Chapters from a rent survey to a listed landlord

Mandate

Bring rent, heating, building and tenant records under one governed structure, make the data behind tenant-facing decisions defensible, and prepare the company for EU sustainability reporting.

Outcomes

• Replaced the comparison-flat selection behind 21,000 rent letters a year; consent without dispute rose from 81 to 93 per cent and actions for consent fell from about 640 to about 140 a year.
• Reconciled heated living area in 1,240 buildings for the carbon-cost split; unresolved buildings fell to about 190 by the 2025 statements.
• Built a €9.1 million condition model for the 1,840 buildings in the €460 million riser programme; 212 buildings brought forward and 165 deferred.
• Deleted 3.4 million unneeded documents about former tenants and applicants; access requests answered in nine days instead of 31.
• Added condition evidence to the valuers' data tape from 2023, with the share of buildings lacking evidence stated each year.

Decisions in context

When the chief financial officer created this role in April 2021, rent data, building records, heating data and tenant files lived in four parts of the company with four definitions of a flat. My mandate was to give them one structure and to make the data behind decisions that reach tenants defensible in a stairwell, a courtroom and an audit. I began by reopening the rent-increase selection I had designed myself in 2018. The carbon-cost law, the riser programme and a data inventory that found 4.1 million documents we should not have kept then set the order of work. I lead 132 people with six direct reports and a €17.5 million operating budget. I have no authority over rents, capital works or valuations; my influence runs through the evidence I put in front of those who do, including the supervisory board's audit committee twice a year.

Reason for the move

Current role. Not seeking an executive move; considering one supervisory board seat with the management board's consent.

Mandate

Build a single rent roll and portfolio analysis for a landlord about to grow by a third through acquisitions.

Outcomes

• Integrated the rent rolls and building records of 48,300 acquired flats from the 2017 and 2019 acquisitions within nine months of each completion.
• Built the first complete valuation data tape, cutting valuers' data queries in the December cycle from about 1,900 to 310.
• Automated comparison-flat selection for rent letters in 2018, a design I later had to correct.
• Organised the company's contributions to rent-index surveys in 23 towns.

Decisions in context

I joined Haselwerder in September 2015 with a team of nine and a rent roll held in three systems. The company was about to buy 28,500 flats in Saxony and Berlin, and later 19,800 more, and my first task was to make sure we knew what we had bought: rents, lease dates, floor areas and the condition notes that sellers prefer not to volunteer. We integrated both acquisitions within nine months of completion. I also built the data tape our external valuers receive each December, which reduced their queries from about 1,900 to 310 a cycle. In 2018 my team automated the choice of comparison flats behind rent-increase letters. It worked as instructed and chose the highest eligible rents, and I did not question it hard enough while objection rates rose. That lesson came with me when I was appointed chief data officer in April 2021.

Reason for the move

Appointed chief data officer when the role was created.

Mandate

Give the valuation department independent evidence to test the rents, vacancies and repair assumptions in valuations of residential portfolios offered as loan collateral.

Outcomes

• Built a database of 410,000 rent and transaction records used to test valuers' market-rent assumptions.
• Introduced a comparison of valuation rents with borrowers' actual rent rolls for every portfolio loan above €50 million.
• Flagged repair assumptions below observed spending in 14 portfolios, leading to lower loan amounts in six.
• Completed a part-time certificate in property valuation in 2012.

Decisions in context

At Ostwall I moved from producing rent evidence to testing other people's use of it. The bank lent against residential portfolios worth billions and relied on valuations prepared for its own lending purposes, which German covered-bond practice expects to be prudent and sustainable rather than a snapshot of the market. My team of twelve built a database of 410,000 rent and transaction records and compared valuation rents with the rent rolls borrowers actually held. The most useful finding was about repairs. In fourteen portfolios the valuation assumed less spending than the owner was already incurring, and in six the bank lent less as a result. I studied valuation part-time and qualified in 2012, which let me argue with valuers on their own terms. I left because I wanted to sit with the owner, where the same evidence decides which buildings are repaired first.

Reason for the move

Wanted to work for an owner, where data decides what happens to buildings and tenants rather than only how much is lent against them.

Mandate

Make heating and service-charge statements arrive sooner and contain fewer errors, and give the managing directors reliable controlling data.

Outcomes

• Cut the share of statements delivered in the last three months of the legal deadline from 64 to 12 per cent.
• Reduced tenant objections to statements from about 2,100 a year to about 700.
• Built the company's first register of buildings, heating plant and floor areas.
• Introduced monthly controlling reports for the managing directors and the city's shareholder representative.

Decisions in context

Mittelweser owned 21,400 flats in one Lower Saxony city and two neighbouring districts, and its tenants included many households on low incomes for whom a heating back-payment was a genuine shock. In my first year about 2,100 households objected to their statements, and most of them had a point: readings had been keyed wrongly, floor areas differed between systems, and nearly two-thirds of statements arrived in the last three months of the legal deadline. I built a register of buildings, heating plant and floor areas, brought the reconciliation of fuel invoices in-house and agreed reading dates with the metering service providers. Objections fell to about 700 a year. Reporting to the commercial managing director, I also prepared controlling reports for the city's shareholder representative, my first experience of writing for a governing body that met only a few times a year.

Reason for the move

Wanted to work on property data at a larger scale and to understand how lenders and valuers judge housing portfolios.

Mandate

Design the samples, questionnaires and statistical models for rent indices that towns publish and courts rely on.

Outcomes

• Designed or rebuilt rent indices for eleven towns, five of them qualified indices.
• Introduced a published explanation of each index's method, written for tenants and landlords rather than statisticians.
• Cut the number of explanatory variables in the institute's standard model from 19 to 11 without loss of accuracy.

Decisions in context

I joined the institute from university in October 1999 and spent my first winter checking questionnaires collected door to door. From 2002 I led projects, with two analysts and up to thirty seasonal interviewers, and before I left I had designed or rebuilt rent indices for eleven towns. The work required statistical rigour and something harder: a method that a judge, a tenants' association and a landlord's lawyer could all follow. I simplified our standard model from nineteen variables to eleven and began publishing a plain explanation of each index's method. I also saw the limits of research from outside. We produced a number and left; landlords and tenants lived with it. In 2005 I moved to a housing company to learn what happens to data once it reaches a household's letterbox.

Reason for the move

Wanted to work inside a landlord and see how data was used once a rent index had been published.