Getting Ahead Of the Demand Shift: How to Win In the Future of Selling

More than half of brands report they’re struggling to satisfy consumer demand for home delivery and buy-online-pickup-in-store — and, of course, most live in fear that a single failure to meet fulfillment expectations could lead to the loss of a customer for life. In the recent “Anticipatory Analytics Help Retailers Adapt to the New Normal” webinar, RIS explored the hottest tech trend helping aid today’s retailers in managing demand, promotions, fulfillment, assortment and more. As part of this, we examined how forward-thinking retailers can get started on their journey to advanced analytics success and the potential benefits in terms of improved availability, margin and lifetime consumer value. Read on for an edited version of the transcript and all of the presentation slides, including comprehensive research from Forrester Research on the future of selling. Tim Denman: Welcome to “Anticipatory Analytics Help Retailers Adapt to the New Normal” webinar, hosted by RIS News and presented in collaboration with Antuit.AI. I am Tim Denmon, editor in chief of RIS, and I thank you for joining us today. With me today to explore these topics as well as Forrester's latest research on the footwear and apparel industry and the shifting shopper trends disrupting this segment in the face of COVID-19 are Forrester's vice president principal analyst serving application development and delivery professionals George Lawrie, and Antuit.AI's COO AI merchandising and marketing Yogesh Kulkarni. George and Yogesh thanks for taking some time today. Yogesh, why don't you get us started? Yogesh Kulkarni: Thank you Tim. I just wanted to take a moment to actually quickly introduce Antuit.AI to our audience. Antuit is actually an AI software solutions company, very fortunate to work with some very leading large retail brands especially in the apparel footwear our department store and specialty. What we are really known for is injecting AI in the end to end merchandising lifecycle for how and how much to buy, how to distribute your product, how to shape the demand using prices, and how to personalize customer experiences. We are extremely well known for our concept, what we call unified demand signal, which is one demand signal helps you power a lot of your inventory decisions such as allocation fulfillment and pricing. We also actually have success with treating inventory management as unified topic. Again we've had a lot of success thanks to many of you, we have won a lot of panels in this space, and as you know we are owned by Goldman Sachs. So with that in mind, I just wanted to quickly give a backdrop to how we came about to conduct this particular study and its findings. As we worked with retailers, especially through pandemic, I think one of our retailer clients actually aptly said that retail has changed in the last six months more so than it has in past several years. We saw a seesaw of various factors. We saw a lot of people actually in nonessential retail have declined sales, e-commerce went through the roof, demand and inventory management have always been a challenge, but it particularly became a huge challenge as customers actually looked at multiple fulfillment options. We all experience that with our retail partners, and what we wanted to understand as a company and for everybody's benefit was as you actually come out of pandemic and get into a new normal, how can artificial intelligence, advanced analytics and anticipatory analytics help you make that transition? And to that end we sort of commissioned a survey with Forrester, and we have some very insightful findings that we are about to share with you. So with that in mind, I'd like to turn this over to George Lawrie from Forrester so we can discuss the findings. George Lawrie: Great, thanks Yogesh. Yes, what we're thinking about here is how can you anticipate what the future of selling is going to be like? You heard Yogesh talking about the challenge of a unified model of demand and a unified view of inventory and how challenging that is, especially when you've got a much broader assortment than ever you had before, and you've got many more locations from which you have to fulfill. And actually, exactly as Yogesh said, the pandemic has really challenged people, and it's shown up the let's say weaknesses they have in their systems. So to do this work, to do this study, what we did was to look just at United States. We looked at retail but we also looked at manufacturers or brands — I should say more correctly around apparel and footwear — and we looked at companies with a certain minimum size of revenue and really some quite large ones as well. So the industries were retail and they were apparel manufacturers. And the sort of respondents we asked what we really wanted to do was to make sure we were talking to people who had responsibility for understanding inventory and allocation. So you can see here that a significant number of them were C-level executives, quite a number were vice presidents, and some were directors as well. So this is the project we undertook to identify these people and to try and understand from them what their view was around let's say what's happening to the business shifting. We know, of course, all of us know that people's behavior has changed during the pandemic, and you heard Yogesh say that maybe that change will be permanent, and it's put enormous demands really on retailers who sometimes feel perhaps a little bit like this package in the middle of the road, unable to react to something coming down the track very fast towards them. And, of course, for a long time uncertainty has been you can say to supply chain professionals certainly no surprise. This is the wonderfully