7 Oct 2020Blog

Thoughts on data quality in the footsteps of Covid-19

Thoughts on data quality in the footsteps of Covid

Is data quality important in the battle against Covid-19, and is it really something to worry about for a restaurant owner in Berlin?

Arriving in Berlin

When the summer was at its hottest, life took me down to a weekend in Berlin with my daughter. It’s a wonderful city in many respects, even in the hot summer heat and prevailing coronavirus pandemic.

Once in Berlin, we left our gear in the apartment we had borrowed for the stay and went out into the neighbourhood to get something to eat. After a couple of minutes of aimless searching, we finally found a nice place, Café am Engelbecken, which I can also recommend if you are in Kreuzberg.

Finding out when to wear a mask

We sat down in the sun, and we had barely had time to sit down before a happy waitress wearing a mask appeared. Unlike Sweden, Germany has introduced requirements for mouth coverings in various contexts, but perhaps they are not as clear as one might hope.

Masks are required in shops and on public transport, but when it comes to restaurants, it is not as clear. Sometimes it is required, sometimes not. Sometimes you have to wear it but only when you move around inside the restaurant.

Thoughts on data quality in the footsteps of Covid

In the combined hat and wine store Hut und Vino in Prenzlauer Berg, you must wear a mouth covering.

Registration before dinner is mandatory

The waitress informed us that before she could take any order, we had to provide some information about ourselves. She gave an explanation in German, and I picked up that it was something about coronavirus and that the information would be used for contact tracing in the event that an outbreak could be traced to the restaurant. She handed over a piece of paper and a pen and asked us to write down our names and telephone numbers. It turned out that it was enough that only one person in a party stated their information. That was my job. I entered my name and phone number in a simple table on A4 paper.

The table already contained about ten handwritten rows of information, which was probably the names and contact information of “contact persons” for other parties who visited the restaurant during the day. After we had given the required information, we finally were allowed to order. A few minutes later, a fantastic breakfast came in, which we enjoyed in the company of about a hundred wasps.

Data quality Data quality blog

Café am Engelbecken served a nice breakfast that we ate together with what must have been all the wasps in Kreuzberg.

Bad ways and less bad ways to collect and store contact information

During the following days, I visited, either alone or with my daughter, several restaurants of different types and classes. What they all had in common was that they collected information about their guests. The same type of information and for the same purpose, i.e. in the event of an Covid-19 outbreak, to be able to trace guests that have been visiting their restaurant. Just as the quality of what was served varied between the restaurants, also the way the information was collected differed between different restaurants. A lot.

In a less fancy – or let’s say a more relaxed – place, the restaurant collected customers’ details on small pieces of paper which were then deposited with other guests’ notes in an old Kleenex box which for the day played the role of an archive. As a guest, you were free to interpret what information was to be written down, so after a bit of internal debate, I ended up writing down my first name, email and my mobile number on a small piece of paper, which I squeezed into the already pretty full Kleenex box.

At a more structured establishment, I was asked to scan a QR code that led to a website with a simple form where I could enter my contact information. When I finished entering my details, I was asked to “check in” and to remember to “check out” when I left the restaurant. Perhaps the intention was to collect information about which times different guests were at the restaurant. Anyway, I forgot to check out.

Data quality blog

At Mundvoll good food was served, but only if you first registered your contact information when you arrived at the restaurant.

That is about how it went. At each restaurant, the visit began with some form of registration. Never the same procedure, but the most common method was to fill in your information more or less structured on A4 paper.

Can the collected data be trusted?

After a while, I started thinking about data quality in this context. I see several challenges for those who will use this information to trace restaurant guests in connection with a possible Covid-19 outbreak.

Though the methods for collecting information were different, there was one common denominator: a lack of controls that the information given was correct.

  1. There was no “input validation” – e.g. is the name complete? Is the phone number a valid phone number? Is there a country code? Is the email address a valid e-mail address? Not even in the cases where some kind of digital form was used were any checks made.
  2. There was no control of ID, so in principle you could give fictitious names and contact information. Last but not least, in the cases where I stated my information in writing, there was no control that my handwriting could be interpreted by an actual reader. It is a well-known problem among my colleagues who have ever had to deal with my handwritten notes.

Why contact tracing most likely will fail

My conclusion is that the collection of information at restaurants is done because you as a business owner are required to do so. It does not seem that anyone has set requirements for exactly what information should be collected and that the quality of the information collected should be good enough to use for tracking disease cases.

I also cannot remember that I approved that they could store my information. Isn’t it personal information regardless of whether it is entered on a website and stored in a database or if it’s recorded on tissue-sized paper and stored in a Kleenex archive?

It must be impossible for several of the restaurants to know if the information is complete or correct. Also, as only one person in each party needs to provide information, it is assumed that the table guests know each other well, which is not always the case.

If, against all odds, an outbreak is traced to one of the restaurants I visited, then the chances are probably slim that I will be contacted. Mainly because the poor bastard who is given the unhappy task of interpreting my handwriting, would give up before he or she reached my last name, but also because it will be almost impossible to sort out these paper-based records in the near future. I assume that the information will be reported on to some authority, which in turn will use it. How do you do that when you get a “data dump” in the form of a Kleenex box?

These are very concrete examples of when people have not thought through how to collect, perform quality control and harmonise the data that they are going to handle. Although it is seldom this bad, it is not uncommon that organisations start collecting data without thinking through the actual requirements for the data: exactly what information should be available in what format, how to ensure correctness and quality, etc. When it comes to the crunch, data you cannot trust is not worth much.