AI, Retail Real Estate, Value: 6 Questions

The Bezestein Bazaar, El Khan Khalil, Cairo, 1872, by John Frederick Lewis. Watercolour. The Higgins Art Gallery & Museum, Bedford

The Bezestein Bazaar, El Khan Khalil, Cairo, 1872, by John Frederick Lewis. Watercolour. The Higgins Art Gallery & Museum, Bedford

Recently there has been much talk of a ‘Retail Apocalypse’, with millions of square feet of retail space shuttering and the demise of many long term household names. The US and UK have been particularly heavily affected but the impact has been felt widely across the developed world.

Just across the US we are now seeing over 100,000,000 sq ft of retail space closing per annum.


And online sales have been, for years, growing much faster than offline:


Though still the overall penetration is not that high and varies markedly across countries.


Looking at all of this positively many real estate people point out that the majority of people still prefer to shop, in shops. A recent US survey results are shown below.


Further comforting the real estate industry is the sight of many previously ‘online only’ Brands opening physical stores. So we may have an industry closing a lot of space but ultimately it seems like ‘online only’ does not work and shoppers, by and large, still want to go to physical stores. Or that is the way the real estate industry likes to look at the market.

What we will discover over the next few years though is that they are both right, and very wrong. Yes, people will always like the experience of shopping but the current experience of shopping, with notable exceptions, does not have product/market fit.

A rising tide will not lift all boats, it will expose the naked.

Fundamentally, and this is the problem for real estate investors, not all retail, or retailers, are equal, and the difference in capabilities of operators is going to become more and more a key driver of returns. As with the office market, where one could previously ignore the particularities of occupiers and focus purely on the NOI, today and going forward the operator of that asset will be a major determinant of performance. Retail Real Estate is going to become an asset class where understanding the dynamics of the retail industry, at a very granular level, will pay of handsomely. 

Real Estate Investment as a numbers game is coming to an end. Our industry is morphing from being about selling a Product to delivering a Service. And that changes everything. You’re no longer buying into an asset class, you are buying into an ecosystem of variables, most of which are not real estate related, that combined, will determine short, medium and long term value.

Where though does AI sit in all of this? 

Below are a series of questions I put together for a panel discussing the impact of AI on the valuation of retail real estate.

1. We can see how AI changes things for retailers but how does it change things for shoppers?

The best retailers have been utilising AI for some time. The recent report from the MIT Sloan Business School ‘Artificial intelligence in Business Gets Real’ shows how the early adopters are doubling down on their AI investments, building competencies, and working to take AI to scale.

Principally retailers are working on understanding their customers at a deep level so that they can:

  • Make personalised product recommendations

  • Tailor the inventory in individual stores to better reflect the wants, need and desires of the local population

  • Optimise pricing

  • Deliver more targeted offers and deals

  • Reduce the incidences of fraud.

All of the above is of course aimed at delivering a more compelling shopping experience for customers, by better understanding the individuals drivers of consumption.

Mostly this data harvesting and analysis is being performed online, which is why the best online retailers (think Amazon or Alibaba) are so successful. They understand what I want, and make it extremely easy for me to purchase goods, any time of the day.

Providing customers with what they know they want is a very hard thing for anyone to do better than an advanced e-commerce company. 

What online is not good at though is ‘discovery’, the serendipitous finding of something I didn't know I want. And generating that frisson of pleasure when this occurs.

Which is why offline shopping will persist, albeit with a huge caveat.

And that is that the offline retailer has to make it a more enjoyable experience for me to get up off my sofa, leave my house, and travel to visit their store. And then, when I am there, know enough about me that this experience is tailored to me. Yes I know, this is all sounding very ‘me, me, me’ but the reality is that is what offline retail is all about. Delighting me, and whoever I choose to go shopping with.

Much of all this personalisation is made possible by carefully collecting the right data, with the requisite ‘Volume, Variety, Velocity and Veracity’ (the famous four V’s of data) to be able to answer the right questions that provide the answers to deliver this personalisation.

So how does AI change things for shoppers? By upping their expectations as to what a shopping experience is.

It was recently said, by the ‘Retail Prophet', Doug Stephens, in response to a question about millennials being fickle:

‘Millennials don’t suffer from shortened attention spans. Rather, they simply have a much higher sensitivity to things that are boring.’

