Building a realtor recommendation tool with the help of data science

In the fiercely competitive property market, “location, location, location” is not the only factor that influences success. Having the right realtor can also make the difference between a big gain or a big loss (or a long sell time versus a short sell time).

Like many other industries, real estate covers a broad range of offerings. The most successful realtors are often specialists in specific property types (ex. condominiums, houses, commercial lots) and/or specific regions (rural, urban, and everywhere in between). But for a non-expert looking to buy or sell a property, finding the right realtor can be difficult, as most advertise that they can do all property types.

Enter Keylo, a free online site that helps property sellers/buyers find a realtor who is best suited to achieve their end goals.

“We’re not tied to any realtor, and we have no stake in making money off of the customer. Our only interest is helping you achieve your actual objective, which is to buy or sell a property,” says Keylo founder Ryan Mracek (pictured right).

Keylo’s website offers statistics on real estate professionals from across Canada, allowing users to see the realtor’s success rates with specific property types or locations.

The company spent the last two years collecting this data. Its next step is to perfect a referral tool that would allow users to type in their location and property goals, and receive recommendations on specific realtors. But when Keylo’s staff began investigating such a tool, they realized they would need some outside AI expertise.

Enter Cybera’s Data Science For Albertans program

“We wanted to create a ‘best realtor’ scoring system that would show users which realtor has the best chance of selling their house. This is more of a qualitative measurement — the best realtor may only sell 15 properties a year, but if those are specifically geared to your property type, they could be the best option for you,” explains Mracek.

In the spring of 2019, Cybera’s data science team worked with Keylo’s staff to build a data-driven model for realtor referrals. With this prototype in hand, Keylo is now able to iterate and flesh out the service.

But for Mracek, the biggest benefit of working with Cybera has been the knowledge and experience he has gained for developing a data science-based product.

“The plan has changed as we’ve gone along. Part of our first model failed, which is good, as we learned from that,” says Mracek. “My background is in accounting, so there was a lot I needed to learn about this technology. But now I feel like I can manage a data science project, and make decisions with confidence.”

A unique program for Alberta businesses

Like many other small-to-medium sized enterprises in Alberta, Keylo did not have the resources to staff an in-house data science team, or hire a consultant to trial different models and iterations.

“We were looking at grants to finance our data science project, but weren’t having any luck,” says Mracek. “Our company doesn’t easily fit into a specific infrastructure category. Because we’re involved in real estate, we weren’t seen as a ‘tech company’, so we couldn’t get the grants we needed.”

Data Science For Albertans is a free program, funded by the Government of Alberta, with the goal of getting more Albertans to utilize and benefit from AI, machine learning, and other data science-related tools. As well as introductory presentations and workshops, Cybera offers short-term, in-depth consultations with small-to-medium sized businesses and entrepreneurs across Alberta.

“We believe having the right data skills and tools will become necessary for Albertans to compete on a global scale,” says David Chan, Data Scientist at Cybera. “We therefore want to share our experience and expertise with Alberta companies so they can make more data-driven services available, or make operational decisions that maximize their potential.

“A major benefit of our Data Science For Albertans program is that it creates a safe (and free) space for companies to trial-and-error these technologies, so they don’t have to risk putting a major investment into an approach that may not work,” Chan adds. “Having this kind of a ‘sandbox’ in which to experiment is both useful and necessary for creating innovation. We hope this program will encourage others — from a wide variety of sectors — to try out these digital technologies.”

Says Mracek: “What’s great about Cybera is that they are helping to accelerate companies like mine. We couldn’t afford to pay someone to do this work, and this process of getting going has been really important to us. And we’re ready to learn more! After this summer, we’ll be in a much better position to compete with the bigger products.

“If we can keep expanding, maybe we could hire an in-house data scientist,” he adds. “We could potentially even go international.”

Cybera will be hosting a series of Data Science Industry Fellowships this August in Calgary and Edmonton. This initiative will provide opportunities for additional companies to build out a proof-of-concept data product. For more information on how to join as a potential fellow or industry partner, visit the program webpage.

2 thoughts on “Building a realtor recommendation tool with the help of data science”

  1. Building a realtor recommendation tool with data science’s help tells about a great technology that is being made. I wanted to know if I could find a home being foreclosed using this.

    1. Cybera doesn’t maintain any IP once we have helped an organization bridge into data science. I’d recommend you reach out to Keylo directly at as they’d be able to provide you with a more thorough answer.

Leave a Comment

Your email address will not be published.