A shaken cocktail takes about 90 seconds of a bartender's hands. A draft cocktail takes ten, and it tastes the same on the three hundredth pour as it did on the first. With 77% of operators reporting labor cost increases in the first half of 2026, taps deserve a serious look. Not for the whole list. For the right part of it.
Who is doing it
Draft cocktails used to signal a compromise. In 2026 they signal a bar that knows its numbers. Hello Hello in Chelsea, opened by alumni of Seamstress and Bathtub Gin, runs house builds on draft with nothing on the menu over $18. In London, All My Gods built its whole concept on tapped cocktails at £10, with an £8 house martini poured from a machine. These are cocktail-first operators choosing taps because the volume math works, not because the craft doesn't matter to them.
The labor math
Average hourly earnings in leisure and hospitality are $23.49 per BLS. Loaded with taxes and benefits, call it $28 an hour, or about $0.47 a minute.
- Shaken or stirred à la minute: ~90 seconds of hands-on build = ~$0.70 of labor per drink
- Draft pour into a chilled glass, garnish: ~10 seconds = ~$0.08 per drink
- Difference: ~$0.62 a drink. At 150 draft-eligible drinks on a Friday night, that is ~$93 and roughly 3.3 hours of rail time back, at the most expensive moment of the week
The point is not to cut a bartender. It is capacity: the crew you already have puts out more drinks with less strain, and ticket times stop stretching at ten o'clock. The same argument as batching your stirred list, taken one step further.
There is a second saving nobody budgets for: variance. A tap cannot overpour, and it cannot shake a short one. On a high-volume house drink, that alone protects a point or two of pour cost.
What belongs in a keg
The rule is shelf stability. If every component holds for weeks refrigerated, the drink can live in a keg.
- Spirit-forward builds. Negroni, Boulevardier, house martini. Batch at serving strength: include the 20-22% water that stirring would have added, keep the keg at 0-2°C, pour into a chilled glass. No ice in the glass for an up serve, ever.
- Carbonated highballs. Spirit, cordial, water force-carbonated at 6-8 g CO2/L in the keg. Serve over clear ice or ice-less in a chilled glass.
- Acid-adjusted builds. Citrus-style drinks work on draft only if the acid comes from a stable cordial (citric, malic, ascorbic) instead of fresh juice. Same technique as our gimlet cordial.
Filter everything through a coffee filter before it goes in the keg. Particles are nucleation sites: in a carbonated line they foam, in a still line they haze. A draft cocktail must pour crystal clear.
What never goes in a keg
- Fresh citrus. It oxidizes in days and drops sediment in the line
- Dairy, cream, egg. Texture drinks need the shake, and the cleaning risk is not worth it
- Anything you sell four of a week. Kegs are for volume; a slow line is wasted cooler space and a stale drink
Two to four taps is the right size. It covers the house drink, one carbonated build, and a seasonal, and it leaves the rest of the list where it belongs: on the rail.
The equipment math
A basic multi-tap draft system starts around $2,500, plus $1,500-5,000 for refrigeration depending on capacity. If an existing beer system has a spare line, conversion is cheaper still: cocktail kegs run on the same couplers, though spirit-based liquid wants a dedicated, regularly cleaned line. Against $90-plus a night of recovered peak labor on a busy program, the hardware pays for itself inside a quarter. The honest caveat: that return only exists if the taps pour drinks people actually order in volume.
The draft test, before you buy hardware
Pull the sales mix; find the top sellers that are spirit-forward or carbonated
Batch one of them as a fridge bottle for two weeks; track how it holds
Time the current build vs a poured serve; multiply by weekly volume
If the drink holds and the minutes are real, price the system
Start with two lines, not six
Choosing which builds to tap, speccing the dilution and carbonation, and writing the cleaning SOP is prep-system design: Prep-Lab Design & Build. If your ticket times fall apart at peak and the list never changes, fifteen minutes will tell you whether taps are the answer.