named World Uncertainty Index which is run by Sanford and by the Economist Magazine. And you can see that because we have such a closely interconnected supply chains, particularly it has to be said in apparel and footwear, that any kind of event that might happen, anything that's happening in the world, world events, changes in currency, perhaps the introduction of tariffs, maybe relaxation of tariffs, maybe even quotas we have that in Europe back in 2005 right in the middle of season a huge quota put on textiles and footwear coming from China. All that kind of uncertainty, it tests our systems again and again. So we have talked about the pandemic, but it's not really just the pandemic, it's trying to build a system that will enable us to be able to anticipate to sense and respond to all these types of changes, because there will be another one along shortly, no doubt about that. And in our own research here at Forrester we think a lot about how you might anticipate collectively what's going to happen to demand. Of course, people have used historical data, time serious data, but more and more and Yogesh really hinted at this they're using artificial intelligence and machine learning but artificial intelligence to look at predictive variables as well and to try and use those to position strategic infantry and perhaps maybe even spare capacity, maybe some on shore capacity and certainly on shore logistics to enable them to anticipate and position their inventory in such a way that they'll be able to respond better to changes in demand. We're also seeing people build, I would say, event pipelines as well that might help them to understand in case of one of my clients they're looking at the reinfection rate, the R coefficient. But lots of people have been looking at perhaps the potential for a strike or a natural disaster or perhaps a vendor becoming insolvent. So all those kind of event pipelines of also feeding their sense, their anticipate their sense, and their respond capabilities. So these are the kinds of things we look at regularly in our research and it's why we're particularly interested in this idea of anticipatory analytics. So we think about this capability to take data and to use it to drive an insight-based alignment of both the assortment breadth, and you heard me talking about the increasing breadth of assortment we've seen in all kinds of things but particularly in apparel and footwear, much shorter private life cycles for those things as well. But also the depth, so what sizes and how many of these to have location by location to meet this anticipated demand. And also to understand let's say what the propensity of different segments of our customer audience might be to fulfill in store or to fill perhaps online or even to respond to promotions or markdowns. So all this has to be driven by a deep and analytical understanding of customer and customer segment propensity location by location channel by channel. And that is really what anticipatory analytics is all around. So in our survey, we ask people to rank what they thought might happen and what would be most challenging in terms of their adaptation to COVID and a post-COVID world. And you can see here quite clearly, the brands don't expect customers to return to yesterday's business as usual. You heard Yogesh already say that we've condensed a lot of change both as businesses but also as customers or consumers in the past few months, we've changed our behavior significantly. And you can see that that's creating quite a challenge really for businesses to overcome as they go forward. But interestingly, we also asked them how prepared they felt in terms of responding to this shift in customer demand. And you can see here that by and large they're not well prepared for it. So they anticipate that these changes will continue, they're not well prepared for it, and that also they feel that this will result in some long-term business challenges. So you can see here that we heard about perhaps they might be out of stock, what might then happen. So we asked them what would happen if they couldn't fulfill a customer's order, particularly if a customer comes into the store because that key item is not in stock. And you can see here how many of them felt that, yes, of course they'd lose that sale, but it might well impair the value of that customer to them over their lifetime. So you can see that people certainly in our sample the people that we surveyed were very well aware that this is not just a one-off; it's something they have to be able to adjust to over time, and it has a very significant impact for their business in the long term. So to adopt this kind of anticipatory analytics approach the brands really need to focus on overcoming some barriers to get there. So they know that they need to do something about it, but when we asked them about the kind of challenges that they have — we asked them here about different aspects in their business strategy and asked them to say what did they think really would be the biggest challenges for them? And you can see here that they're anticipating these huge changes in demand and they're anticipating significant challenges around that. So when we ask them, you can see here that they were concerned, for example, about being able to predict what demand there might be by location but also by channel, remember how important that is. And to be able to be sure that they could meet the fulfillment requirements, which are now I would say rather more demanding than they used to be. So you can see here some significant challenges or worries that they have. And also that these are both issues that they have to worry about right now. So we asked them about the challenges they have in finding the data that they need, but