Real Estate AI Chatbot Case Study

Real Estate AI Chatbot Case Study

Imperial Properties is a luxury real estate agency in Mallorca. They came to us because they wanted to increase conversions on their website, and we recommended our AI chatbot to them.

The idea was that by adding an AI chatbot to their website, users would be given an extra service, resulting in more trust and more people willingly leaving their email address or phone numbers with them.

The primary goal of the AI chatbot was to increase conversions and make it easier to collect leads

Of course, most people won't simply give you their phone number and email address if you ask them, unless you provide services to them first, resulting in increased trust. We fixed this by ingesting "related general information" into the chatbot, allowing it to answer general questions related to the real estate market in Mallorca.

Below is an example of how the chatbot answers "related questions", resulting in that their website becomes a destination for information related to the real estate market, increasing trust amongst its users, and hopefully over time increasing traffic.

Real Estate AI chatbot answers questions about tax

Recommending properties

In addition to asking questions related to the different areas in Mallorca, and some basic facts, it is also a very good interface for searching for properties. In the screenshot below I am testing it out with a query that would be a typical query a family asks it when looking for an apartment in Mallorca.

Real Estate AI chatbot shows me a 3 bedroom apartment

Notice how the chatbot displays images, while also returning a link to the prospect, allowing the user to have additional information about the property. The AI chatbot will by default return 3 images for each property, and if you click the image, it will maximize, allowing you to see more details.

The chatbot is also configured to use a language that guides the user in all aspects related to renting or purchasing properties. However, the primary purpose was to have an "alternative search interface" providing answers using natural language instead of simply showing links, to increase trust and therefor increase conversions.

Implementation

The core trait Imperial Properties wanted for the chatbot to feature was that it should do its best to connect its users to the real estate agency, and do its best to encourage its users to leave their email address or phone numbers, such that a human being could contact them and try to provide them with properties.

Imperial Properties wanted the AI chatbot to be primarily a lead generator. By providing additional services, such as being able to answer questions about real estate regulations, the cities and towns in Mallorca, this increases trust, resulting in more leads and potential buyers.

Full Screen AI Chatbot

One detail Imperial wanted for the AI chatbot was for it to be a "full screen" AI chatbot, implying it should be one of the most prominent user interface widgets for users to navigate the website.

Most of our clients wants to have the AI chatbot as a button on the page, similarly to how we have a button in the bottom/right corner of this page. However, for the real estate market, displaying the AI chatbot by default, at the top of the page, almost like a "search interface", might make more sense.

Real estate customers don't typically "browse" your website, they start out by searching for what they're looking for. Displaying the AI chatbot at the top of the page the way Imperial does, reflects such usage better, and effectively turns it into an "AI search for properties type of user interface".

Priority Properties

Another detail Imperial needed was for the AI chatbot to prioritise some of their properties, in particular the properties they were exclusively selling or renting out, that no other real estate agents had in their portfolio. This was easily accomplished by using our "Search Postfix" feature, that allows some properties to "bubble upwards" during search.

This was a feature we originally implemented to have the AI chatbot prioritise best selling products, but became handy as we made the chatbot prioritise specific properties that Imperial wanted to feature higher up on their search results.

Collecting Phone Numbers

Having the AI chatbot being able to collect phone numbers was also important. This is not a feature we have out of the box, but something we rapidly implemented for Imperial Properties.

You can read more about how to collect phone numbers in the above link. However in theory, we can collect any type of information using logic similarly to how we implemented the phone collection feature above.

Live Data

Imperial had a backend system that would export properties in an XML file format. This allowed us to implement it such that it could deal with real time information. It works by importing this XML file once every 24 hours, and re-creating the RAG database, allowing it to always have the latest properties and information.

This allows the user to get information only about properties that are actually available, making sure the AI chatbot never suggests properties that are not in their portfolio.

Conclusion

The project took us about 4 weeks from start to finish, and even though we'll need to monitor the chatbot for some weeks to sanity check its responses, already we can see the chatbot providing users with high quality information, that are relevant to their requests.

We worked together with their software developers responsible for having delivered their CMS and backend system, and in total we had 4 orchestration meetings before we were able to deliver a working product in production.

The project implementation phase went fairly smooth, and although there are some features we intend to add to it in the future, already we're all quite happy with how it has proceeded so far. If you want to try it out for yourself, you can find a link below.

Thomas Hansen

Thomas Hansen I am the CEO and Founder of AINIRO.IO, Ltd. I am a software developer with more than 25 years of experience. I write about Machine Learning, AI, and how to help organizations adopt said technologies. You can follow me on LinkedIn if you want to read more of what I write.

Published 22. Feb 2024