Sunday
December 17, 2017

Chatbots’ Potential Beyond Lead Generation

      |
-A A +A

Chatbots’ Potential Beyond Lead Generation

Chatbots have a long way to go in real estate, but there are interesting ideas for internal use that you and your agents may find valuable for accountability.
Chatbot conversation on mobile device

Chatbots are on the rise in real estate, promising to work for agents, brokerages, and consumers. I’ve had an opportunity to explore the options, and I’ve come to the realization that chatbots have great potential beyond their current application in real estate, and in ways yet to be implemented.

I recently read an opinion piece by real estate tech guru Mike DelPrete. He referenced several sources, including my August 2016 REALTOR® Magazine article, Can Chatbots Work for Real Estate?, and raised some concerns that current chatbots appear to have more value for agents and brokerages than consumers.

I agree with Mike that so far bots have been less than satisfying for consumer engagement. First, they’re not as smart as they need to be. Second, they don’t really offer anything new to the consumer. Third, they don’t create the trust people have come to expect in an agent.

Redefy started a journey in search of a viable chatbot for real estate lead generation, and we continue to explore new options, but what we have learned is that there are better uses for the bantering bots.

Caught in a whirlpool

We recently had the opportunity to test an artificial intelligence–powered “real estate assistant.” Billed as a way to answer commonly asked questions without using employee or agent resources, we were intrigued about its potential to vet leads for our call center. Imagine fielding hundreds of contacts’ queries simultaneously, impressing them with our techno power and informative answers.

We added the bot to our website and put it to work, eagerly anticipating a flood of leads. The logic looked sound, even epic, but the results were disappointing to say the least. Here’s what happened.

In the first few weeks, the bot had just 18 conversations with real people (although I doubt the intentions of the alias “Poop monster”). Given the hundreds of thousands of users it could have engaged, site visitors simply weren’t interested in chatting about buying or selling a home. The poor volume of engagement was surpassed only by the abysmal quality. The bot continually created an annoying vortex, circling back to the same phrases regardless of what the person asked. Here’s one of the failed “conversations”:

Jo: Marc, can you please explain in further detail the “broker co-op may apply”?

Marc (our bot): Got it, what city are you planning on moving to?

Jo: We are selling in Raleigh…Just want a quick answer to this question.

Marc: Got it, what city are you planning on moving to?

Jo: Please see last comment.

Marc: Got it, what city are you planning on moving to?

Jo: Raleigh

Marc: And how many bedrooms will you want in your new home?

Jo: Is there a real person here?

Well, that was embarrassing. Everything about this “conversation” said artificial and definitely not intelligent. Our call center staff, who received documentation of the conversations, noted that not one person got to the point of giving their contact information (would you?) so they couldn’t salvage the situation by contacting them. Documentation happened after the fact, so there was also no ability to jump in on the conversation.

At the core of the chatbot-consumer interface is language. Given the complexities—from spelling to sentence structure to slang—it’s a huge undertaking. Clearly, the bot we tried didn’t understand our customers. Without a true machine-learning base, bots such as this can’t learn anything new or adapt. Before the stream of comments starts, let me say this: I understand all products are not created equal; I’m only relaying my direct first-hand experience.

We know that consumers don’t want to talk to a bot about real estate. They want to believe there’s a real person involved in the chat that has experience with buying and selling homes in the real world. Unless a bot can pass a Turing test (the programming is human enough to fool an interrogator), it won’t be successful. It’s a great idea, but a lead generating chatbot done poorly can be more damaging to a brand than no technology at all.

My company found greater success with our Alexa home valuation skill. Everyone who uses Amazon’s Echo or Echo Dot knows they’re speaking to a bot. It’s a data and action device, not an adviser, so they’re comfortable asking what their home is worth. Once they get a valuation, they’re given the option to speak with an agent; if they say yes, our capable call center takes over. The response has been tremendous and we’ve put our efforts toward a similar Google Assistant app (soon to be released) instead of chatbot research for lead generation.

Should we give up on B2C chatbots?

Instead of throwing away the idea of a B2C chatbot altogether, especially for non-retail services like real estate, it’s more practical for bot-creating companies to continue development but redirect current capabilities from critical lead generation to customer service. Again, there’s a language issue, but the variables can be narrowed considerably. Bots connected to a brokerage CRM can employ all the existing data to provide real-time information.

Linking chatbots to existing, ongoing services to keep clients informed creates a sense of participation, excitement, and loyalty. Money-saving apps like Chip and Cleo employ AI chatbots to update and celebrate with customers in words and emojis. It encourages interaction, but there’s no expectation of a long, drawn-out conversation that could end in a whirlpool of misunderstanding.