also in finding the analytics they need in order to respond immediately and in the long-term to these strategic challenges that they have. And as you can see, some significant challenges for them there. Kulkarni: George? Lawrie: Yes, please go ahead Yogesh. Kulkarni: On that note, on the previous slide, we've always seen retailers actually use last year as a foundation to making next year's decisions. And as you see up there, a lot of people don't have trust on their data rightfully so, and they're really in an unprecedented situation on how they plan forward right? If you see a lot of them are actually talking about they have challenges in anticipating how COVID-19 impacts their demand. They actually know that last year's data is not good enough. They don't know how to forecast the demand. There seems to be a fundamental challenge in actually using the data and analytics to actually move forward. Lawrie: I've got to say, Yogesh, it's one of the most common questions that I get from our readers and our clients. They've been calling me over the past few months, saying our historical data now doesn't make sense because it's been disrupted either by panic buying, but more usually in the case of apparel by people not buying anything at all because they just postponing or perhaps they're sitting at home wearing their pajamas and they don't need any dress clothes anymore or dress shoes, they're just wearing their slippers. So I had a lot of people say historic data is now horribly disturbed, what can I do about it? So you're exactly right. But I've also had a very interesting conversation with quite a lot of our retail clients about, is historic data really as useful anyway as it used to be? Even the short product lifecycle of many apparel and footwear items now, actually how much does that history really help? Is there something else we should be looking at? Should there be other variables other than just this history? And I think you and I, Yogesh, have probably had this conversation that people often when you ask them, “What do you think you'll sell next Tuesday?” They look at the same Tuesday last year, and say something probably a bit like that. The difficulty is that people and people's behavior is changing such a lot that that kind of that's a historical view is less and less dependable even without the COVID crisis. Let's press onto the next slide. And see what they thought might be the impact of this. So it's pretty interesting to hear to see that the top thing they worry about actually is the customer lifetime value. Notice also that they're thinking about what this might do to the balance sheets as well. So you might end up, and this generally speaking happens when it's difficult to match your inventory to demand, you always end up with the wrong thing in stock in the wrong place and that means your inventory turn is depressed. And for most retailers, that's the main thing on the balance sheet so that really impacts their return to shareholders. So you can see they're thinking yeah, it's not just about I would say the obvious things like losing a sale and obviously losing the margin on it so losing the revenue, losing the margin. But they're thinking quite a lot about the long-term impact of that. And I have had this actually in some surveys we do directly with consumers that we've seen some of them are saying we remember who is able to anticipate our demand, we remember who was able to keep continuity of supply during this crisis and we intend to reward them with more of our business in future. So some situations there I would say around the difficulty in anticipating those shifts in demand and the impact that has, not just on the top line but also on the bottom line and also on their inventory turnover which as I say is very significant for retailers. So what we're seeing is people asking or thinking at any rate about leveraging these anticipatory analytics to unlock short and long-term business benefits. And what we mean by anticipatory, well the word really actually was originally used and I'm sure many of you will think about this, was used by Amazon. The idea of anticipatory shipping. It's about collecting data that will help us to know in advance where to position that inventory in our supply chain, which locations which stores people might pick it up from where we might need to ship it from. And that really is I think pretty important if we want to get ahead of what's happening demand to try and understand to anticipate what their demand might be and to position the inventory there. And not just let's say inventory in depth but the right range as well to understand, particularly what colors and what sizes should be in which locations, as well as which styles. Brands are beginning to do that and helping their outlets do it as well because they have sometimes a wider view than the outlets or retailers do, so brands in particular trying to do this on behalf. We ask them, when we ask the brands about adopting tools to boost their capabilities, and we ask them about the kind of timeline they had in mind, when would they be doing this. And you can see a significant number of them already well on the way to doing this. So I think the takeaway from this is that yes it's perceived as a challenge, yes it's very concerning in terms of financial results and financial impact, customer lifetime value and loyalty, but actually people are already do understand what they need to do and they're beginning to take their steps towards it. So when we ask them about the different aspects of responding to these challenges and we ask them which are the top priorities for them and which they worry about most, you can see here that for the most part what they're thinking about is, yes, data but also tools. Notice also