From door-to-door rent surveys to 196,400 flats

The first rents I ever analysed were written on paper questionnaires, collected door to door in a small town on the Weser in the winter of 1999. I was twenty-six, newly qualified in statistics, and the town wanted a qualified rent index that would hold up when landlords and tenants took each other to court. Half my job was regression. The other half was learning why a pensioner in a 1960s block and a young family in a refurbished flat would both distrust any number we produced. They had reason to. A rent index decides what a landlord may ask, and the households who answer the survey rarely see how their answers are used.

My mother kept the books for a small housing cooperative in Hildesheim, so I grew up hearing about heating statements at the kitchen table: which flats had complained, which meter reader had misread a radiator, why a widow on the third floor owed more than her neighbours. I studied statistics at Universität Haldenau and completed my Diplom in September 1999 with a thesis on sampling errors in rent surveys. It was not a glamorous subject. It turned out to be the subject of my working life.

From October 1999 I worked at Institut Weserland für Wohnungsmarktforschung in Bremen, a non-profit institute that towns commission to produce rent indices. By 2002 I led the projects, and before I left I had designed or rebuilt indices for eleven towns. I learnt that a method is only as good as one's ability to explain it to a judge, a tenants' association and a landlord's lawyer in the same week, and that a defensible method usually has fewer variables than a clever one.

In 2005 I moved into operations at Wohnungsbaugesellschaft Mittelweser, a housing company owned by a city in Lower Saxony, with 21,400 flats. As head of controlling and service-charge data I was responsible for the annual heating and service-charge statements. In my first year about 2,100 households objected to theirs. Most objections were justified. I spent five years making the statements arrive sooner and contain fewer errors, and I learnt to read an objection as a free audit.

In 2010 I joined Ostwall Pfandbriefbank in Hamburg as head of valuation data and market research. The bank lent against residential portfolios, and my team checked the rents, vacancies and repair assumptions in valuations against the evidence. I completed a part-time certificate in property valuation in 2012 so that I could argue with valuers on their own terms. What I took from the bank was a habit: before trusting a value, ask what the owner knew that the valuer did not.

I joined Haselwerder Wohnen in September 2015 as head of rent and portfolio analytics, reporting to the head of asset management. We integrated two acquisitions with 48,300 flats, built the data tape our valuers receive each year and, in 2018, automated the selection of comparison flats behind rent-increase letters in towns without a usable rent index. That last piece of work is the decision I explain most often, because I designed it badly. It chose the three highest eligible rents. It was lawful and it was unfair, and when I became chief data officer in April 2021 I reopened it myself.

Today I lead 132 people in data engineering, analytics, data governance, building records and data-protection operations, with an operating budget of €17.5 million. I report to our chief financial officer and present to the supervisory board's audit committee twice a year on the controls behind valuation inputs and the data in our sustainability reporting. Our data-protection officer is independent of me, as the law requires, and I would not want it otherwise.

I ask every analyst to be able to explain a figure to the person it affects. That rule sounds gentle. In practice it is demanding, because it rules out comparison sets a tenant cannot see, heating figures in units nobody understands and valuation inputs that leave out what our caretakers already know. Inside the sector I am probably known for three positions: that a rent increase letter should show its working, that heating data belongs to the tenant before it belongs to the bill, and that valuers should receive our bad news along with our rent roll.

Since June 2019 I have also sat, unpaid apart from an attendance allowance, on the supervisory board of Baugenossenschaft Lindenkamp eG, a Bremen housing cooperative with 4,800 flats, with my employer's consent. It has taught me what it is like to be on the receiving end of a management paper, and how little a board can see unless someone decides to show it.

Open to Board appointments · Speaking invitations

Languages / German · English