This kind of bot initiates with the customer’s foreknowledge. No one’s trying to fool anyone into thinking a human is making this contact; it’s for information only and the customer deserves to know what to expect. When send out showing information via SMS or automated email, they know it’s not the agent but it’s so convenient they are happy to use it.

Likewise, a service chatbot can keep home sellers informed of a new offer, or a showing, with the ability for the customer to answer certain questions quickly and with minimal language issues.

Bot: A buyer wants to see your house today at 5 p.m. Is that a good time?

Client: Yes

Bot: Great! I’ll schedule it. Please be out of your property by 4:50 p.m. Could this be the one??

OR

Bot: A buyer wants to see your house today at 5 p.m. Is that a good time?

Client: No

Bot: Ok. Is there a better time tonight?

Client: Yes

Bot: What time?

Client: 6 p.m.

Bot: OK, we’ll ask the buyer.

Bot: Yes, 6 p.m. tonight works for the buyer. Please be out of your property at 5:50 p.m. Hope they love it!

The same tech works for home buyers, with search possibilities and more. Success with this kind of a bot comes down to knowing how the customer prefers to interact, so they need to be given a choice. Those who like bots will enjoy it. Those who don’t may not respond, so it has to be just one of several options to communicate with clients. This is another reminder that chatbots are good only if they make the customer’s experience better.

If the conversation progresses to where the customer suddenly asks an off-topic question, your bot may get stuck. This is the danger of a consumer-facing bot. If you suddenly take over with a call or email, the customer is suddenly reminded that the bot is a bot and has failed. The ability to have a function set up for a human user to take over the chat, emojis and all, is a missing critical piece to existing bots, but soon to be corrected, I’m sure.

An indirect lead gen or service bot that could be useful is the chatbot assistant. It could help a “stuck” agent redirect a conversation on the fly. Cogito, used in the healthcare sector, analyzes customer service reps’ conversations and makes suggestions to the rep in a messaging format (such as “you’re talking too fast” or “you’re interrupting”). If the bot hiccups along the way, the consumer will never know.

I haven’t given up on a client-facing bot. I’m just waiting for someone to get it right. There are some promising new developers I’m watching, but I see some better uses for the technology in real estate and beyond.

Thinking beyond B2C

Here’s where chatbots get really interesting. In addition to being Redefy’s CTO, I also oversee operations. Managing assets starts with pulling reports and interpreting the data.

But people, especially contractors and remote operators, can be hard to quantify when acquiring data takes an email, phone call, or in-person conversation. Do I know what our hundreds of agents are doing right this moment? Do I or my regional managers or even my brokerage managers really have time to find out if a particular agent contacted a lead today? Enter the accountability chatbot.

This bot could send reminders that require an answer. Have you contacted John Q. yet? If yes, there’s a trackable data point that’s good for both manager and agent. If no (or no response), ask why (there may be a good reason and they should have the opportunity to explain); the bot could send reminders in a designated time period before the manager is notified to reassign the lead. Dropping the ball in any step of the transaction could be eliminated.

This creates a more efficient and detail-oriented workforce. Decreasing the human capital need (such as entry-level managers) and increasing customer experience provides the best follow-up and execution possible. I always tell our managers I would rather have 10 leads with 10 listings than 100 leads with 10 listings. The latter stat makes me question the ability of our agents, while the first number makes me praise them.

In addition, documentation of these chats creates an indisputable body of evidence. It’s completely unemotional, doesn’t lie, and shows no favoritism. In an era of litigation and claims of discrimination, an accountability bot can be an asset to employer, employee, and independent contractor.

It goes well beyond real estate to almost every sector of business. Did you stock the latest shipment? Have you filed your report? From pilots to retail associates to doctors, chatbots can offer reminders, send new action points, give safety warnings, take feedback, fill out forms, and take on many other roles.

Think of the time no longer wasted tracking down employees to see if they’ve completed a job. Imagine the employee’s joy at not having a manager constantly breathing down their neck. In some ways, it takes stress away from workers worrying they may forget something on the to-do list. On a more serious note, consider accidents that have happened because someone forgot a step in the routine. Perhaps I should stop right here and get on this idea before one of you does!

Chatbots have a long way to go, and probably a longer way in real estate. The cost of programming and marketing such a feature keeps most real estate companies from designing their own. Developers are hindered by access to data beyond public records. Multiple listing services are splintered across the country, with no single point of entry. Add to that the various platform and CRMs that different brokerages use.

Until someone creates a smarter bot with a compatible interface, B2C chatbots in real estate will take a backseat to better uses for AI chats.



Broker-to-Broker is an information network that provides insights and tools with business value through timely articles, videos, Q&As, and sales meeting tips for brokerage owners and managers. Get more Broker-to-Broker content here.

 

1.75
Average: 1.8 (4 votes)
Your rating: None