Two hundred and fifty beers, and no way to choose between them.

The problem

El Bait Shop, in downtown Des Moines, carried one of the largest selections of American craft beer anywhere — more than two hundred and fifty of them.

A selection that size defeats the format it arrives in. A printed list of 250 names is hard to use, and it goes out of date the moment a keg blows. Behind the bar, staff answered the same question all night — what should I get? — and the useful answer depended on what the person already liked, which is exactly what a list cannot ask.

The owner came to us wanting a beer menu for phones. He had seen one we built for the Des Moines Social Club, liked that customers could browse more easily on their own screens, and was clear he did not want online ordering.

A phone menu would have missed most of the room. The people who most need help choosing are the ones standing at the bar looking up at a wall of taps, and they are not going to pull out a phone and type in a URL to solve it.

At the second meeting, sitting at the bar, I noticed one of the coin-op arcade cabinets in the corner — a format everybody already knows how to use without being told. That was the answer: put it in the room, at the bar, where the question actually gets asked.

The selection is the interface

The screen does not show a list. It shows the beer.

Every bottle and glass sits on the shelf as itself, arranged the way the wall behind the bar is arranged, so the thing on screen matches the thing in the room. You touch a beer to learn about it. You touch a style to narrow the wall down to it.

That distinction is the whole product. A list asks you to already know the name of what you want. A wall lets you point at something because it looks right, which is how people actually choose in a bar.

It works for somebody who knows nothing, and for somebody who knows everything

A general customer who could not tell a saison from a stout needed to get somewhere useful in a few touches. A serious enthusiast — and El Bait Shop had plenty, the launch drew a crowd of homebrewers — needed enough depth to be worth their time, or they would dismiss the whole thing as a toy.

Those two people want opposite interfaces. The resolution was to make browsing shallow and detail deep: the top level asks nothing of you and demands no vocabulary, and every level after it rewards knowing more.

Touching something dark gives its name, what it tastes like, and whether it is on tap right now — so it gets ordered by name rather than by pointing at a tap handle, and the staff spend less of the night answering the same question.

One inventory, three places it shows up

The bar's stock changes constantly. Any system that needed updating in more than one place was going to be wrong within a week, and a wrong menu is worse than a paper one.

So there is a single backend the owner maintains himself. It populates the in-venue browser, the beer menu on phones, and the bar's own website. Update the stock once and all three agree.

This is the part that turned a one-off installation into something worth repeating.

What changed

It launched on 17 February 2011, at the bar, in front of a room of homebrew enthusiasts.

KCCI ran a segment on it and the Des Moines Register wrote it up. Silicon Prairie News covered how it came about. After that we were fielding enquiries rather than making them, including from craft brewers in other states.

It also stopped being a single installation. The same approach shipped again as a wine browser, and again as a tablet menu for Americana across town — a full beverage list with food pairings, concept through build.

The useful part was the second meeting. The brief was a phone menu; the problem was people standing at the bar unable to choose. Those are not the same thing.