How Can Estate Agencies Benefit From Artificial Intelligence

As the world becomes increasingly interested in the use of AI, one area stands out where AI is likely to have only minimal impact Direct Sales. Why? Because Sales are about relationships, personalities, persistence, working to overcome objections, empathy, creating rapport and most importantly the human touch.

Companies such as Insidesales.com offer interesting new tools, which use AI to help guide sales teams, providing them with insight on who to target and predicting pipeline. There has also been increadible leaps forwards thanks to LLMs and their use within chatbots but I struggle to see them closing significant high-value sales.

No matter the hype, AI is nowhere near real human intelligence. AI can take huge swathes of data and through training infer patterns. This can be useful in diagnosing medical conditions, finding relevant legal precedence, identifying security risks or recognizing customer buying patterns. AI can also be used to replicate some innately human acts, such as conversations, recognizing faces or matching a previous experience to a current situation.

So given these fundamental limitations how could the role of an estate agency benefit from the use of AI?

Traditional high street agencies have faced numerous challenges from the self-managed online services. According to a 2017 report a fifth of all agencies are at risk of bankruptcy and profits would seem to be declining. Recent statistics indicate that slightly less than 1% of all house sales in the UK are via online agents, however, online sales are growing by about 20% per quarter. Can AI help traditional agencies to redress the balance?

I interviewed Steven Schiller, Manager of North London Agency Preston Bennett, part of Countrywide, the UK’s largest estate agency. My aim was to see what were his pain points and, as an AI consultant, where could I see AI might help him to be more effective.

VALUATIONS

Initially, we looked at valuations. Agents use local knowledge, comparative property prices, square footage calculations, condition, location and a general feeling for where the property would fit within their current portfolio. AI can perform this element of the role very effectively by virtue of being able to handle so much more data concurrently.

A ‘general feeling’ is actually the human brain’s version of deep learning. We have all the data stored and use our neural network to assimilate the patterns but at a semi-conscious level. We are simply unaware of the huge amount of data and calculations being performed by the brain but sense the results as gut instincts.

In the case of valuations, we can merge the gut instincts of 1000s of agents into a single system by combining more data than any single person could ever want to review or learn.

An example of this is Houseprice.ai — they claim that their “Root Mean Square Error (RMSE) model is an industry leading 2.6%, meaning our middle most projected price is just 2.6% from the actual sales price”.

However, one element is still missing, the seller’s desire to get more for the property than any true or realistic valuation could offer. So, while AI can deliver the data, it will still be up to the Agent to convince the seller of the validity of the valuation and this becomes even more pertinent when the valuation is created by a black box system, which takes in a variety of information and throws-out an apparently random number.

MATCHING APPLICANTS TO PROPERTIES

The next area of interest for me was the process of matching applicants to properties. Agents rely on 3 mechanisms to achieve this, their existing relationships, a hit and hope approach of showing everything to everyone or their existing CRM and matching systems.

Given the innate pattern matching capabilities of neural networks this is an area where AI can really help. For example a study in the US set up a competition between an AI and local agents of matching applicants to properties with the AI winning.

One area that this could really benefit is new housing developments, where the need to match thousands of applicants to thousands of new properties can be incredibly time consuming, while working on relatively low commissions.

RELATIONSHIPS

The one word that came up more in my interview than any other was ‘relationship’. Anyone who has ever achieved any significant sales will know just how important their relationships are to making the sale or even finding the opportunity. Top sales people are hired based on their LinkedIn Contacts. Making that call to and old customer, a lunch meeting or round of golf are things that no AI is likely to replace for many years. Has anyone ever sat down for dinner with Alexa?

Estate Agents are sales people, they sell their services to sellers and sell properties to buyers. And like all sales people building relationships is their differentiator. We all know the common criticisms that surround agents, however, good agents transcend these cliches to make allegiances and relationships that close deals.

However, Amazon has proved that you don’t need a salesman to sell. And so the online agencies, such as PurpleBricks, are proving that you don’t need an agent to sell a property. High Street agents need to fight back. By using AI they may find a way to add value and reduce costs. Through better and automated applicant to property matching and accurate and realistic valuations, while retaining their innate non-replicable skills of being able to sell and build relationships.