The Retail Podcast by Vue.ai: Leaders in Retail | Ashwini Asokan & Julia Dietmar28 min read

Akshara: Hey everyone! How’s it going? Welcome to The Retail Podcast by Vue.ai, our first ever series on Leaders in Retail. We are so excited to have you listen and I’m Akshara your host for today. Just until 12 months ago, we were in the middle of the great retail apocalypse. A disastrous year that leads over 9 huge retail bankruptcies in 2017. JC Penney, Radio Shack, and Sears each announced more than a hundred store closures. There were also so many other apparel companies whose stocks hit new multi-year lows including Urban Outfitters, American Eagle, and Ralph Lauren.

So what’s really going on with retail? Is Amazon eating retail? Is it because e-commerce has blown up and 60% of global e-commerce unicorns are from China? Have consumption patterns of shoppers evolved so much that brands can’t keep up? The success of startups like Casper, Bonobos and Warby Parker has really forced physical store retailers to offer similar deals and convenience online.

Brands that have been successful in 2018 know that it’s not just about selling products anymore, it’s actually about selling personalized experiences. And with labels like Stitch-Fix raking in millions of dollars using data science and personalization, we know one thing for sure – the algorithms have arrived.

And whether it’s a physical store or online channel consumers are now placing a lot of value on experiences. The retail value chain involves a series of change makers coming together to enable businesses to sell their products to customers. Each of these actions in the chain really brings a portion of the value to the entire process, but who is responsible for designing these retail experiences? We are bringing you the answer.

Our podcast features the true champions of retail. We are interviewing CEOs, CTOs, heads of product, merchandising, heads of innovation, investors and so many more people who are part of this massive retail value chain. Today, we are in conversation with Ashwini Asokan and Julia Dietmar, the Founder-CEO and Chief Product Officer of Vue.ai®by Mad Street Den.

Akshara: Ashwini, there has clearly been a rapid technological revolution in fashion, right? Brands are now so focused on personalizing experiences in a way that they are enabling shoppers to rebuild products and customize them to the very last detail. What do you think is going on with retail right now?

Ashwini: I think something fascinating is going on in the retail industry, especially this year, 2018. Just like you said I think 2017… it seemed like retail had hit a complete rock bottom. But this year, I think, it got kicked-off with the IPO with Stitch-Fix. Just this last quarter, Lulu Lemon’s shares just went soaring and they say that greater than 50% of that growth is actually attributed to e-commerce sales. There was just a piece last week in the New York Times about how old-school retail stores are seeing sales stronger than they have seen in years. I think any time you go out there in the market this sentiment for the most part of this year has been absolutely positive. And I think, you know, it’s a sign. It’s a sign of things to come for retail. The industry did hit rock bottom. It’s been interesting to watch what has happened in 2018… a lot of these brands, the industry as a whole has started to figure out how to pick itself up and move on.

There is no doubt I think, we just absolutely have to give credits where it’s due and Amazon has significantly changed the way consumers shop today, than anything we have seen in recent past. Then there are social media, which is the second biggest thing I say right after Amazon. Kudos to Instagram and Snapchat and what these social platforms are doing to how people buy; I think both these things have been a significant change in the industry. Thanks to these two big moments that we have seen or to the introduction of technology into this otherwise conservative, slightly more traditional industry.

Akshara: Ashwini, it’s interesting that you are talking about a change in this typically traditional industry so, in terms of the future, what kind of trends are you seeing in retail or things that you actually feel will blow up?

Ashwini: The first is that with retail… it’s been very interesting to see how while the retail industry has been a bit traditional a bit conservative, fashion is always been ahead of the curve when it comes to adopting the technology, experimenting. Whether it’s on the runway, off the runway it’s been fascinating to see all the kinds of new technology they adopt, but what’s been missing in this, are experiments that scale. The industry has been experimenting for a very long time – whether its sustainability or moving past, issues with body types, culture, race, it’s always been the center of conversation on things that are cutting it from a cultural perspective from a societal perspective. But if you think about the context of the technology itself I think, the industry has been experimenting for a very long time with AR, VR, new types of experimental store, but none of these experiments really scaled. My bet for the next five years is actually on starting to see experiments that scale and I think that is the first biggest shift that we are gonna see in retail.

