Rostering from the sales curve, not from habit

Labour is usually the second largest controllable cost in a shop, and the rota is usually built by copying last week's.

Start with the hourly sales curve

Pull sales by hour, by day of week, for the last eight weeks. You will see a shape — and it is almost never flat.

Typical Saudi retail: quiet mornings, a modest lunchtime, a build from late afternoon, and a strong evening peak that runs late. Thursday and the weekend are different again. A restaurant's curve is even sharper around Maghrib.

Then ask: does the rota match the curve?

In most shops the answer is no, in both directions. There are three people at 11am serving four customers, and two people at 8pm serving thirty.

Build the rota from the curve

1. Work out transactions per staff hour at your busiest and quietest periods
2. Set a target — a number of transactions per staff hour that is comfortable, from observation not theory
3. Staff each hour to that target, rounding up at peak
4. Use short shifts to cover the peak. A four-hour evening shift is cheaper than extending a full shift and it fits the shape

Shops that do this for the first time typically find 5–10% of labour cost moving from quiet hours to busy ones — with no reduction in total hours, and a visible improvement in service at peak.

Attendance that is actually recorded

A rota is a plan. Attendance is what happened, and the gap between them is where labour cost leaks.

You want:

• Clock in and out on the POS or a dedicated device, tied to the individual's account
• Breaks recorded
• Automatic clock-out after a threshold, so a forgotten clock-out does not record an eighteen-hour shift
• A variance report: scheduled hours against actual hours, per person, per week

That last report is the one that matters. A person consistently clocking in ten minutes late and out twenty minutes early is two and a half hours a week, which is a conversation, not a crisis — but you cannot have the conversation without the data.

Linking labour to sales

The useful number islabour as a percentage of revenue, by hour, not for the month.

A month at 18% can contain hours at 40% and hours at 8%. The monthly figure hides exactly the problem you would act on.

Look at the worst hours. Usually they are the quiet ones where the rota did not shrink, and the fix is a shorter opening or a smaller crew rather than anything dramatic.

Saudi specifics

Friday and Saturday are the weekend, and trade patterns shift accordingly — often with a distinct post-Friday-prayer pattern.

Ramadan is a different business. Different hours, a concentrated iftar peak, and late-night trade. Build a separate Ramadan rota from last Ramadan's data, not from an ordinary month. SeeRamadan operations.

Prayer times affect both trading and breaks and should be in the rota rather than handled ad hoc.

Labour regulations govern hours, overtime and leave. This article is about the operational side; confirm your obligations with a qualified adviser.

What the system should give you

• A rota built and published to staff
• Clock in/out tied to individual accounts
• Auto clock-out on a threshold
• Scheduled versus actual hours per person
• Labour cost against revenue, by hour and by day
• Sales by hour and day of week for building the rota in the first place

Wameed includes shift scheduling, attendance and automatic clock-out; per our August 2026 audit, labour management is present in Wameed, Oracle, Geidea and Odoo, and absent in Foodics, NCR, Rewaa, Marn, Revel, Sapaad and Hala.

See thestaff features.