On this page
In 2020, pubs faced a difficult combination of practical constraints. They needed to serve customers while reducing unnecessary contact and managing changing public-health requirements. Technology could help, but it could also make an already difficult job more complicated.
I developed BarWaiter, a browser-based table-service application, during that period. The central design problem was not how to move ordering onto a phone. It was how to let customers order from their tables without expecting the pub to replace the working practices and equipment it already depended on.
Start with the actual service
A familiar mistake in software development is to treat the existing business as an obstacle that ought to be redesigned around the application. That was the wrong starting point here. A pub has staff, customers, stock, tills, payment terminals and established routines. Adding a digital order can help only if it fits within those realities.
BarWaiter let customers browse a menu and place an order using their own phone, without installing a dedicated app. Staff received the order in a back-office view that could run on an ordinary device. The order could then be processed through the pub’s existing till or electronic point-of-sale arrangement, with payment taken using its established card terminals.
This sounds deliberately unremarkable. It should. The design objective was to remove friction from placing and handling an order, not to impose a new operational model for its own sake.
A real field trial shaped the implementation
A field trial began at The Eagle public house in Norwich on 4 July 2020. The surviving BarWaiter website records the trial, including feedback dated 1 September 2020 from the owner and staff. The descriptions focus on practical benefits: managing table orders, making administration straightforward, handling product variations and working with the equipment the pub already had.
That evidence is more useful than a hypothetical demonstration because it describes a system used in a working venue. It also has limits. The testimonial material is hosted on the product’s own website; it is attributable feedback, not a controlled independent study of labour savings or service performance.
I used the trial to observe where the interface helped and where it still placed work on customers or staff. A digital system may be straightforward to its developer and still feel cumbersome to somebody trying to complete a normal shift. Feedback from everyday use matters because it exposes those mismatches.
Fit into existing responsibilities
One of the important choices was to keep the ordering process distinct from the pub’s existing payment and sales systems. Staff could read an incoming order, put it through their usual till and fulfil it, instead of being asked to adopt a new end-to-end business platform before they could use table ordering.
That boundary made the product narrower, but also more approachable. It avoided making a small operational improvement dependent on a large migration. The customer needed a short route to placing an order; the staff needed something legible and actionable; the manager needed to keep a menu and product variations up to date.
Different people encountered the same application with different responsibilities. Designing those interactions meant considering the sequence of work rather than merely counting screens or features.
Foundations figure 01
Table ordering without replacing the pub’s systems
A simplified operational path from the 2020 product description
- CUSTOMERPlace a table orderUse a phone and a browser, without installing an app
- BARWAITERReceive the orderStaff see the request on an ordinary device
- EXISTING SERVICEProcess and fulfilUse the pub’s existing till, payment process and service routine
Illustrative workflow, not a technical system architecture or transaction audit.
What the product taught me about simplicity
The most useful engineering lesson was that an interface should carry complexity on behalf of its users. Product variations, availability and order handling can be complicated, but that does not mean every complication needs to be exposed to the customer or bar staff.
The second lesson concerned evidence. Before the trial, I could check whether features operated as specified. During the trial, I could observe whether those features fitted the way the venue worked. Those are different forms of feedback. Neither should be mistaken for a universal performance claim.
BarWaiter was also a practical reminder that designing for people includes those who may not be comfortable installing applications, learning a new interface or changing established habits simply to complete a routine task.
The connection to later research
There is a recognisable thread between this work and the user experience of determinIQ. Both ask how to make a capable system accessible to someone whose expertise lies elsewhere. In BarWaiter, that person may be running a bar or ordering a drink. In determinIQ, it may be a manager trying to understand data rather than construct an analytical query.
The systems solve different problems, and there is no suggestion that the governed analytical architecture existed in the earlier application. The connection is in the approach: understand the actual task, reduce unnecessary exposure to the machinery, and keep the underlying responsibilities clear.
The experience also informs how I think about building software now. A working prototype is a beginning; the measure of its usefulness is what happens when it encounters the ordinary constraints of the people it is meant to serve.
Evidence and limits
The surviving BarWaiter product site describes the operating model, and the dated field-trial page records the venue and testimonial accounts. They support a real-world trial and the product’s described functionality. They do not independently establish comparative productivity, commercial uptake beyond the documented venue, or long-term adoption. Those claims are unnecessary to explain what the work taught me.
Historical context
Topic: Foundations
Explore further
Supporting material is listed below.
Further reading is listed below.