The 2026 context makes this the sharpest number in the building: National Insurance thresholds and rate changes have raised the cost of every employed hour since 2025, the living-wage floor keeps climbing, and the venues absorbing it through price alone are watching spend per head fight back. The ones absorbing it through scheduling are the ones reading this page.
Getting the denominator and numerator honest
Numerator, loaded: wages plus employer NI, holiday pay accrual (12.07% accrual on worked hours for eligible irregular-hours and part-year workers under the applicable rules, with holiday pay calculated on the appropriate basis), pension employer contribution, meals-on-shift if you provide them, and agency premium when it happens. A £13.85 base hourly wage can cost more once the applicable employer costs are included. Calculate those from the actual worker and payroll assumptions rather than adding an unverified universal percentage before a single scheduling decision.
Denominator, net: ex-VAT, weekly, same definition forever. A percentage that divides by gross till receipts understates cost relative to net revenue; the difference depends on the labour ratio and is wrong in the flattering direction, which is the dangerous one.
With honest inputs: under about 28% you're lean (watch service quality), 28-33% is the normal healthy UK band, 33-38% means the rota is fighting the trade shape, above 38% sustained is a structural conversation about either pricing or hours. Trade-estimate bands, indicative; a food-heavy kitchen legitimately shifts the whole frame.
The decomposition that actually schedules people
Weekly, split the percentage three ways:
Rate (loaded £/hour): mostly fixed, moved by policy, not by you this week. Hours (total rostered): the lever. Coverage mismatch: the hours where you were staffed and empty, plus the hours where you were slammed and short. This is the one nobody computes and it's where the fat hides.
Compute mismatch from the daypart curve: hourly net revenue against hourly rostered cost, four weeks overlaid. Most venues find the same two things: a fully staffed dead shoulder before peak (one bar person for 90 minutes on four nights is six paid hours; multiply by the actual loaded hourly cost) and a short-staffed peak where round times stretch and second drinks die. The first is found money; the second is the expensive kind, because slower peak service may reduce drinks-per-head point, which is usually worth more than the shift you saved.
The practical shape: fewer flat rosters, more stepped ones. The trade is a curve that lives at 22:30; the rota is usually a rectangle centred on 19:00. Two shorter, offset shifts covering shoulders and closes beat one long one wherever the law and the team's preferences allow, and every hour you're not paying a third bar person to watch the first table arrive is margin with your name on it.
The quality line, said plainly
Labour percentage has a floor that isn't a spreadsheet: service speed, responsible-drinking vigilance, and the difference between a venue people return to and a venue that survives the lease. Scheduling to 25% through thin bars is how venues discover, via reviews and shrinkage, that the last three points of labour percentage were doing real work. The goal is the same output with fewer wasted hours, not fewer hours doing the output.
The weekly review workspace includes an hourly comparison block: enter net sales, paid hours and loaded cost to see where coverage and demand diverge. Use the pattern to ask better rota questions, then check service quality after any change. Better Bartender's venue pilot explores automated reporting inputs separately; an hourly overlay needs actual rota data as well as sales.
Convert elapsed time into paid hours
One employee working ninety minutes creates 1.5 paid hours; two employees doing that together create three paid hours. Include paid setup, closing and relevant breaks according to the actual payroll treatment. Do not add a holiday allowance twice if it is already included in a loaded hourly rate. Employer NI thresholds and pension eligibility affect the cost of individual workers, so a single multiplier is a planning simplification. Use actual payroll cost for historical reviews, then test realistic marginal staffing costs when changing the rota. Service quality, licensing requirements and safe working arrangements remain constraints on any apparent saving.
FAQ
What is a good labour cost percentage for a UK bar?
25-35% of net (ex-VAT) turnover is the normal working range for wet-led venues in 2026, with healthy independents around 28-33%; loaded cost means wages plus employer NI, pension and holiday accrual. Trade-estimate bands, indicative.
Why is my labour percentage suddenly high when I cut hours?
Because the ratio has two levers. Cutting hours into a soft trade week raises the percentage while reducing the £ problem it was meant to fix; check whether revenue fell faster than hours before cutting deeper.
Should I staff for forecast sales or actual footfall?
Both, at different times: the roster is built on forecast (last-year same-week, weather, events), and the shift flexes on actuals (send-home policies, cross-site moves). Venues that only do the first waste hours; venues that only do the second run on panic.
Does more staff mean more sales?
At the margin during peak, frequently yes: round time drives second-and-third drinks, and a bar person per how-many-covers is a real production function. Outside peak, no: which is the whole argument for stepped rotas instead of bigger ones.
Method and assumptions
Bands are illustrative working ranges, labelled indicative; NI and statutory-holiday references follow published UK rules as at October 2026 (check current rates when deploying). Hourly-overlay method per data standards. Worked figures illustrative.
Last reviewed 19 October 2026.