Customer Story: Kiinteistötahkola
In the very first month, the prediction accuracy of invoices covered by the automation increased to over 80%, and at the moment we are slightly above 90%.
Olli-Pekka Kuha
Enterprise Architect
Branches
9 branches across Finland
Revenue
Inudstry
Property management & real estate
Personnel
Kiinteistötahkola
Kiinteistötahkola is a management and brokerage agency that offers high-quality management services and real estate accounting for housing and real estate companies. They also offer rental and real estate brokerage, especially for apartment owners.
Purchase invoices in the real estate industry are usually repeated month after month. Despite that, there has been surprisingly little use of automation. How should the purchase invoice process be structured so that it makes the everyday life of accountants working in the real estate industry easier?
This is the question that Olli-Pekka Kuha recently had to face. He analyzed the purchase invoice process of Kiinteistötahkola, one of Finland’s largest property management companies, and realised something needs to be done.
He works as an architect in the product development of Premis property management software. An essential part of his job is also the automation of the financial management processes of housing associations, and increasing transparency.
“Premis is an enterprise resource planning software developed for the needs of housing associations. It includes all the functionalities needed to run a housing association. The Premis handles, for example, accounting, invoicing, hosting certificates, meeting functions – and also the accounts payable,” Olli-Pekka describes the software.
The challenge: half a million invoices a year
Kiinteistötahkola has a strong desire to be involved in the automation of financial management. The renewal of the purchase invoice process came to the product development table at the end of 2019. The goals of the project were to modernize the purchase invoice process, and increase transparency and the level of automation.
“Almost half a million invoices go through the purchase invoice process annually, so the number is significant. However, there is a lot of repetition in invoices, so we already knew when we started that with smart solutions we can really make things a lot smoother,” says Kuha.
The degree of automation of purchase invoices is close to zero - why on earth?
In the real estate industry, it is typical for purchase invoices to repeat largely in the same way from month to month. Invoices arrive from the same operators and the amounts tend to remain the same throughout the billing period. These are usually the types of invoices that would be easy to automate.
Even so, the degree of automation has been close to zero. Why on earth?
“In most purchase invoice systems available on the market, the automation of purchase invoices is done using rule-based automation,” explains Kuha.
Rule-based automation requires, as the name suggests, that different processing rules are entered into the system by a human, on the basis of which, for example, accounting is done.
“If you think of an accountant who manages, say, ten housing associations, each of which receives invoices from ten different suppliers every month, it’s easy to calculate how laborious the equation becomes.”
In Kuha’s example, there could be a hundred rules that need to be created and maintained, which of course takes its own time and is also quite tedious. At the beginning of the renewal project, it became clear that rule-based automation that had been available for a long time had not really been used at Kiinteistötahkola due to these experiences.
“I am not saying that rule-based solutions are pointless. On the contrary: they are still well suited to certain cases and environments. It is also worth noting that the use of rule-based automation and artificial intelligence solutions is not mutually exclusive,” Kuha clarifies.
Snowfox's artificial intelligence solution “fits like a glove”
Olli says that Snowfox.AI came into the picture already in the planning phase of the renewal process.
“After reading about Snowfox’s solution, I concluded that this kind of algorithm-based artificial intelligence service would fit our needs perfectly. There was no similar product on the market before, but I immediately understood that there was potential here.”
The discussion started by checking that the interfaces with the new purchase invoice system are correct. Kiinteistötahkola’s ambitious goal was to build automation where creating and maintaining rules would take no time at all, and Snowfox.AI seemed like a very promising tool for that.
“We agreed that we will try Snowfox.AI as soon as we get our own development to a suitable stage,” says Olli.
Easy implementation and low-risk investment
Integrating artificial intelligence into financial management software can sound like a laborious project. However, according to Kiinteistötahkola’s experience, this is not the case.
“Snowfox has productized its artificial intelligence solution really well and in a customer-oriented way. For example, we didn’t have to base our purchase decision on marketing speeches alone, but Snowfox offered a free test with our own data,” Olli-Pekka tells joyously.
Before the purchase decision, the test period carried out using Kiinteistötahkola’s own data went as expected. The end result was quite a high forecast accuracy, even though only three months of data were used. This proved the prior assumptions about the good suitability of the data for algorithm-based automation to be true.
After the test period, the company quickly moved on to the actual launch phase, which also received praise from Kuha for how smoothly it went.
“The implementation of the purchase invoice system, which is part of their partner network, worked very straightforwardly and only required a few meetings and participation in the final testing from us. Now we only pay for the correct predictions, which means that Snowfox also shoulders its part of the outcome.”
The degree of automation rose like a rocket
How has the original prediction of improvement in the automation levels in the purchase invoice process come true?
Handsomely, says Olli.
“In the very first month, the forecast accuracy of invoices covered by the automation increased to over 80%, and at the moment we are slightly above 90%. At this stage, the automation covers single-line invoices. In the future, our aim is to expand the automation to cover multi-line invoices as well, in which case we will reach an even higher level in overall automation.”
The accountants involved in the development have been very satisfied with the help provided by Snowfox – you could say that hunger grows as you eat.
“We have seen that the solution works excellently, so we want to move forward even faster. We haven’t calculated the exact figures for time savings yet, but we can say that time is being saved already, and with the introduction of multi-line invoices even more.”
Olli Kuha encourages all operators in the real estate sector to boldly consider and test the automation solutions that are best suited to the industry.
“If possible, you should test the automation with your own invoice mass and data. I dare to say that a good algorithm-based solution in real estate can raise the level of automation above average – and bring with it all the benefits.”
Want to know how much easier handling purchase invoices could be?
Let’s have a chat, no strings attached.
If you would like to talk to us about conducting AI accuracy test or would just like to ask some questions, just fill out this handy little form and we’ll get back to you within 24 hours (unless it is the weekend or a national holiday, then it might take a tad longer.)