eight, that is fine, as long as it is written down before 27 November and not decided on the morning.

What do you do when a number moves the wrong way?

Decide now, not then. Each number needs a trigger and a pre-agreed response, because judgement is the first thing to fail at 11pm on Cyber Monday.

Sensible starting points for a small business: if the checkout gap widens by more than a few percentage points overnight, someone places a real test order with a real card within the hour. If the oldest unshipped order passes twenty-four hours, you stop promising next-day delivery on the website that day. If stock variance appears on any top-twenty item, that item goes off sale until it is recounted, because selling air costs more than missing a sale. If messages older than four hours exceed about thirty, you pull someone off packing and put them on replies.

Write the trigger and the response on one page. A rule written in advance gets followed under pressure. A discussion started under pressure does not finish.

How do you set this up before 27 November?

You have five days and a change freeze, so build nothing new. Every one of these six numbers can be read from reports you already have: your store admin, your courier portal, your payment dashboard and a clipboard for the stock count. A shared spreadsheet with six rows and one column per day is a complete implementation. Resist the urge to automate it this week, because a dashboard built in a hurry during a freeze is itself a risk.

Spend the time instead on the two things that cannot be improvised: assigning each number to a person by name, and agreeing the trigger for each one. Then run it once as a rehearsal, ideally on 25 or 26 November, so the first real reading on the morning of 28 November is routine rather than novel.

What happens to this after peak week?

Keep the spreadsheet. Those six rows across two weeks are the clearest record you will ever have of where your operation bends under load, and they arrive just as you need to make January decisions on tools, headcount and renewals. A business that can point to three days of checkout failures or a persistent twenty-hour shipping backlog is making those decisions from evidence. One that cannot is making them from memory of a stressful fortnight, which is how the wrong problem gets funded.

This is the systems-over-headcount argument in its most practical form. The question in December is rarely whether you need another pair of hands. It is which specific step broke under volume, and whether a tool or a process change fixes it more cheaply than a salary.

Frequently asked questions

Is six numbers too few for a business doing high peak volume?

Six is the number a small team will genuinely read every day for two weeks. A list of twenty gets checked once on day one and abandoned by day three, which is worse than having no list at all. If you have a dedicated operations person with capacity, add delivery exceptions from your courier as a seventh. Otherwise hold the line at six.

Should we be watching these in real time instead of daily?

Only failed payments and the checkout gap justify a live alert, and only if your payment provider already offers it without new setup. Everything else is better read once a day. Real-time monitoring during peak week usually produces reaction rather than decision, and it pulls your most capable person away from the work that clears the backlog.

What if a number looks bad but we cannot find the cause?

Protect the customer first and investigate afterwards. Take the item off sale, pause the delivery promise, or switch off the affected payment method. These are reversible in minutes and cost you some sales. Trading on while you diagnose costs refunds, chargebacks and reviews that last well past December. Record what you saw and what you did, and keep the diagnosis for the first week of January.

Ready to Transform Your Business?

Let Digital Perpetual help you automate, streamline, and grow.

Get Started with Digital Perpetual →
black friday 2026 peak week order monitoring singapore sme operations ecommerce