Happy hour ROI: the maths nobody does before the chalkboard goes up
By Jack Hawkins · Updated
Happy hour pays back when extra contribution from genuinely additional trade exceeds the contribution lost through discounting existing demand and the extra operating cost. A busier offer window is not enough. Compare the whole relevant trading period with a credible baseline, then separate new sales, switched full-price sales and extra staffing. This guide gives UK operators a repeatable calculation before a chalkboard becomes a permanent margin commitment.
The reason the topic survives in every venue's marketing kit is that it looks like growth: bar gets busier, revenue line ticks up, chalkboard validated. The reason it dies in the P&L is that revenue during 17:00-19:00 isn't the question; the question is total weekly contribution with the offer versus without it, and that comparison almost never gets computed, because computing it means turning the offer off and measuring.
The four ways happy hours leak money, priced
Cannibalisation. The regular who would have paid £11.50 at 20:30 now arrives at 18:15 for £7.50. Every converted full-price serve costs you the £3.33 net gap at 20% VAT plus the margin on whatever they'd have drunk second at full price. In a neighbourhood venue, the size of this effect must be estimated or measured rather than assumed.
The early-shift discount on late staff. Happy hour builds an early crowd that clears before peak; if your rota peak-staffs 17:00 anyway, the offer costs margin and buys nothing. Daypart data first, offers second — the labour guide has the curve method.
Unit-price erosion. Weeks of discount pricing re-anchor what your regulars believe a pint "costs", and the menu's £5.80 starts feeling like a rip-off against the chalkboard's £4.20. A change in perceived value is not directly measured by the POS; investigate it through guest feedback and actual demand rather than assuming it is permanent.
Serves-per-head substitution. A discounted first drink that replaces the second full-price drink — the guest who comes at 6 and leaves at 8 instead of coming at 8 and staying till midnight — is negative-ROI by a wide margin, and it's common on wet weeknights.
The maths before the chalkboard
Four numbers from two weeks of daypart data, all ex-VAT:
- Baseline 17:00-19:00 net revenue and contribution (a venue that's already comfortably busy at 18:00 has nothing to fill — stop here).
- Incremental units the offer plausibly adds (be honest: a chalkboard adds tens of guests, not hundreds; steal your own Friday footfall as the ceiling of "new").
- Cannibalised units: existing-guest serves repriced down. Use staff observations alongside booking or loyalty information where available, and test low, middle and high switching assumptions. Halving an estimate does not make it a statistical lower bound.
- Drag: discount per serve × all discounted serves including the cannibalised ones.
Incremental offer profit = (incremental units × discounted contribution) − (cannibalised units × lost contribution) − fixed extras (extra early staff hour). In a mid-weekroom venue with genuine empty capacity, low cannibalisation and a narrow discount (£1.50 off two house cocktails), it clears comfortably. In a Friday-adjacent room with a £3-off rail, test whether the result remains positive under realistic switching assumptions.
The structural alternatives usually beat it: a value tier rather than a value hour (a £6.50 house option on-menu all week, no anchoring shock, no training to arrive at 18:00), an event (quiz, vinyl, tasting) that fills capacity with full-price spending and a reason, or a food-led early offer where the margin math is cleaner because it's a new occasion, not a repriced old one.
If you're going to do it anyway
Narrow it: two or three specific products (the ones with the best contribution-to-price ratio at discount, from the matrix — the house lager is almost never one of them), a hard window, and an internal agreement about when it gets killed. Then measure like a scientist for four weeks: daypart revenue and contribution vs the same weekdays prior month, plus a bartender headcount of "regulars during the offer". A keep/cut decision made on those four weeks of numbers is worth more than a decade of happy-hour folklore, including this article's.
Put a sensitivity range beside the offer
Illustratively, 100 additional serves at £4 discounted contribution generate £400. If 60 existing serves lose £2 contribution each and extra staffing costs £80, incremental profit is £200. At 160 switched serves, the same offer produces zero incremental profit: £400 minus £320 minus £80. This makes the uncertain switching assumption visible. If the extra investment is the £80 staffing cost, the first scenario is 250% incremental ROI on that explicitly defined investment; a wider campaign budget gives a different ratio. Keep the pounds result visible alongside any percentage and measure the whole relevant trading period.
FAQ
Does happy hour actually make money? Sometimes: when it fills genuinely empty capacity, the discount is narrow, and cannibalisation is low. Discounts can attract incremental demand and can shift existing purchases into the offer window. The balance is venue-specific: evaluate total contribution over comparable periods and show a sensitivity range for uncertain switching.
How do I measure my bar's happy hour? Compare total weekly contribution (not offer-window revenue) across four offer weeks vs matched non-offer weekdays, and count discounted serves taken by existing regulars. Contribution up with the room fuller is a keep; the result should be judged on contribution, with differences in weather, events and coverage noted.
What's better than happy hour for filling quiet hours? Events with a reason (quiz, tasting, live) fill on full price and build habit; a permanent value tier captures price-sensitive demand without an anchoring cliff; early food offers create a new occasion. All three avoid repricing the drinks your evening trade would have bought anyway.
Why does our happy hour crowd not stay? Because the offer's structure trained a departure time. Add a second beat (kitchen open till 21:00, DJ at 20:00, late list on one night) so the offer is a doorway rather than a destination, and watch whether the 21:00 covers line moves before extending the discount further.
Method and assumptions
The approach here is an illustrative incrementality model, not a claim that one study proves all happy hours fail. Incremental profit equals additional discounted contribution less contribution lost on switched sales and extra costs. If quoting ROI as a percentage, divide that incremental profit by the defined additional investment and multiply by 100; if investment is zero, show profit in pounds rather than an undefined ROI. Promotions must comply with the rules in the relevant UK jurisdiction, including Scotland's separate on-sales restrictions.
Last reviewed 19 October 2026.