It's 3:15pm on a Tuesday and you've got two people stood behind the counter polishing the same patch of stainless steel for the third time. Nobody's come in for twenty minutes. Rewind to 9:05am and it was the opposite: a queue out the door, one barista on the machine, and a customer giving up and walking out.

Same day. Same rota. The labour was just in the wrong place.

Most café rotas aren't built around when customers actually turn up. They're built around what you've always done, and that quiet habit is costing you money at both ends of the day.

Why "what we always do" overstaffs the lulls and understaffs the rush

Walk into most independent cafés and the rota looks suspiciously tidy. Two people 8 till 4, one on 10 till 6, maybe a third for the lunch push. Round numbers, full shifts, the same shape every week.

It feels fair and it's easy to write. But footfall doesn't arrive in tidy blocks. It arrives in a curve, with one or two peaks and a long sag in the afternoon. When you staff a flat line against a curvy demand, you get exactly the Tuesday above: idle hands at 3pm, chaos at 9am.

The basics of building a rota and reading your labour cost are worth nailing first, and I've covered the basics of rota and labour costs separately. This post is the next step: shaping that rota to your real demand instead of guessing.

Wage percentage is the number, but a single figure hides the story

Wage percentage is your fully loaded wages divided by net-of-VAT sales. It's the single most useful ratio on a café P&L, and for most UK indies a healthy figure sits somewhere in the 28-35% of turnover range, depending on format. Coffee-led takeaway sites run leaner; brunch and full-service sites run higher because the tickets are bigger and the service is heavier. The café wage percentage benchmarks post has the full table by café type.

Here's the catch. A monthly wage percentage of 31% looks fine on paper. But that one number is an average of brilliant hours and dreadful ones blended together. The 9am rush where you're doing £90 of sales per labour hour is subsidising the 3pm lull where you're doing £18. The average looks healthy while the afternoon quietly bleeds.

Reading where your covers actually land

You already have the data. Every transaction your till takes is stamped with a time. Pull a few weeks of sales out of your POS and group it by hour and by day of the week, and the demand curve appears almost immediately.

You're looking for a few things:

The peaks. When do most of your covers land? For a lot of cafés it's a 9 to 11am coffee-and-pastry wave and a 12 to 2pm lunch push.
The dead zones. The classic is the 3 to 5pm sag, after lunch and before any after-school or commuter trade.
The day-of-week shape. Saturday at 10am is a different business from Tuesday at 10am. A rota that ignores that is leaving money on the table on both days.

This isn't a new system to install. It's reading what your till already knows. The sooner the shape of your week is in front of you instead of in your head, the sooner you can build to it.

Turning covers into the headcount you actually need

Knowing the curve is half of it. The other half is converting covers into people, and the way you do that is a rough labour standard: how many covers can one member of staff comfortably handle in an hour on your floor, with your layout and your menu?

Say you settle on roughly 20 covers per staff hour for counter service. If your data shows 60 covers landing between 9 and 10am, that's three people for that hour. If 3 to 4pm only brings 15 covers, that's one person, maybe with prep duties layered on top.

That's the whole trick. Instead of guessing "we always have two on in the afternoon," you're sizing each hour against what the hour actually does.

Building the rota to the curve

Once you've got the shape and a working standard, you stop writing tidy blocks and start writing to demand.

Stagger your starts

You don't need everyone in at 8. Bring your opener in to set up, then stagger the second and third arrivals to land just before the 9am wave, not an hour before it. Every hour of someone stood ready before trade exists is pure cost.

Boost the midday peak, trim the afternoon

Pile your bodies into the peaks and pull them out of the sag. If the data says 3 to 5pm is dead, that's where one person covers the floor and gets the prep done, not where you've got two idling. There's a separate question worth asking too, which is whether you can lift that trough at all by filling the quiet afternoon trade rather than just staffing it leaner. Most cafés find they can take 10-15% of labour cost out of the week just by reshaping shifts like this, without losing a single minute of peak coverage.

Prep ahead so peak labour goes on service

The worst use of a 9am barista is having them slicing cake or filling the fridge while a queue builds. Push setup and prep into the genuinely quiet hours, so that when the rush hits, every paid minute is going on serving customers, not getting ready to.

Don't strip the rush to the bone

There's a wrong way to do all this, and it's to read "cut labour" and start hacking at every shift until the wage percentage looks lovely on the spreadsheet.

Cut the peaks too hard and queues build, regulars drift off, and the team you've got left burns out covering the gaps. Sales per labour hour goes up on paper while total takings quietly fall, and your wage percentage ends up no better or worse. Protecting the rush is the point. The labour you're trimming is the idle afternoon, not the moment customers are actually trying to give you money.

The aim isn't a skeleton crew. It's the right crew at the right hour.

Every hour costs more now, so this matters more

This used to be a nice-to-have. After April it's urgent. The National Minimum Wage rose to £12.71 an hour for 21-and-overs, and with employer NI on top a barista on the headline rate actually costs you closer to £15.50 to £16 fully loaded.

That means every hour of misplaced labour is dearer than it was last spring. The 3pm idle hour that cost you £24 in March costs you noticeably more now. If you want to test whether a shoulder hour pays at all, profit per trading hour after wages sets the gross profit that hour earns against the wages of whoever's on. I've gone through the true cost of the April NMW rise line by line, and if you're juggling a pay rise for a long-serving member of staff against a tight rota, the pay rise planner is worth a few minutes. The cost cliff is real, and a demand-shaped rota is one of the few levers that takes cost out without taking pay off anyone.

It also pairs naturally with knowing your break-even covers per day: once you know how many covers you need just to cover the day, you can see at a glance which hours are pulling their weight and which are passengers.

Review against actuals, every week

A demand-shaped rota isn't set once and forgotten. Trade shifts with the seasons, with the weather, with a new café opening down the road. The rota you built in May won't be right in November.

So every week, sit the rota you ran next to what actually happened. Where were people idle? Where was service stretched thin? Nudge the standard, move a start time, adjust the peak boost. It's ten minutes and it keeps the rota honest.

This weekly loop is exactly the discipline StaffBrik is built around: it tracks your fully loaded labour cost against your sales so you can see the wage percentage for the week just gone, not six weeks later on the monthly P&L. We're building it toward demand-aware rotas that read your till patterns for you, so the curve does the heavy lifting instead of a spreadsheet you maintain by hand.

The takeaway

Most café rotas are built on habit, which overstaffs the quiet hours and understaffs the rush.
Wage percentage is the number to watch (roughly 28-35% of turnover for most indies), but a single figure hides where the waste actually is.
Pull a few weeks of sales from your till and group by hour and day to see where covers really land.
Set a rough labour standard (covers per staff hour) and size each hour against it, instead of guessing.
Build to the curve: stagger starts, boost the peaks, trim the 3 to 5pm lull, and prep ahead so peak labour goes on service.
Protect the rush. Trim the idle hours, never the moment customers are trying to spend.
After the April NMW rise every misplaced hour costs more, so this is urgent, not optional.
Give it a couple of weeks to get your standard right, and review the rota against actuals weekly.

Ed O'Brien has run Hunters Cake Company for 17 years across cafés in Witney, Burford, and a bakery in Carterton, Oxfordshire. He's building Brikly - modular tools that give independent café owners the same data the big chains have, without the big chain price tag.