The modern data stack enables reporting and customer-facing recommendations via machine learning.
“When we only used Fivetran for reporting, it was important but not absolutely critical. But now, with the part it plays in updating the behaviour of the website on an hourly basis, Fivetran is business critical.”
Gil Machado, Data Engineer, Wunderflats
Destination: Google BigQuery
Business Intelligence: Sisense
Doubled conversion rate from 30-40 % to over 80 %
Identified new market segments (e.g. student accommodation)
Advanced the use of AI and machine learning
ELT data kept flowing with Data Engineers on holiday or sick leave
Deeper levels of data exploration
Saved time, money and resources
Launched in 2014 to make the housing market more accessible and transparent in Berlin, Wunderflats is Germany’s leading platform for temporary furnished housing. By combining personal contact and artificial intelligence, they create the best user experience in the real estate industry while providing more services at more affordable prices than traditional brokers. The site has grown rapidly to cover housing markets all over Germany, thus building up industry-critical knowledge and data about temporary furnished housing needs and offers. This has been pivotal to Jan Hase, CEO, who, together with his co-founder Arkadi Jampolski, is building a data-driven business and making decisions based on the analysis of multiple data points rather than hunches. So far, this approach has proven right: Wunderflats boasts exceptional matching rates between tenants and landlords, and their product, catered specifically to their users’ needs, is rated with the highest customer satisfaction in the real estate industry.
There was no dedicated BI (Business Intelligence) tool when the company started, so they used a combination of manual data extractions and Google Sheets to provide insights into web activity. The firm then advanced its analytics capability by investing in Sisense for BI and Google BigQuery as the data warehouse.
While the company’s software engineers were enthusiastically developing features and functionality for the web platform, there was a feeling that data was still not being used to its full potential. Analysts had created manual pipelines for transactional data, but volumes were low and insights limited.
Gil Machado was recruited as Data Engineer in 2019 and set about creating pipelines and self-service access for employees, enabling them to drive business improvement in their departments. “What was missing was a reliable way to sync the data from the databases to the data warehouse,” he said. “Fivetran was introduced, allowing us to do much more and identify which data points present the greatest opportunities for us.”
Fivetran connectors are used to mine two main data sources – SendGrid, an email marketing application, and Mongo DB, which provides transactional data from the production database. Around 80 percent of the data is now loaded into the warehouse, enabling the business to carry out more advanced analytics.
Before, insights were restricted to basic indicators, such as monthly revenue and number of apartments. Now Fivetran facilitates much deeper levels of data exploration. The big win for Wunderflats is freeing up time to focus on higher value projects. He gives the example of Fivetran connectors accessing data for heuristic systems and machine learning models that determine the way apartments are ranked and displayed on the search results page.
Drawing on data that is updated with Fivetran connectors every 15 minutes, the list matches supply to demand and benefits the bottom line. “Basically, we are making sure that the properties that are most likely to convert with the needs of an individual user appear highest in their search results page,” said Machado. “Every hour on the dot, the scores are updated and optimised. We improved our matching to the point that more than 80% of tenants who request an apartment with Wunderflats proceed to conclude with a signed contract compared to 30%-40% before.” This is a significant innovation for the traditional housing market, because Wunderflats’ tenants can effectively rent an apartment completely online without having to peruse huge numbers of listings.
Large volumes of data provided by Fivetran are now accessed via Sisense in self-service dashboards and reports. The company’s different departments – sales, product, finance and real estate – all have access to data with around 10 employees using dashboards and up to 50 receiving regular reports.
The role of Fivetran is evolving, however, and becoming more strategic. “When we only used Fivetran for reporting, it was important but not absolutely critical,” said Gil Machado. “But now, with the part it plays in updating the behaviour of the website on an hourly basis, Fivetran is business critical.”
The data also enables more ad hoc analysis that reveals new market opportunities. For example, Wunderflats was able to identify specific challenges around student accommodation. As a result, the specific needs of students are now reflected in a specific user flow on the website.
Product innovation and oversight of conversion rates has benefitted the bottom line and there are other cost benefits. Automation and scale offered by Fivetran have saved time, money and human resources that were previously allocated to manually building pipelines and then maintaining them. “It allows us to focus on feature engineering to support new machine learning models, delivering more value to the business,” said Gil Machado.
Even before the higher value benefits, outsourcing the development and maintenance of pipelines had justified the Fivetran investment, according to Machado. “Something like Fivetran is really advisable in ensuring critical pipelines keep flowing,“ he said. “It has delivered a huge return on investment.“
Fivetran, the leader in automated data integration, delivers ready-to-use connectors that automatically adapt as schemas and APIs change, ensuring consistent, reliable access to data. Fivetran improves the accuracy of data-driven decisions by continuously synchronizing data from source applications to any destination, allowing analysts to work with the freshest possible data. To accelerate analytics, Fivetran automates in-warehouse transformations and programmatically manages ready-to-query schemas. Fivetran is headquartered in Oakland, California, with offices around the globe. For more information, visit fivetran.com.