that they're thinking about the people and skills here. I think this is very significant that people need to understand that these tools are here that seem to complement or amplify their judgment. And in the past, I think all of us have been involved in projects where there's been perhaps some resistance maybe because people haven't been well aligned or their bonuses have not been well aligned and they haven't understood actually how much this will help them to benefit all of them in fact. So what are we seeing in terms of results from this? We're seeing already that they're beginning to manage demand better, they've used it to anticipate and managing their promotions their allocation their assortment better and also their fulfillment. And not too surprisingly because they're beginning now to apply this kind of science I would say some machine learning, but also using AI to incorporate the experience people have, they've been doing this for a long time and this is a very well-trodden path particularly around markdown to use people's experience to help to understand when to take the first and the best cut. But all of that is beginning, at least they're anticipating it, making some impact on these worrying things the challenges that they were facing around their top line and their bottom line and also their inventory turnover. So I think the thing that really came over to us out of all this data and this study was that in order to put the foundation in place, you've got to have some metrics or some measures which will help to, let's say, anticipate changes in demand and disruptions in supply and to give sufficient notice at the right level of granularity for the system and the people who are working the system to take corrective actions. So I talked about the idea of anticipating perhaps an increase in the COVID infection rate or perhaps the illiquidity of one of your suppliers. So all those kind of things are important to establish those metrics but also to implement technology. We saw in the previous slides that for many of them they've got to outpace their peers in selecting and implementing the best technologies here. There is a kind of an arms race around machine learning and AI to do this well. So I would say probably most people that I talked to think how can they more quickly get this actionable insight that's critical to them adapting to the volatile markets they have? But also very importantly to retrain their people. So to make their people more I'd say not data illiterate is perhaps a bit harsh way of saying it, but to understand how the tools can scale their judgment and their capability and to understand what these kind of insights mean to them and how they can act on them. So with that I'd like really to hand back to Yogesh because I think maybe you have had some questions already ready. Kulkarni: I think this is great. Thank you, George, for that insightful results as you rightly concluded, as selling in the new age actually requires foresight not just hindsight and I think that's extremely important. So one of the things that we have seen in the past is the retail organizations have always been focused if you look at how they've structured their organizations, how they've structured their operations. They're focused a lot on what I would call is product and store demand. So a lot of them actually think about what products they're going to sell in what seasons, how much, and where are they going to sell it to their stores? They organize a lot of their operations around stores. But e-commerce demand surge continues along, doesn't it? Now you have to start to think a lot about who the customer is, how will they buy, what's the buying occasion? So how should actually retailers think about their shift in thought process from product and location specific demand to more of a customer-specific view? Lawrie: It's a great question Yogesh. It's quite funny because people have been talking about customer-centric retailing for at least I would say 10 years when people first started talking about a customer-centric assortment. And of course, to some extent that was the job of the merchant. The merchant would say that he or she knew their customer and knew what they would buy, and they were using their intuition and perhaps some data to put together an assortment that was compelling for what they thought of as their customer. The trouble is that their customer isn't really attached to a store anymore. The customers don't really care about stores. It's very harsh to say this. But they do care about the brand. That is very important to them. Then, as you correctly said, it's about the occasion. So there are occasions and I think we all know this when we need to fulfill today. So I'm going to a party or maybe I'm going to a funeral more sadly, so I need a new suit or I need a new tie or I need new shoes but I need them really today because I'm going to do that tomorrow. And people are so trained now I would say by their use of the internet to go search for things, that what they want to do is to see local availability in my size and my color. Can I reserve it and pick it up today? Now, that is a huge logistic problem to be able to anticipate that, to place that inventory somewhere close. That's very, very hard. But it's not really a store problem, it's a problem which is about understanding the center of gravity of demand and the center of gravity for my customer audience and my customer segment. So we're seeing a lot of people doing a lot more work I would say at the beginning of the process to think about the customer, to think about customer segments that we're serving to think about the occasions on which we serve them, to think about their propensity then to fulfill on which occasions will they wait a