Drive-thru hospitality · working file

Experiment design: the two arms of the delight hour

A giveaway arm against a handwritten-note arm, in matched store pairs with a control group, for one pre-selected low-traffic hour a day, one reporting period only, designed inside the six guiding principles.

Employment workKFC South PacificPilot two

The question

Does a surprise moment layered on the hospitality foundation lift satisfaction and word of mouth without breaking the guiding principles? The design compresses the intervention to its smallest honest unit: one pre-selected hour a day, one reporting period, one token with every receipt in the hour, promoted only by word of mouth. The word of mouth doubles as a free uptake detector, because guests asking about the hour at other times are proof the surprise is spreading.

The two arms

	Giveaway arm	Handwritten-note arm
Token	Small branded merchandise, occasionally a small discount	Branded affirmation cards; the crew member may add a handwritten line
Delivery ritual	Gamified: spin a wheel, pick a number, lucky dip, flip a card	Handed over with the receipt and a genuine line
Cost profile	Low cost, with stock and storage, sized from historical transactions	Zero cost
Execution	One token with every receipt in the hour, one reporting period	The same
What the pilot found	Edged the note arm on the satisfaction averages	Positive; won on cost, lost narrowly on outcome

Neither token appeared in guest comments. The gift was not memorable; the interaction around it was. That is the finding the recommendation rests on, and the lift figures themselves are held by the business.

Why one hour, and why a low-traffic one

The window was chosen by operational economics. A low-traffic hour carries fewer transactions, less stock, less storage, less staffing stress and a higher baseline satisfaction. Bounding the hour also made the hype exclusive, one hour, every receipt, maximum effort, nothing to sustain all day, which cut the training burden. The hour is fixed and pre-registered; stock is sized from historical transactions with an uplift allowance.

Safeguards

Every concept passed through the six guiding principles as a decision filter; anything that failed one was dead on arrival.
Rejected: food and voucher giveaways (register codes and tracking friction); numbered-guest targeting (complexity and gimmick risk); a standalone discount (needed commercial sign-off, so it folded into the giveaway arm).
Apology gestures are banned by principle: the gesture is never a reaction to a negative event.
An adversarial design review flagged the automation rollout as a confound, so the two arms were balanced across automated and non-automated restaurants.
No margin erosion: the giveaway arm's cost is absorbed by the low-traffic window; the note arm costs nothing.

Crew first

Surprise the crew before the guests. For a week before the guest hour launched, managers were given tokens to surprise their own team, so that people experienced being surprised and then passed it on. On top ran a gifting licence: one crew member per shift may gift a free upsize, item or discount at their own judgement, capped at one gesture per shift by the margin review. The internal gesture is a training mechanic that feels like a perk.

Measures

Measure	Definition	Source
Service satisfaction	Top-box share, test against national, intervention hour only	Feedback platform, weekly
Service quality	Top-box share, same comparison	Feedback platform, weekly
Order accuracy	Error-free orders, test against national, hour only	Hourly platform cuts
Order speed	No regression	Operations telemetry, weekly
Positive sentiment	Share of positive comments, hour only	Comment classification
Friendliness mentions	Share of comments naming friendly service	Comment classification
Execution	A token with every receipt in the hour; no complaints about the gesture	Mystery-shopper visits and the platform
Crew	Confidence for the guest interaction	Crew survey, week one

The design pre-registered the fixed hour and asked the feedback platform for hourly-level reporting, a capability it did not have and which was never fully delivered. The gap is reported openly, and attribution relies on the window the platform could produce.

Working file from employment at KFC South Pacific, anonymised: no restaurant, colleague or vendor is named, and the figures are the ones the case study states. Back to the case.