That’s how AI has changed things for shoppers.

2. Amazon Go is pioneering AI powered automated stores - is this going to be a big thing?

You’ve probably seen the promotional video Amazon released, showing how the experience of shopping in one of their Go stores was different. You simply walk in, tap your phone to begin shopping, then pick up whatever you want, and simply Go.

Frictionless retail. Just pick it up and off you go. How easy it that?

Essentially this is only possible because of the incredible advances in the AI behind ‘Computer Vision’, which is the ability of computers to understand photos, videos or the world around them. Today, a computer can ‘see’ better than a human. This is the same technology behind many of the autonomous vehicles you hear about, and the ‘self driving’ capabilities of a Tesla.

In computing terms, this is now largely a solved problem. The technology will improve (it is mostly a function of computing power and the availability of training data) so these prototype small Go stores are very likely to grow in size over the next five years. They will be a big thing. Autonomous shopping will be a big thing.

Paradoxically the aim is not solely to replace humans in store. In fact new prototypes include open kitchens, where you can see your food being freshly prepared, right in front of you.

Remove the boring bits of shopping to concentrate on the product and service.

It is speculative at the moment whether or not Amazon will release this technology as a ‘Software as a Service’ product but if they do, close attention needs to be paid to who picks up on these new capabilities, as they will almost certainly be the most innovative retailers in the market. Possibly, even probably, they will enable a new range of entrants.

Critical from a real estate perspective will be whether your asset is equipped to enable a retailer to ‘go autonomous’. Could it be that, in five years time, not having such infrastructure will be like having an office building with poor broadband? i.e effectively useless.

3. Alibaba talk a lot about ‘New’ Retail, especially connecting offline with online shopping. And are huge investors in AI. How does this impact on real estate?

Alibaba is prodigiously good at AI. They are also prodigiously good at e-commerce. The two are connected.

80% of all e-commerce in China touches one or more Alibaba entity, as they control the 3 largest marketplaces in the world, the B2B (, C2C (Taobao), and the B2C (Tmall). 

Simply put, this enables them to have almost perfect knowledge about the nature of demand across China, in real time.

They have opened 65 of their Hema branded supermarkets in the last year. They own 29 department stores and 17 shopping malls across the country. And through their ‘New Retail’ platform they are helping to digitise hundreds of thousands (yes, hundreds of thousands) of ‘Mum and Pop’ stores.

Jack Ma explained what he means by ’New Retail’ in a shareholder letter in 2017:

 “E-commerce is rapidly evolving into New Retail. The boundary between offline and online commerce disappears as we focus on fulfilling the personalised needs of each customer.” 

The key here is that not only does 80% of Chinese e-commerce pass through Alibaba, but a large percentage of the merchants also run Alibaba software in their stores. So the two way insight, between demand and supply, is something new within retail, and immensely powerful

How does this impact real estate? 

First, as above, it is vital that any retail asset that wishes to plug into this sort of ecosystem (Alibaba is not the only one, just the largest) is technically capable of doing so. 

Secondly, investing in ‘New Retail’ locations should mean investing in areas where retailers are likely to be more successful, as the level of data and AI powered analytics should enable better matching of demand and supply.

And thirdly, hunting out locations where ‘New Retail’ does not yet exist but is likely to arrive in the near future should offer good returns as the quality of the retail experience will improve as these new technologies and analytical capabilities are put into practice. Asset values should rise accordingly.

4. New retailers like b8ta - who say their mission is ‘Retail designed for discovery’ - provide a very different in store experience. What role does AI have in their business?

What actually is the point of a store today? What will it be in five, ten years?

Most likely, it will not be about product distribution. You used to have to go to a shop because that was where the goods were. Today you do not; they can get to you in a myriad of ways. Delivered to your home, or office, even to your car boot. Delivery has largely been solved. Sure, we still have issues of deliveries when you are not at home but one way or another, that will be resolved in the near future. In a drone delivery world, they would only ever deliver when you were in because the system would know, you are in.

Amazon in particular are spending vast sums trying to solve this ‘last mile’ problem, so it will be solved.

So what then is a store for? Prefaced like that it is obvious isn’t it? A store is for ‘discovery’ and fun. Finding out about things you didn't know, or more about things that you did. All wrapped up in a human-centred environment that is enjoyable and a pleasure to experience.

