When I was a commercial real estate broker, I spent a good deal of my time trying to figure out how to find property owners who wanted to sell. “Isn’t there a place where all these owners gather and hang out?” I thought. Is there an association I could target? What is the best way to find sellers?
The best I could come up with at the time was to simply grind it out. Do the data research, find the ‘true owner’, contact the owner, and rinse and repeat. What amazed me about this process was that it actually worked. However, it always bothered me how so much of my effort was wasted in researching and contacting owners that would never sell or had no need for my services.
Regardless, I was very confident that if I found a building owner who wanted to sell or exchange, I could provide the best services. So I continued with the shotgun approach to marketing to building owners.
Fast forward a few years and I started ProspectNow to make it easier to market to building owners. For the first few years our focus has been around the property and owner contact data. However, I was always still bothered by the wasted marketing effort of contacting owners that would never sell.
Enter Predictive Analytics.
One thing I noticed about trying to find sellers is that there were some logical things I could do with the data to improve my chances; for example, if a property had a loan coming due in the next 12 months. I thought if I contacted these owners it would improve my chances. Another one was owners that own multiple properties. At ProspectNow, we have all kinds of filters to help our users target these properties and we have been sharing these best practices for year…but, how do we know for sure that these things are directly correlated to selling? The answer is that we don’t. It’s logical, and it “feels” like it should correlate, but we don’t know for sure.
Predictive analytics uses math and data science to tell us the answer. About 18 months ago we decided to set out to get the answer using a predictive algorithm. We built a machine learning model that looks at thousands of properties that actually sell each week…then we analyze the properties that have not sold based on the characteristics of the properties that did.
We did not want to launch a product without the confidence that it worked. So we picked a group of properties that our algorithm said were more likely to sell. Then we watched over the next several months. What we found was the properties in the group we picked were ~70% more likely to sell than the random group. Obviously results will vary by region, but our initial result was incredibly encouraging. Since the model is machine learning, we are now seeing in some cases that properties are more than 2x as likely to sell.
Imagine half the direct mail expense or half the number of cold calls. If you are an owner, you are also going to receive fewer marketing communications since you are not a candidate to sell your building. It’s a win for everyone.
Predictive analytics is not new at all. Facebook & Google use this technology to serve the right ads to you. Insidesales.com uses it for lead scoring. Zillow uses it to forecast home prices. The applications are substantial. While ProspectNow’s application of this technology is a first for CRE Tech, I expect there will be many different ways predictive analytics can be useful for our industry.
In the meantime, if we can help our customers cash more commission checks then we have succeeded.
Which Properties Will Be Listed for Sale This Year? Predictive Analytics and the Future of Marketing. Guest Post Steve Wayne.
Connect with Steve:
+1 I met Steve waaaay back in the day. He is quietly killing it!