week to have it sent to them? On which occasions do they need it right now? And to use that to try and put together the assortment but also the allocation and to allocate it into the locations that will meet each of these different kind of occasions. Does that answer the question Yogesh? Kulkarni: I think it does, and I think to the point that you made several minutes ago that customer centricity is key and I think we saw a lot of retailers, actually you expressed surprised on the metric, but a lot of retailers are actually extremely worried about customer and customer lifetime values for the first time much more than inventory and inventory profitability alone, right? So I think that's exactly how I think people should think about it. George, I have sort of a follow-on question to it. And I know you probably look at a number of retailers. Do you see, in the recent past we've also seen a number of retailers that haven't been able to do well through this pandemic. Are you seeing some, in your view, some difference in approach when it comes to anticipatory analytics between winners and others? Lawrie: Well, it's a good question because I think we all know that people like grocers and home improvement did well. It's partly because what they're selling was in the case of grocers essential, in the case of home improvement well when people are at home they want to do home improvement stuff. But the other thing behind this was that those segments of retail were already very well organized in terms of the level of data they had at the item level so they had all the attributes already there. Now, what I have seen is that some types of apparel and footwear brands have also done pretty well by thinking about this. The ones that were well prepared in terms of yes data about the items, but also data about their customer. So people who have had good, I would say, understanding of their customer segments have done well on this, and I think we all know that people who have been in the outdoor business have done fantastically out of this. Why? Because people are locked at home, they've all got cabin fever, and they're thinking about when can I go hiking again? When can I go camping? When can I go fishing? And some of them are buying in anticipation of the time when they can. And, well, none of us can fly anywhere, so perhaps there's going to be a lot more vacationing close to home and a lot more of this kind of outdoor activity. So I've seen people do that. But the real, I would say the common denominator, is that they have great data about their customers but really more important the intersection between the customer, the occasion, and the merchandiser they have to the item attributes. Does that answer the question, Yogesh? Kulkarni: Yeah it does, thanks a lot for that insight George. I will turn this back over to Tim to see if there is any select audience questions that we need to actually present to George. Denman: Absolutely, thanks Yogesh and thanks George. It's a fascinating presentation and great conversation. We do have a few minutes left, and we have some questions from the audience. After the current scenario, referring to COVID-19, don't you think there should be a new baseline for predictive analytics instead of relying on historical data? Lawrie: If I was to answer that, I'd say absolutely, Tim. And whoever said that, that's great. It's a bit like companies have been thinking about doing zero-based budgeting. I think there will be a new baseline, yes, because I don't know if the history, and the history has always been a little bit misleading and distorting because I actually, if you look at sales, you're not seeing people maybe bought a substitute instead of this thing they really wanted. You don't see the losses, you don't see the people who are going to store and went no they don't have it. So I think, yeah, we need to rebase. And I'm seeing more and more retailers who are saying what I'd like to do is to rebase on something else that I know about the customer, so stuff that I know but maybe stuff that the market knows as well. So I've seen people say I'd like to look at search engine arguments, I'd like to look at more general propensity of people to do things so where can I find this more, let's say industrywide or market-wide data. So I'm seeing a lot of appetite and interest for that to do exactly the rebasing that you're talking about. Kulkarni: George, to add to your great points here, I think baselining is also about actually looking at, at least some of the partners we worked with have is actually understanding and adapting to new normal. As an example, we have seen increasing trend at city center or downtown stores as an example have lost traffic, so now you are not going to have the same level of sales. We're seeing an uptick or return to normal on some suburban stores where you can follow social distancing. There's different types of products that maybe considered potential, like clothing for children as an example. They have started to return back to normal whereas dresses may not because you're no longer going into the offices right? But I think all of these factors actually are in place and have to be taken into account when you start to look at establishing a new base. And to your point, I think over reliance on history and last year has to go away. You have to use more and more of changing signals or emerging trends to react to the new normal. Lawrie: Predictive signals. I love what you said, by the way, about downtown and out of town or suburban because it's so interesting. Previously there was actually quite a sort of move to, say, do we even need out of town stores? I don't know if you saw this, but a couple of years ago there was a lot of data came out saying people aren't learning to