So how does AI help with that?

Enter the likes of b8ta. They started a store where they work directly with Brands, and act as a marketing platform for them. Each Brands products are displayed in dedicated areas, with Brand trained staff on hand to demonstrate and inform. Sort of like the Apple Store but for multiple vendors.

They are beautifully designed stores dedicated to presenting a Brands products in the best possible way. You cannot buy anything in store, the whole point is nurturing and positioning each Brand.

In the background every movement and preference of every customer is recorded; what Brands they looked at, if they looked at Brand X did they look at Brand Y, what questions did they ask, what did they ask for that was not there, and so on. All of this data is then analysed by AI and fed back to the Brands, together with suggestions as to how to refine, optimise and improve on every aspect of the experience.

This is the software industries ‘Build, Measure, Learn’ for stores.

Since launch in 2015 they have opened 78 stores across the US.

The clever bit though, is that they realised that designing these types of environments was complicated, so they started ‘Built by b8ta’ and now offer their software and systems ‘as a service’ to third parties.

This pay monthly solution includes checkout, inventory, point of sale, inventory management, staff scheduling services and more.

There are large Brands that can do all of this on their own, but there is also a long tail of companies where this is beyond their capability to deliver but nevertheless could benefit greatly from having access to it.

This is analogous to the office market, where the workplace is becoming a too difficult problem for many companies, who are then ‘low hanging fruit’ targets for Flex space operators to sell to.

AI is a great enabler, but it is hard. All companies can benefit from AI powered services but most likely, they will buy into a complete, turnkey solution. Who owns the solution could be sitting on a great deal of value. 

The OneMarket spinoff from Westfield is essentially mining the same seam of need. Off the shelf, managed, highly sophisticated AI powered services. No capex, no long delivery cycle, just go.

5. In the future will we all have personal AI’s that do our shopping for us? That understand our needs, wants and desires so well they just order what we want before we even ask for it? Do we even need shops in that world?

There is a meme going around the internet that a future Amazon service will be one where they send you two boxes each month. One is full of all the things that they ‘think’ you need or want, and the other is empty.

You simply keep what you want and put the rest in the empty box.

Pre-emptive shopping. They ‘know' you, so just give you what you want. And if it is your husband's birthday they will also give you a selection of presents that they ‘know’ he would like.

Now tell me that that is not a service that a large percentage of people would buy in to. Sure, there is also a large percentage of people for whom it is the devils work and they would hate it. But that is the point.

In an AI powered world we will start to know the difference between people and how they want to shop. Our job is then to give each party exactly what they want.

6. Thinking about the role of technology and AI what characteristics of a retailer should we pay attention to? How can we know who needs what type of space, and will they be able to pay the rent?

The starting point from a real estate perspective is to get to grips with what types of retailer one has in one’s portfolio and to make a judgement as to whether they are providers to group 1 or group 2 in the answer above? And then ask, are they setup for their market?

Knowing how well the retailers in your portfolio are ready, willing and able to make the most of this coming world is a key risk factor. Some will not be, in which case they are a bad bet, whereas some others are fully aligned and should be aided and abetted in their efforts in whatever way the owner of their stores can help.

AI is becoming more and more central to the operations of retailers, either handled by themselves, or provided by ‘SaaS’ providers like b8ta.

I would contend that where these types of retailers, and technology providers, operate is where the successful retail of the future will be. Online or offline is now irrelevant, everything is one. Online behaviour feeds offline and vice versa. And much of this data needs to come from the physical environments these companies operate in, be it a single store, or a giant shopping centre. The physical environment has to be able to be analysed, and has to be responsive. Data will increasingly inform how we setup, manage and optimise our physical spaces. They have to be equipped to enable this.

In summary then, AI is profoundly impacting the retail market because it is changing the capabilities of retailers to provide better products, services and experiences. And that is raising the bar for everyone. Owners of physical assets need to be fully au fait with the new needs the best retailers will have. Ideally they will work in partnership, as the often confrontational nature of landlord/tenant relations is damaging the ability to satisfy the end user.

The bottom line is that shoppers do not need shops to go shopping. They need to be made to want them. AI can help make that happen. 

Ultimately human + machine wins!