• Christian Schitton

Digital Transformation challenges in the real estate industry - the Analytics Level

Digital transformation unfolds in three main levels, i.e. the data gathering level, the analytics/ operational level and there is the result level. In other words, a company has to implement a good functioning system for data collection. Then it has to use the collected data for its respective purpose in an efficient way. And finally, it has to communicate the results to the relevant addressees in a form those recipients can work most effectively with.


This is not really rocket science. Though, the challenges lay in the fact that those levels are built up in a lot of separate layers, which represent the complexity of digital transformation processes.


And it's the management's responsibility to take care for the right balance among those levels on the one hand. On the other hand, management has to make sure that digital transformation is taking place within the company's current constraints (e.g. time frame, stage of transformation, needs at hand, cost budgets and similar) in order not to overstretch the whole system.


Last time we talked about the data collection level. This time we will have a look at the analytics/ operational level.



What could possibly go wrong?


Here are two examples which crossed our path recently.


Imagine a company which puts a lot of efforts into the data collection. Effort means that spreadsheets were created with utmost details. As those details have to be partly provided manually, so it takes a lot of personnel and time resources to hold up the information flow.


However, on the analytical level there is no tool available to make proper use of all the data provided. Instead, the company has to rely on a standardised business analytics tool, which came along as a side product of an accounting system.


Standardised analytical tools face the following challenges:

  • In general they require a standardised data input. This can result in a lot of dead ends, as this kind of input may not exist as needed respectively the gathering of that data takes too much of the company resources. Another flaw may be that the format of the data provided and the format in which the standardised analytical tool is able to work may not harmonise.

  • They usually offer a standardised range of business analytics metrics. Standardisation does not take into account the special requirements a company might have in achieving its goals. This often leads to the point where a company is not using part of the collected data. The company might try to circumvent the issue by additionally creating its own business analytics metrics (most often in Excel spreadsheets or even Google Analytics). The result is an uncoordinated grid of business metrics solutions stored in different systems and fostering the build-up of unfavourable data silos.

Consider another company which organised its data collection by means of a data warehouse. Again, we have a situation where along with the data base a standardised business analytics tool is offered. In this case the management paid for a partly-customised version in order to reflect their special needs.


However, in practice we witness, for no evident reason, that the semi-customised analytics tool hasn't been used so far. Rather, the available data pool was converted into Excel sheets on an on-demand basis while a somehow customised in-house business analytics tool was created on Excel as well. So the digitalisation process which was started on the data collection level was reversed and the parallel flow of info made its way back to the lists...


With the internal analytical demands growing, this company faces the situation that meanwhile those Excel frames have become hardly manageable and the use of the semi-customised application gets interesting again. In order to switch back to it, the company would have to get an upgrade. Though doing this, all of the semi-customised parts would get lost as the upgrade just addresses the standardised product.


You get the picture.



Customised Analytical Tools


The fact is that one of the main reasons we are gathering data is to analyse them - this is something which is often getting forgotten! However, it seems that companies are so stuck in the data collection stage that the analytics phase does not get the right attention and therefore it is very hard to retrieve relevant business insights.


We mused in a recent article that digital transformation does not automatically have to be a deep burden for a company's infrastructure and that open source solutions, like R or Python, can be of great assistance in this respect.


The point is that there are highly adaptable tools available for backing the development of customised environments for any of those tasks. And this, of course, is also true for all of the levels in digital transformation, including the analytical/ operational level.


The span of possibilities runs from simple exploratory data analysis, business analytics, predictive analytical approaches to computational intensive machine learning applications in a real-time environment. There are countless ways for a company to create exactly this kind of management information system it needs to aim for its goals without having to compromise on standardised applications which stop short of delivering sophisticated results.


Exploratory Data Analysis


This is the first step to get to know your data - on portfolio, company and market level.


