I put a display system in a real shop and asked everyone what happened.

The title is Konzeption und Weiterentwicklung eines interaktiven Anzeigesystems zur medienübergreifenden Kommunikation, with the emphasis on social media. 118 pages, Martin Gessert and me, August 2014.

The premise was of its moment. Screens were appearing everywhere in public and all of them were broadcasting, one way, at people. The web had gone the other way and turned into something people took part in. So: what happens if you connect the two, and let the screen in the room carry what people are posting rather than talk over them.

We already had a prototype from earlier work, so the thesis was about developing it further and then verifying two interaction scenarios, one for a trade fair and one for retail. Point of information and point of sale.

Instead of testing it in a lab we adapted a usability engineering process, the one Mayhew wrote down in 1999, and used its shape to verify the scenarios in the places they were meant for.

The first was the university’s second IT contact fair on 27 March 2014, with 41 companies exhibiting. The job there was giving students a fast view of a whole day’s programme across a room full of stands. We surveyed the visitors and, separately, the companies standing at those stands, because they are two different users of the same screen and they want opposite things from it.

The second was the shop. A clothes, shoes and drinks shop, ten staff, on the second floor of the Anger1 in Erfurt. We installed the system, showed the staff how it worked in a short introduction, and then it was theirs. On 11 August we came back and sat down with the manager for an interview against a written guide, recorded with his agreement. Martin asked the questions, I took the record.

They used every module. Mostly they uploaded videos, ran sales promotions and put the drinks list on the screen. After the introduction there was no trouble operating it, and they set it up themselves each day.

The thing I did not expect is in one line of the interview record. The staff liked arranging what went on the screen. Putting the content together was something they could be creative with, and it lifted the mood in the shop.

Nobody had designed for that. We had built a system for owners to broadcast and for customers to interact, and the people who got the most out of it were the ten people working there, because we had accidentally given them a thing to make.

The thesis reports that customers noticed the window and that footfall rose noticeably. It also records the manager’s figure for how many people came through the shop on a normal day, 700 to 800.

Both of those are things a manager told us in an interview. Neither is a measurement. We had a dashboard in the system that counted QR code scans, and we did not build anything that could have told us whether the screen in the window changed how many people walked in. Three weeks is also not long enough to know, and we had no idea what the baseline was beyond what he remembered.

I would write that chapter very differently now. Not because the interview was worthless, it was the most useful two hours in the whole project, but because an impression and an effect are different claims, and we let them sit next to each other as though they were the same kind of thing.

The other thing I got wrong was the whole premise. In 2014 it was reasonable to assume people wanted their posts on a screen in a shop. They did not, much. And the platforms closed those interfaces soon afterwards anyway. Building on somebody else’s API is renting, and I was not thinking about the lease.

Chapter 4.1 is called Point of Sale. The following year I joined GoodsCloud, which NewStore acquired two years after that, and I have been building retail point of sale ever since. I did not plan that. I had been looking at the same problem from the other end and did not notice.

It became a product too. The digital signage system we sold through Zebresel, my studio at the time, came straight out of this research. That was the first time I turned something I had written into something people paid for, and it is why I stopped treating research and product as two activities.

The social layer is the part that aged into a warning rather than a lesson. Everything we built on top of other people’s platforms went away on their schedule, not ours.

And the question itself never left. Twelve years later I am building iPad as Display, which is an old tablet in the corner of a desk showing you your day. Same question, much smaller screen, nobody else’s API.