Inman

Realtor.com’s AI integration is a case of ‘garbage in, garbage out’

Editor’s note: Realtor.com responded in the comments section below to this contributor’s critique. 

Everyone wants in on the AI goldrush, but in the race to monetize the new technology, are they making better tools or simply distracting toys?

In April, Realtor.com launched a new AI-enhanced way of searching for a home that would seem like a match made in heaven. Home searchers have not seen innovation in the home searching experience in decades. 

Called “Dream Home,” the tool promises to help buyers by matching information from the MLS with user inputs. As with many things AI generates, the reality is less impressive than the premise.

A flawed home search experience

The home search experience has many difficulties. For one, home searchers often don’t know the complexities of data entry when it comes to adding a home to the MLS. They know what they want but might not know how it’s classified in the MLS. 

This problem is exacerbated when homebuyers move from one state to another and don’t know common amenities found in homes. For example, home buyers who are relocating from Texas have to learn what a full-daylight, walk-out basement is. 

On top of all that, the data is entered by humans hoping for more exposure for a listing who might “unintentionally” add features to a real estate listing that are not technically accurate. If you have a bedroom on the main level and a bathroom, is that a master on the main feature? No, but the real estate agent thinks there might be a buyer who wants an in-law suite, and a full bath and bedroom could work, right?

If that isn’t enough, much of the MLS is left up to manual entry. Neighborhoods get misspelled, square footage is either misentered or flat-out not entered, and the hotter the market the more likely information has been left off. 

Enter AI

This is where a competent AI could really help with the home search. AI excels in taking large amounts of data and helping it make sense. An AI can sift through terabytes of data and find just the right thing you’re looking for. 

However, the same flaw found in home searching is found when using this tool. There’s an old adage around computing: “Garbage in, garbage out.” That seems to be the case in this execution. Instead of relying on any stable data set, we have the MLS, and you can only imagine the output. 

If you’ve ever tried to craft a prompt for an AI, then you know there’s a bit of a learning curve to get the desired result. Even after that, there’s a less than 100 percent success rate on simple prompts.

Every output has the potential to be “technically” correct but just a bit off.

Realtor.com’s AI

Realtor.com’s AI is exactly what you’d expect if someone from the C-suite made a decision to push an AI out the door. That’s more of a reflection on the decision-making processes at Realtor.com and less about the engineers tasked with creating this. 

The Good

The AI takes MLS data and attempts to show you what you’re looking for. In other words, it purports to find your IRL “dream home” based on the wishlist items you input. 

The Bad

It’s not really a search at all. It often misinterprets your input. It doesn’t handle the price range well. I put in a search in trying to look for homes in my area and it told me not to use landmarks. At best, it’s a random luxury home generator, a toy. 

Hot take

An MVP-level AI execution is being fed a bad overall data set that is best at generating a delightful distraction while looking at luxury homes. However, it’s months (many months from the look of it) from being able to deliver a definitely different search experience than the flawed one we have now.

I may sound like I’m a bit harsh on Realtor.com, and I am. I’m biased. This is a brand that’s been mismanaged for decades, much to the detriment of the people it represents. Watching the leaders at Realtor.com make technology decisions is a bit like watching our elected officials ask the TikTok CEO how Wi-Fi works.

While I applaud the ingenuity and the courage to bring this to the market, there’s also likely a reason you haven’t seen this particular toy on every screen.

You can test it out here: realtor.com/ai

Am I wrong? Too harsh? On point? Let me know in the comments.

Joshua Jarvis is a digital marketer with 4rd Marketing in Atlanta, Georgia. Connect with him on LinkedIn or Facebook.