Through the day without data - on the digitalisation of technical due diligence

Through the day without data - on the digitalisation of technical due diligence

Why technical due diligence (TDD) will remain largely manual work even in times of digitalisation. An inconsistent or sometimes barely existing database for many, especially older properties, hardly allows for automated data processing. However, there are good opportunities for the digitalisation of processes within the framework of a TDD, both with clients and the corresponding service companies, which enable a partially automated processing of data. Digitalisation is a big topic in all areas of the real estate industry: digital data rooms, BIM, robotics, AI, virtual and augmented reality or premaintenance. There is a lot of talk and writing about all these topics among experts. The example of a TDD shows which hurdles have to be overcome in order to digitise processes and services in a meaningful way and to answer the question where digitisation can be helpful and where it reaches its limits.

Technical due diligence maps the technical and structural condition of a property and is thus, among other things, an important basis for risk assessment in the transaction of real estate. It consists of an intensive on-site inspection of the technical installations and the detection of structural defects, the inclusion of existing data in the form of maintenance logs and other documents, the careful analysis of the available data and a subsequent assessment of existing risks in the form of a report.

Human expertise on site still indispensable for a long time to come

The automation of processes through AI and robotics is believed to have a lot of potential. Even if it sounds like science fiction, the question arises whether the inspection of a property as part of a TDD could not be digitised or automated in this way. "When assessing technical and structural defects, a large number of different parameters have to be taken into account to clarify possible damage patterns and their causes," explains Yannis Hien, technical real estate analyst at REC Partners GmbH, which specialises in the technical assessment of real estate, among other things. "To be able to assess damage or damage patterns on site, a long experience and professional judgement is required. Often, several causes for the same damage pattern come into consideration. Being able to recognise and assign this is, among other things, what makes the quality of a TDD. And for the foreseeable future, this will remain a matter of people, which only experienced experts can do," Hien is certain. After all, the transfer services required to deduce the actual cause of damage from a damage pattern and to evaluate it accordingly are far too complex to be able to depict them even remotely reliably with the help of an algorithm.

No learning libraries for AI

There is also another problem: AI learns on the basis of images within the framework of so-called deep learning. A Google project to recognise cat faces has already shown that tens of thousands of images are needed as a learning basis. And even then, the hit rate was only 75 per cent. "Establishing a clear cause-and-effect relationship between structural defects on the basis of images is not only much more difficult, but we also don't have a correspondingly extensive library of clear damage images that AI could use as a basis for learning," says Hien.

The database is crucial

Even if AI and robotics are thus not suitable for creating a sound TDD in the long term, the documents available in digitalised form, such as maintenance logs, data rooms or documents on technical equipment, could be processed in automated form. "Of course, it would be useful and very helpful if we could draw on existing data pools for this process," explains Hien. "That would make our work much easier. But in many cases, unfortunately, that is not a given either. When we provide technical advice to companies in transactions, in many cases we don't even have access to documentation that gives us an approximate picture of the condition of the technical building equipment, let alone a data stream." Especially in the case of older properties, there is often no documentation available at all on the technical systems from which the year of construction, inspection intervals and maintenance can be taken. "Sometimes there are handwritten maintenance records that have been scanned in several times," Hien knows. "But that is not the form of digitised documents that can be processed automatically." Therefore, currently in many cases, the processing of the data elicited during an inspection is still manual work here as well.

Premaintenance as a solution?

Today, very modern plants may already be maintained online as part of so-called premaintenance. Plant components are replaced at an early stage before they wear out. It could be the key to an always up-to-date and available database, at least with regard to technical equipment. But this procedure is expensive, and that will discourage many property owners from applying it to older installations, if that is at all possible in individual cases. "The age of the systems is a fundamental problem," explains Hien. "Because as long as ventilation and heating systems are running without major problems, people are reluctant to invest in renewal or replacement." The consequence: there is often no documentation or data on their actual condition.

Storage locations and interface problems - lifting the treasure trove of data

But even if all information were available in digital form, this does not mean that it is available and usable for TDD. This is because the required data is often stored in different places and an overall directory does not exist. "This is especially the case when, for example, facility and asset management do not have a common database, a classic interface problem and, in our experience, the rule rather than the exception," explains Hien. "But despite all the difficulties and obstacles that in many cases stand in the way of a complete digitalisation of technical due diligence, there will be great progress in the near future, at least in the area of data availability, which will make our work easier." This is because both the clients, primarily the large property owners, and service companies above a certain size are increasingly digitising their processes. "And here, the first thing is to lift the treasure trove of data in your own company, structure it and transfer it into automated processes," adds Hien. "For us, this means systematising and preparing the data from hundreds of TDDs available to us in such a way that we can use it automatically for future projects. And that will lead to us getting one step closer to digitalising a technical due diligence and making it more efficient and uniform."