A restaurant can know quite a lot about the people interested in its food and still make ordering lunch unnecessarily adventurous.
The promotion gets your attention. The website opens. The menu is a PDF best appreciated on a cinema screen. By the time you’ve established whether the lunch offer includes Tuesday, you’ve begun considering a sandwich from somewhere less mysterious.
Better customer data matters. So does giving the customer a decent next step.
Refresh Miami’s January 25, 2024 story about the launch of Aro, a Miami restaurant-marketing startup, makes this a timely local conversation.1 Its interest in data-informed promotions is the news hook. The website implications here are our own analysis.
A promotion, website and ordering system may each report success separately. One records clicks, another records visits and the third records purchases. What remains unclear is whether people can move comfortably between them.
For a fictional South Florida restaurant running a weekday lunch offer, I’d start at the customer’s end. The destination should explain the offer, its availability, the price and how to use it. Pickup and dine-in are different plans for a lunch break, so let’s not save that distinction for the final screen.
Then follow the intended action into the booking or ordering tool. Does the offer still make sense there? Has the customer lost their selection? Are they being asked for information they already provided?
These aren’t glamorous questions. They are, however, rather closely related to somebody getting fed.
When off-the-shelf stops fitting, see how we approach custom development.
Connect the journey before expanding the dashboard
First-party data comes from the business’s own customer interactions. It can help explain what people do, but the event definitions need to mean something.
Viewing a page, clicking Order and completing an order are different outcomes. A canceled booking is different again. If those distinctions disappear in the reporting, a very tidy chart can give the owner the wrong impression.
The systems need clear handoffs too. A webhook can notify an application when something changes. Martin Fowler’s discussion of event-driven systems helps distinguish different ways those updates can be used; the practical question is who receives the information and what they do next.2
Svix’s 2023 webhook work is relevant to the delivery side of that arrangement.3 An update arriving late or twice shouldn’t trigger repeated messages or leave a canceled order looking completed.
Start with one campaign and reconcile its reported outcome with actual orders or visits. Include the offer’s operating constraints. Lots of interest isn’t much comfort if the kitchen can’t fulfill what was promised.
Pairs well with Miami has robot kitchens. Someone still has to handle the missing chicken.
Keep information requests proportionate as well. Helping someone reserve a table doesn’t require learning their entire biography. If you’d like permission for future marketing, make that choice clear rather than treating a dinner reservation as a long-term relationship agreement.
There are useful opportunities after the first visit. A returning customer might appreciate finding a previous order, provided current availability is accurate. Relevant follow-up can help. Repeatedly sending the same discount because an automation once enjoyed a successful afternoon may help rather less.
I’d begin with one journey the team can explain and a customer can complete. Improve the destination, define the important events and check the results against reality.
A restaurant website doesn’t have to do everything. Helping someone choose, order and understand what happens next is a very respectable start. Ideally, while they’re still hungry.