I remember the day like it was yesterday. My first college class as a wide-eyed freshman at the University of Washington. Econ 101. I walked into this cavernous auditorium with 1,000 other wide-eyed freshmen and found my place. Front row, balcony center, where I sat twice a week for an entire quarter and listened to a renowned professor — whom I never actually met or even saw his face (the big screen with fascinating supply-and-demand overheads was plenty to capture my attention).
Thank goodness for my TA, who helped me understand important concepts that I use today like, uh, the difference between classical and Keynesian economic theory.
Ah, the trusty college teacher’s assistant; that person who adds the human touch to college by helping students navigate the mundane and not-so-mundane concepts taught by professors who are too busy professing to answer questions about what they are professing on.
So get this: Ashok Goel, a professor at Georgia Tech University who teaches an online class in knowledge-based artificial intelligence, had a problem. Goel estimates that the 300 or so students who take this class post roughly 10,000 messages during the quarter in online forums — far too many inquiries for him and his eight TAs to handle.
So, Goel did what most profs with this problem would do. He created Jill Watson, an artificial intelligence bot who became his ninth TA.
After extensive programming and refinement, Jill was answering the students’ questions with 97 percent certainty. When “she” did, the human TAs would upload her responses to the students. As the quarter went on, Jill didn’t need any assistance from the human TAs, and she would write the class directly if she was 97 percent positive her answer was correct.
And the very best part of this story is that these computer science students, who were taking a class in artificial intelligence, did not know that they were talking with a machine. Oh, the irony.
OK, I know, you can fool computer science master’s degree students, but the average homebuyer or seller is way too smart to fall for this trick.
Goel’s “experiment” is another example of why artificial intelligence, at least in this form, is coming to real estate and specifically, the portals. It’s a no-brainer. Here’s why.
1. Better user experience
Under their current business model, the portals have an inherent challenge. They want to deliver an exceptional user experience (as that brings users back), but the current models are reliant on the masses of agents who buy leads (or in Zillow Group’s case, ad impressions) to respond to a user who needs assistance or has a question. And we know that a large percentage of inquiries either go unanswered or the response time is well beyond consumer expectations.
ZG has tried to deal with this through their Premier Agent Assist program, but that solution does not scale well, and ZG is all about scale. Intelligent bots can answer questions instantly, every time, without fail.
An instant response, even one that might be less than optimal in substance, is better than no response at all — especially if it leads the user on a path to get what they need.
2. Better user experience
No, that’s not a typo. One of the challenges with the current portal lead model is that many lead recipients have no clue about the house that the buyer might be inquiring about.
A bot can quickly process vast amounts of data to answer basic questions about a particular property likely far better than an agent who doesn’t even know where the neighborhood is.
3. Better lead segmentation
The portals spit out millions of leads, and presently, there is little done up front to segment leads in any meaningful way. It’s usually when the agent talks with the prospect that they can start to understand if they need to drop everything to sell a home or if this person is someone whose time horizon is the next millennium.
With a basic question-and-answer dialogue between the prospect and the bot, it will be far easier to segment the lead before it ever reaches the agent. Segmented leads create new monetization opportunities for portals, and they provide lead recipients with the types of leads that they want to work. Win-win.
4. The ROI of online leads goes up
Many agents that use online lead generation as a meaningful part of their business have teams. No wonder ZG targets high-performing teams. As lead volume grows, so does the need to add bodies, similar to Professor Goel’s need to add more TAs.
By automating basic question response, lead bots could reduce this need and allow the humans to deal with the types of questions and interactions that are far more likely to result in a client relationship. The result is less cost which drives positive ROI. And the better the ROI, the more the portals can charge.
So, how do I know that the portals are working on this? I don’t. But I do know that they are always looking to improve the user experience and make more money, and well-designed bots can help with both. And, Spencer Rascoff’s informal Twitter poll supports the notion that ZG is at least noodling on it.
Realtor.com might have a bit more difficulty moving on this as NAR might not think that bots are in line with the motto of “keeping the Realtor in the center of the transaction.” But I don’t doubt its new data science exec is all over this.
On that note, I am not suggesting that lead bots are the beginning of the end for Realtors. Just like I don’t think all college TAs are going away. Only the bad TAs. The ones that don’t respond quickly (or at all), don’t answer basic questions well, are unable to communicate well, don’t keep up with the subject matter or are rude or ill-tempered.
With Jill Watson, Professor Goel may not need eight human TAs. But he still needs the good ones for the human-to-human interaction that is an essential part of the learning process.
The analogy for real estate is unmistakable.
Formerly Senior Vice President of Industry Relations at realtor.com, Russ Cofano is an industry consultant and speaker with 25 years of executive-level experience in technology, association, MLS, brokerage and law. You can find him at Cofano Consulting or LinkedIn.