It’s not just embracing things that or technologies that just allow for different types of show and tell marketing. Embracing technology that allows changing the consumer experience in-store, changing the consumer experience online, connecting those experiences across different channels, improving operationally when you are going up against a mammoth a giant like Amazon. You got to know what you’re going up against. Growth is just one side of the story. I think the other side of the story is competing with the kind of economies of scale that somebody like Amazon has, which means you got to start thinking about things like operations. How do you make the supply chain more efficient etc.? How do you make production more efficient? How do you begin to use data in a way that your operational efficiency just sores? And I think, these are all things we are going to start seeing experiments that scale across different aspects and different parts of the value chain.

Akshara: Right, that’s so true! But where does personalization as a concept feature when it’s across multiple channels?

Ashwini: I think in the last eight to ten years we saw all things e-commerce soar, right? And it was almost like a swing from all things offline and in-store to the other extreme which is everything online all the time. What we’re beginning to see now is a correction of that swing from left, from one end to other end and we are beginning to see that retailers across the globe are thinking about this whole online to offline, offline to online, social channels and the role social channels play in this.

I think we have been talking about omnichannel and personalization for long as we have as an industry but I think we have been so far away from anything or anything personalization across channels. I think that is about to change and you already have, you beginning to see signs of it order online, order in store and get it delivered at home, order online and pick it at a warehouse that’s nearby – so many different types of experiments that companies are doing in terms of just bringing that online and offline experience together. I think that is going to be the second trend to watch something that is going to define who survives successfully, who grows, versus the brands that get left out and a part of that story is, of course, is social.

I don’t think you can ignore what the social platforms are going to do in this online to offline story. Last week, there was a new research that was published that stated that until recently all of us kept saying Google is not where people start their shopping search it’s all Amazon. Amazon had overtaken Google as the platform where consumers did their first searches for products. As of last week, that’s now changed to Instagram. Instagram has officially beaten Amazon as the platform where consumers are going to start their searches and I think that’s HUGE! Think about how and in such a short period of time we have gone from Google to Amazon to Instagram. I think online, offline, social channels, email even something interesting. We have been talking to a bunch of retailers who do marketing offline as well, I think we are going to have to start seeing how all these pieces actually come together over the course of the next five years. That is the other space that’s going to look significantly different in the next five years.

Akshara: What would you say AI’s role is in the future of retail, in the bigger scheme of things?

Ashwini: The future of retail is AI. I know that comes across a bit biased given we are an AI company, to begin with, but you know Li Jin of Andreessen Horowitz in her Twitter feed was asked if “algorithms are the new brand”, which I thought was such a powerful question. To say that customer behavior is all changed completely and nothing like it used to be and the name of the game right now is about ease of access, it’s instant gratification, it’s about Kylie Jenner wears something, you post it on Instagram you can’t wait for an entire season before it goes through runway and then gets off to store to wear it. It’s now a thing and it’s not going to be a thing two weeks from now when Rihanna picks… you know… another designer locally off the LA streets for that’s how quick things are changing in the industry and in order to keep up I think AI has a role to play across the board.

Whether it’s giving people just what they want, when they want, just think about the idea of timeline and what the timeline has done to way people surf, to the way people look through stuff they expect things to be automatically curated in that’s where the customer is that is the mindset where the shopper is today. And I think you know no one wants to go to a hundred different sites anymore and which begs the question – what is it that’s going to be important for all these brands as they move forward?

I think the first thing is to get off legacy platforms we have been seeing brands across the globe work with these mammoth platforms from like 20 years ago. And that’s not how it is going to be; companies are going to be agile and really light in their feet and I think the name of the game is relevance. It’s about user intent and I think finding a way for AI to crack and what is it going to take to get a loyal shopper, somebody who believes you in your brand for some reason or for some value. I think to have the right kind of technology platforms that can cash in on that, that can continue to grow that across the different channels is going to be absolutely key.

Akshara: Ya I mean automated processes are definitely a big solution for fashion retailers, but before we get to that what is really the root cause of these problems that need automation in the first place, there is so much talk about the cost of bad data to companies. What is it about inaccurate data that makes it such a massive problem?

