What is predictive lead scoring, and how can it help you land more real estate listings?

And one last question I forgot to answer -- territories are divided (at the smallest level) by Census block. From there, you'd move up to Block Group, then Tract and so on. 

https://en.wikipedia.org/wiki/Census_block

Sorry for the delay in response, Tom. 

When the computer model predicts turnover in your area, they do so for past years first -- this is called back-testing. It allows the model to (at a very fast speed!) understand the market conditions of your area, and what common local seller signals are. The model tries over and over to get the most accurate predictions for past years before it is "allowed" to predict the turnover and likely sellers for the upcoming year. Then, we continue to monitor our predictions against actual turnover throughout the year (and this metric is exposed within the product so you can see it too). 

The size of territories vary greatly, but most are at least 1,000 homes and we rank the "top 20%" or 200 homes as the ones most likely to sell. Some agents have 10,000 homes in their territories, and have a large number of team members or agents following up on leads. You can really choose the size of your farm, but in order to have the most success, we recommend a farm of at least 1,000 homes total.

Hi Jim, 

We actually aren't associated with Zip Realty -- we just share part of our name :)

SmartZip is a predictive analytics company located in the Bay Area. And we agree that predictions and big data have the power to change real estate!

Check us out here: www.smartzip.com

Hi Mark, 

We do have predictions for nearly every neighborhood in the U.S. if you're interested. No pressure but we do have a great number of clients who are working in Idaho we could put you in touch with :)

www.smartzip.com 

Such a good point! You can definitely use personal scoring methods in your everyday practice to help uncover the motivations of your leads. And, those scoring methods can help you determine your follow up!

It may be that some leads you have to walk away from entirely because they're a bad fit. Or in others, you may determine that they'll be ready to sell in 6 months so you can keep in touch regularly as they prep their home to sell. 

Yes, Olga, most of the farms we are able to create for people are at least 1,000 homes -- but you don't need 1,000 leads to work from. The beauty of lead scoring is that it can help you identify the top 20% of seller prospects from a larger group, and focus on them. So it's a lot like traditional farming -- but instead of waiting for the "right people" to identify themselves, the predictive scoring identifies those people for you upfront.

Yes James (Jim) Lawson, DBA i spent the money and got nothing. Never again will I ever part with money without the ability to test and prove first. My leads are far far better.

Great question, Deb. Obviously, data isn't perfect -- and the accuracy of the output (ranked predictions) is tied to how good the original data was.

I can say that in my experience, this is where having a professional algorithm standing behind you is important. The computer model can scrub the outlier data points (like when someone a 1 million dollar home is logged as a 100k home) so the predictions are working from the data that best matches the seller triggers in a given area. 

It's frankly pretty complicated, which is I think why so few people are able to incorporate oredictive lead scoring into their business now. 

I definitely agree with you that deep insights (and sometimes anecdotal evidence -- like knowing that a neighbor's father is an agent) is required for having the best predictive scores. By combining your personal knowledge with what the data says, I think you'd have a killer ranked list to work from.

Thanks so much, Saul! I really appreciate your kind words. I can't wait to hear how lead scoring helps you change your mindset and business!

I agree that predictions could be considered "leadguesstimates" so it's important to have a way for the predicted sellers in your area to show their selling intent. At SmartZip, we have a marketing platform with lead gen pages that convert potential sellers... so the "predictions" become real people who know your brand and who are able to respond to your marketing. 

Hopefully, by starting the dialogue with an offer of value, you would be able to ensure a fair commission with these sellers once they convert from leads.

Prafal, sorry to hear that the program didn't work for you. I'm not sure what you mean by "additional features" -- do you  mean other products? 

We arrive at most-likely-to-sell owners by running predictive algorithms that assess the market conditions and local seller signals for your area in years past. By forcing the models to work from past years first, we can back test their predictions against real-life turnover to ensure they are predicting home sales at a high rate of accuracy. Then, we tweak the models to reflect next year's predicted market conditions, and output next year's predicted sellers. We continue to measure our predictions against real-life turnover throughout a customer's term of service.

Hi Tom Bailey -- yes we do have a lot of agents whose territory is comprised of mostly vacation/second homeowners. Because the marketing can properly target the primary address of the owner -- rather than the renter -- we see agents with a lot of success getting listings from the vacation homeowners. In many cases, they have never heard from an agent about their rental/second home because no one knows how to get ahold of them. 

As for the variance in values in small areas, our algorithms are pretty dang smart. So if it's clear that waterfront homes are selling for nearly a half million more than ones off the water, we would likely be able to pick up on that. No pressure at all, but if you'd like to see firsthand how well the home values in your area (down to that waterfront valuation) look, we'd be happy to show you. You can ping me here or request a demo here: 

www.smartzip.com/demo