Proptech Weekly #53 — Are real estate agents AI safe?

JANUARY 11, 2017 | Rayhan Rafiq-Omar

Friends, Feudal Landlords, Commoners,

When Michael Bruce recently announced the results of his firm Purplebricks, much of the presentation focused on their success at generating online enquiries for people looking to sell and buy homes.

It seems fairly obvious that this is the part of agency that generates revenue — finding new customers.

But it’s also the part of agency that’s executed least well, resulting in a fragmented offering — not just in the UK, but worldwide — and has potentially left the door wide open for a new entrant to more reliably win instructions.

And then respected property journalist Damian Wild posed whether there are any professions that are ‘safe from AI’.

Is it much a leap to think that selling a home for many isn’t a ‘people’ business but a data one.

This isn’t a simple problem — there are nuances in establishing trust with the public. Can you sell property, why are you better vs the competition and most importantly does the property owner ‘like’ you. You may snigger, but it’s a bigger factor than ‘competence’ at selling homes.

So, which company can use data to win instructions, price and market property ‘optimally’ and even negotiate and close the sale?

Well obviously there isn’t one even close to existence today… or is there?

Of course there is: Opendoor in the US has been doing exactly this!

They’ve established a way to price property online so that people feel confident enough to hand over the keys and accept a cheque for their home.

This process happens as quickly as 72 hours within receiving the online valuation.

Now obviously it’s early days for Opendoor — they started with just one city and a type of home that was homogenous — all the homes they were pricing and buying looked largely the same.

But they quickly cornered the market in their first city and since expanding to their second city have raised an additional $200m+ to help scale to 10 cities in the US.

I can hear you saying what’s the big deal. Honestly, visit Opendoor.com.

Valuing property with an algorithm — in a way that gives someone confidence to accept the valuation — could be the most profitable business model of all time.

There’ll be an inflection point where they’ve ploughed so my time/money/data into their machine that expanding across the US and potentially across the world would be more rapid than even Uber’s expansion. And we all now accept that Uber is far superior as a service than someone ‘taking the Knowledge’ having to tout for their own business.

Currently Opendoor have a big issue with having to hold stock, fix it up and sell it on themselves.

In the US this is made easier by the existence of Buying Agents, which Opendoor are very happy to pay a 3% commission for — yes, US real estate commissions are still 6%.

But Opendoor ‘buyers’ don’t need to use a buying agent — and I’d bet largely they don’t.

You see, Opendoor’s technology extends to helping buyers as well as sellers: they install electronic locks and security cameras to enable 24 hour, unassisted viewings; they fix up every home to the ‘ Opendoor standard’ and provide not only a warranty but a money back guarantee.

Yes, you heard that right: if you aren’t satisfied after living in the home for a few months, they’ll buy it back.

So this one company helping sellers — current data analysis by Proptech consultant Mike Delprete shows Opendoor is making an average of 5.5% reselling those home, on top of their fee — and transforming the experience of buying property. All without the need to engage with a real estate agent, and largely powered by technology.

There are ‘holes’ in the model — but probably not so big that $200m of funding can’t fix.

They can certainly learn from Nested.com’s ‘we’ll lend you the money if we don’t sell your home in 3 months’ provides for a better balance sheet and less disruption to the existing model of appointing an agent to sell your home.

Opendoor can also massively increase the diversity of stock they can service as they become a larger ‘market participant’ and see more transactional and sentiment data than anybody else.

And of course, most importantly, they can improve their pricing so that they are ‘liked’ more by homeowners looking to sell.

For the time-being agents are going to be losing their jobs more because of legislative changes — stamp taxes and banning of lucrative double charging practices — but there’s a threat on the horizon from what can only be described as the precursor to Skynet.

But, if Artificial Intelligence really does come to realise its potential, surely it has bigger plans than disrupting real estate agency…