Ashwini: The underlying building block of all this which is just is the data that the retail industry is sitting on even. I think, that is probably the most important question of having worth to the retailers for the last few years continuously we keep coming back to this problem – saying yes, we can go and analyze and start looking this data to build interesting products but it’s important to acknowledge that the data is available there is inaccurate. Let’s just take catalog data is an example people across the world is sitting and tagging is sitting and manually tagging all the stuff and there are extensive inaccuracies in just the way you label the products that you have. And as a result, search becomes an issue (personalized search), discovery becomes an issue and this kind of becomes a bit of waterfall effect and when you have data that is not right or that is flawed.

And I think, for us as a computer vision company, this has been a big value proposition just being out there and saying you know it’s important to get past manual-only forms of creating data. That’s the first part of this data problem. A second part of the data problem is actually how many different types of people or vendors are creating the same data across the different value chain. You have got retail work with so many different types of technology vendors, technology vendors merchants all kind of partners their process and everybody is creating data, so no two departments are looking at the same consumer in the same way. Everybody has a very different view about the customers because everybody is producing their own data. So again, how any given department looks at that data, is also extremely flawed and especially we are talking about online to offline to social channel. We are talking about this omnichannel strategy in personalization. It’s important that everybody is getting a singular view of the customer as opposed to how broken it is today. And then, the last but not the least, is actually sharing of that data since everybody is busy creating their own data, working with their own vendors, having their own perspective of what’s going on both in the market as well as with their shoppers there is a big issue. Just them not sharing data – like buying is not looking at marketing data, marketing is not looking at merchandising data, merchandising is not looking at the seller data. And you think that before somebody goes in and decides the next season’s clothing or before the marketing team puts out a certain set of discounts even whatever kind of campaigns, they have that data instantly accessible to them for decision making at any given point of time. So I think the biggest set of problems that I believe algorithms can solve or AI can solve, is actually the best framed in the context of a data problem.

Akshara: And Julia what do you think of the biggest inefficiencies in the retail value chain there is process inefficiencies and overall operational inefficiencies but is it just that or does it depend on the architecture of the company or the technology they are investing in or not investing in can you tell well about that.

Julia: Traditionally, there has been a team that is responsible for a channel and they are creating an own set of data for distribution to heir channel separately sometimes not even collaborating with their partners or throughout their wholesale department so what that results in is not necessarily inaccuracies in data but a lot of inconsistencies and looking from the other angle so when the data comes to a retailer from different channels how could those inconsistencies be solved and taking into example something as simple as color – one single color can be described in many different ways, it can be orange or it can tangerine sunset who searches for how do your customers actually look for products with those type of attributes so that those are the very very simple issues with data that can be easily solved especially right now with AI and image recognition technologies. Also, going off the traditional organizational setups because the same chunks of data exist across organizations in different formats, different databases, different systems, however you wanna look at it a lot of times its very difficult to bring this all together and create that unified view of your product, unified view of your customers so that is something is not very easy to solve its a problem that is not of your sexy problem no one really wants to spend time looking at that but that is something absolutely sensual for retail to get to the next level.

Akshara: From a retail respective what do you think some of the specific problems that AI could solve?

Julia: So now, with technology that is available, we can solve inefficiencies like labor force. Just think about the example – there is a company called Bossa Nova that developed this amazing robot that walks through the aisles of a grocery store and scans the shelves for restocking and the signals are immediately sent to the back room if it has determined that there is no product on the shelf. It’s a lot more efficient than deploying an army of workers who are, who may or not know these things. And of course, you cannot sell what you don’t have on your shelf so that’s like they lost revenue right there and easily solve problems with technology. Another problem specifically which is just kind of close to what we are doing here with fashion is being able to do something with technology that you are not even able to do today. Just imagine, if you are a brand or a retailer that sells clothing in various sizes and obviously we know and it’s been proven through A/B tests, all kinds of A/B tests, that showing pieces of clothing on a figure increases your conversion rates significantly. So, of course, you need to photograph everything on models and that is very expensive. So if you do all dresses from 0 to 24, at most you can afford to photograph the most regular size and maybe the smallest plus size. But most customers are somewhere in between. Not being able to visualize how that garment looks in my size, not necessarily on me but on a model of a standard size, is something that would increase intend to purchase trifold and without the technology cost prohibitive to use something like that.

Akshara: Julia, can you tell us something about how personalization has evolved as a retail concept?