Crawling into different aspects of the data pool allows a company to get familiar with the data structure and to examine separate details of the current situation. In this stage we are also able to go through the data using statistical modelling tools in order to check for outliers, suspicious behaviours or irregularities.


Main purpose is of course to explore the data, according to the tasks we have at hand and to prepare them properly, Necessary data sets get pre-processed, i.e. getting them into a data format the analytical tool needs to be fed with. Side effect in terms of data setting is that we do not have to rely on the structure of the database or IT infrastructure of the company anymore.


Business Analytics


Business analytics builds up the backbone of any management information system. Saying this, every industry has different requirements which is also the case for the real estate business. The more important it is to have an adaptable, flexible and customised information system in place.


Here, the scope of applications moves from e.g. liquidity checks, Cash Flow scenarios, yield considerations to market value calculations, monitoring the debt funding frame, screening the portfolio rent payment patterns or Value-at-Risk assessments and similar. Business analytics is focused on an as-is stage and establishes the foundation for a more advanced management information system.


Predictive Analytics


In case decision makers are thinking steps ahead of the competition, the advancement of a management information system can be achieved with predictive analytics. These applications can be handled as an add-on to existing business analytics tools. It is backed by the business analytics environment and it works best when interdependencies between business and predictive analytics are incorporated.


Predictive analytics allows us to get a short and mid-long term perspective of a company's situation and its risk exposures. The range of assignments in the predictive analytics area is manifold:

  • outlook of single market parameters dependent on developments in a defined market cluster

  • mid- to longterm simulation of certain risk/ performance parameters (e.g. equity yield, DSC - ratio) in an investment portfolio,

  • prognosis of price developments of construction parts during a development,

  • recognising exiting tenants in an early stage by their communication behaviour, or

  • the impact of market sentiment on a potential changing market behaviour and its impact on the risk exposure of a portfolio in this market

This is just to name a few. Predictive analytics engages a lot of statistics as well as machine learning/ deep learning applications. This always depends on the task at hand, size of the available data pool and the frequency of the incoming data.

 

Needless to say, that all of those stages (i.e. exploratory data analysis, business analytics, predictive analytics) are to be linked to each other and are embedded on asset level and further aggregated to portfolio and corporate level.


Some Examples


In several previous articles we showed the range of possible analytical applications and the high flexibility those tools are offering.


As an example, here is one which discusses the impact of interest rate changes on the performance of real estate assets taking into account current low yield levels: https://medium.com/analytics-vidhya/market-correction-in-alternatives-real-estate-a50200dae32


This article discusses the impact of external markets on the own investment area when market clusters were built up: https://medium.com/p/27760eb651cb. By the way, this is an excellent example of how important exploratory data analysis can be also on the market data level.


Here is another example, illustrating Value-at-Risk (VaR) assessments with respect to share price movements of a public listed real estate company.


This case shows the development of VaR-positions in several "quality" stages, starting with historic volatility (in blue - highly inefficient), to moving averages (in red - reacting delayed on current market circumstances) to taking into account the changing volatility over time by means of ARIMA-GARCH models (in black - reflecting market risk on spot).


Final Thoughts


The real beauty comes with the notion that it has not to be pre-defined which kind of problems or which kind of metrics is to be handled. Any creation of a business metrics can be adapted to changing situations, be enlarged according to new necessities or be downscaled when the business task shrinks in a fast and efficient way.


This is possible because all can be packed up in modules. Read more about our modular approach in digital transformation for real estate here: https://www.d-darks.com/post/d-darks-modular-solutions-for-the-real-estate-industry.


The analytical stage is the most underestimated stage in the digital transformation process in the real estate industry. We have so many technical tools available to help businesses thrive and become a part of the data driven decision making process in order to get the best insights for their decision making and therefore keep their risks under control.


This is what tech development is capable of at this very moment and we do not see why real estate companies would not use all of it in order to achieve more decisive progress in the digital transformation process while strengthening their competitive advantage.