Julia: Being able to take one step further even with technologies that do exist today, we are kind of tend to say, “okay this is working let’s move on to the next thing”. And what’s happening is, we stop optimizing and even if you look at something as simple as a recommendation that is based on the good data, we have gone through kind of traditional statistical models of people looking at this also looking at that for furniture and clothing. Something little bit more efficient is visually similar looking at the dress I am gonna show you things that are more similar. Now, we are actually in the third generation or third iteration of that and that is actually determining customers intent in that particular session and dynamically personalizing for that individual customer, not based on what thousand other shoppers have done but what this particular customer has done in this session. And what she is looking for today and we have just run on A/B tests for one of our largest clients, and we have clearly shown that dynamically personalized product recommendation engine increase conversions by more than 100% over visually similar which are the second generation of recommendations. So, those are kind of the most I would say, lower hanging fruits that we could have right now that could be easily solved with AI and these types of technologies.

Ashwini: One thing that came to my mind is you were talking about just we try and do new these experiments. And then we are already moving onto the next things like the next new technology. Let’s go as opposed to really doubling down on a handful of these and making sure that it’s kind of goes back to that point that I was talking about experiments on the scale. But just think about what people – like Twitter, if you would look at Twitter at any given week, it is filled with jokes of people just saying “I bought a vacuum cleaner on Amazon, I am going to be, you know completely just they are gonna dump vacuum cleaners on me for the next I dunno how many weeks!” and you have got thing following you everywhere and that’s insane! And I think what you are pointing out is just so incredible how people are like oh personalization is such an old thing! Or you know any of this stuff is all over the place. And NO! We are just scratching the surface just there because I think being able to double down on what’s working and pushing the same thing across channels. Imagine if we could take what talking about the dynamic personalization engine the minute I get off your site what if my email is now a continuation of that journey, and if I go to social media it’s a continuation of that journey there and I think just suddenly being reminded, being it’s almost like subtly being way persuasive at that point to across different channels and I think that’s a fabulous point.

Julia: There is one more thing I wanted to bring up that is about the omnichannel everybody talks about omnichannel but no one store or no one brand has really solved it. Just imagine like a very simple scenario that all of us face every day – you go to the clothing store, physical store, you bring a bunch of garments with you in the dressing room you try them on, maybe you like one or two and maybe you buy one or two. So the only thing that brand, that store knows about you, is what you bought and even that information is not always translated to other channels. But how amazing it would be if the store knew what I tried on and didn’t like for whatever reason or maybe liked but found too expensive so they offer me a promotion or while I’m trying it on in the dressing room I could get recommendation of how to style it for various occasions right there somewhere on the screen in the room without even an associate being on my back and knocking on my door.
It would be amazing if this whole experience was brought together. now, of course, I understand there are again I am going back to data I am going back to processes online data offline data is not always connected, the in-store inventory is not always connected to overall inventory. The investments in hardware that needs to be installed in stores are not always justified but I pretty much guarantee if there is one brand or one retailer that actually did that and prove that there was a real value that will have it come very quickly, to all of the stores.

Akshara: So there was a recent article on Forbes that talked about investors heavily investing in AI-based platforms because they look at this profound enabling technology which is cutting across all sectors and AI brand hire than SaaS, Marketplaces and even e-commerce has the technology to invest In why do you think that is.

Ashwini: Every type of new technology goes through that cycle hype cycle. And I think, to be honest for the longest time, I have been saying AI is a technology looking for a problem dissolve. And especially, even in the context of computer vision more so, and as a company that deals, whose computer vision is at the core of our business, we have been asking ourselves this question – why computer vision can something else do this better like is there an easier way to do you know something that AI is already doing or is been built to do? When I think of that question, I don’t think we are at a point as an industry where we can confidently say you know all the answers to what AI can do. And to a large extent, I think the hype cycle typically happens before or just before a product goes on or a piece of technology something begins to scale because once it scales it passed that peak. I would say, in some many ways 2017-18 has been like the year of the peak or the years of peak AI. But that said, I think what is exciting about the technology and the technologies that comprise AI is that an inflection point. People are moving, it kinda goes back to my point about experiments that scale as well I think especially in the retail industry I remember when we started working in 2015-16 with retails, retailers brands across the globe everybody was all about you know image search and visual search and singular piece of like use case that is going to change all things retail. And I remember because we had image searches as a big piece of the beginning of Vue.ai®. One of the things we started realizing was that people are actually not taking a picture of what they see behaviour out and you know looking for it online as many times during the course of the month which was not yet there it was 2015. And as repeatable behavior has something that as a habit it wasn’t yet a habit we instruct people are looking for what Kylie Jenner just wore a poster on her Instagram. This is a bit of nuance they in terms of use case or how people were embracing the need that they where having and then to look back and go okay is integrating image searching simplest way? Which like is putting a camera in the app is that the only way to do it or are there any other type of cases that basically derive off? That looking at how people are behaving so I think that is what’s beginning to happen right now with AI.

That is the most exciting part of what’s going on and I think the investor’s sentiment gaps the investors across the board and the market has been super excited about AI. But I think, you know, everybody is waiting to see like that one or two companies that are really going to take off. I think there’s a bit of that going on as well people are waiting to see who is going to be cutting it and making it there. I think, with that effect, I would say Stitch-Fix has actually done a massive contribution to the industry that you know algorithms-driven personal stylist. That’s the beginning. And I think that in so many ways is the beginning of the story of AI that scales. I think yes that hype cycle the technology is the most kind of promising and the excitement is there. But I think, what’s coming within the next year or two, to be honest, is probably gonna be way more exciting than just the fact technology exist until now.

Julia: Ya absolutely. And towards that, there is a lot of hype in AI because there are the many problems that AI can potentially help solve. Unfortunately, for now, whenever somebody says we are investing in AI very often it doesn’t, it’s not very clear what problem they are trying to solve by investing in AI itself. Without knowing what problem you are trying to solve exists make much sense. AI can help with logistics, AI can help with marketing, AI can help with data, AI can help with merchandising there are lots and lots of different problems just within retail and I’m not even going outside retail. So obviously, there is a lot more in that but what would I see right now is the neural networks, deep learning and all of that advances in the technology that have been happening in the last 5 years now been looking into generative networks that are actually creating something out of nothing, almost creating content. We here as Vue.ai®, are solving or at least starting with their problem of solving cost and efficiencies in model photography, not being able to create as much content as is needed for optimizing conversion. So that’s what we are starting but this is just the beginning we are really really at the very beginning of this what the technology can do.

Akshara: And from a retail perspective what do you think some of the specific problems that AI could solve?

Julia: We have already seen the emergence of influencers that are not real people Shudu, Lil Miquela, and we have seen Whole companies with virtual models yes those virtual models were created by humans all graphic designers and the way they maintained right now is probably very cost inefficient but there is technology that is coming very close and the heels of that idea that will actually allow for just to be created pretty much out of the thin air. So just imagine that those influencers can be every day there could be new influencer for new audience or new body type or new geography or however you wanna imagine this and it’s not just imageries its videos, I can even imagine we could have one day runway shows that not real human models will walk in the runway show it could be holograms or something of that nature, so ya this is super exciting, it’s super exciting because we are at in the beginning and we get to shape the future will look like so ya there is definitely a bright future for AI in fashion at least, ahead of us.

Akshara: I totally agree with you Julia there is so much happening and the retail space is so much excitement. You know, and with the proliferation of pop-up stores, the digital phenomenon, retail is just exploding!

There’s so much going on and before you know there is already this big war between Amazon and Walmart getting ready for their next battle. There is Jet.com’s theme-based shopping experience, there is prime wardrobe and then there are also chat bots that are changing the way people are engaging with brands. So no brand can really do away with technology entirely. In fact there is a report by Forrester called ‘Future of Jobs’ that forecasted by 2022 about 76% of sales task is casual work restocking shelves and inventory control will be done by robot. And it’s not just robotics there is also Shudu Gram, Lil Miquela – they are the newest virtual models of fashion industry and they are not even human! So, it’s all about leveraging technology and giving consumers these amazing personalized experiences and also giving them something that’s relatable.

So going back to what Ashwini said algorithms are the new brand and there is gonna be so many cool amazing speakers that we have on here that will talk about the future of retail and where AI stands in this whole scheme of things with retail. So, watch this space and don’t forget to tune-in to the next episode of The Retail Podcast by Vue.ai : Leaders in Retail. Thanks for joining us today and we hope you had a good time listening to us see you soon. Bye!

You may also want to see more episodes of Leaders in Retail.

The Vue Podcast: Leaders In Retail | Miya Knights
The Vue Podcast: Leaders in Retail | Ravi N. Raj
The Vue Podcast: Leaders in Retail | Anjana Reddy