drive. Even in the U.S., youngsters aren't getting their driving permit. It's been that way in Europe for a while. But when I noticed that data for the U.S. I thought what does that mean for out of town stores? And almost immediately afterwards I noticed, not apparel but IKEA said we might deemphasize the out of town stores and do much more city center so people just see what they want, it's just like a showroom and then they have it delivered to them. So, interesting point there. How do you manage the social distancing there? Kulkarni: Exactly. Denman: Great, we have a question here about demand, which is obviously a big topic here. What are the sources of enterprise data to derive universal demand signals? And a follow-up, what is the dependency on the freshness of that data? Lawrie: I could just say a couple of words about this. Typically retailers always looking at point of sale data even by the half hour. And, of course, they're looking at click stream all the time. And most of them are combining the click stream and the point of sale to try and understand something from that. And many of them of course are sharing it with the brands. But increasingly brands themselves are let's say forging a direct relationship with consumers and they're using that, I didn't show the slide in here but we collect this data a lot and we know that people actually quite want to have a direct relationship with the brand maybe buy directly from them, or be directed from the brand's website to the outlet or the stock where they can find it. So there is, I mean that puts a lot of challenge on let's say understanding the local availability of that inventory. And I've worked certainly with some people in let's say the athletic shoe world around how do we feed that availability so when people go into let's say Google Shopping, they might be arsing about on their mobile device, they think ah I'm going to pick up a new pair of trainers how do they know which store has their size and which color? There's a lot of work going into that. But of course they won't have it unless they anticipated they were going to be there. So you need that anticipatory element to it. Yogesh, maybe you have a view on this too. Kulkarni: Yeah I think you've covered it. I think what we're seeing is actually like you said, a lot of people would start with the POS data as the holy grail. A lot of them are starting to combine a lot of online activity data, merging the two to actually understand the demand a lot better. They're trying to understand as an example, if someone actually came in within a ZIP code and is starting to order a certain type of product, what does that actually mean for how much inventory I carry in that store within that ZIP code? No longer am I actually looking at this in silos in terms of online demand versus store demand, right? So the combination of the two is becoming more powerful. And like your IT said, I think to take a step further wherever customer data is available if retailers actually have customer data you start to see a pattern in what they want so you can start to anticipate that when you actually go into your next buying season to decide on what you're going to buy. Because today cash is so scarce that you have to put the right kind of investment, everybody is actually buying down less than they used to right? I think the confluence of POS data, the online data, and the customer data is becoming more and more common. Lawrie: Yeah. Sound point about the working capital as well that you need to conserve it. That was actually the first thing we found with most people we were working with, they were saying actually during this crisis we've got to keep our customers with us but we've got to keep our cash with us too. Denman: Yeah, certainly cash is king. For our parts of our conversation, data is king. I heard a question here about obviously COVID-19 has injected a whole bunch of outlier data into data sets. How will this impact the change of retail analytics going forward with a year or maybe two years worth of what we consider outlier data? Lawrie: Of course, you know there are algorithms to help you determine what's an outlier and what isn't. And I've had this conversation lots of times with people. They've gone, can we just you know, eliminate these outliers? And actually generally speaking if you've got good systems they will help you determine what's a significant piece of data and what isn't a significant piece of data. I think the people who are struggling are people who have really quite simple weighted moving average … sales. Because depending on what your weighting is, it could take quite a long time for that outlier to work its way through the system. Interestingly, one person I work with not in apparel but in grocery, one of the things that they did was just to bump up safety stock for everything. They just changed the safety stock basically in their core systems just to make sure they always had plenty of everything. They said we aren't going to worry about having too much. But of course now they have to move all that out again. So I think the thing is yes there will be lots of outlier data, but I think whoever said we ought to start rebasing, I think that's, if we do rebase perhaps we'll ignore that outlier data as we go forward. Sorry Yogesh, you probably had a better answer on that did you? Kulkarni: No, no, I think you actually said it the right way. I mean, I come from a world where data is valuable and you don't throw out any data as much as you can. So what I would say is, look, simply say if you had a product that was selling 100 units last year, it's selling less than 50 units this year, you want to use last year's data to understand the trend of when it sells and how much it sells. But you also want to know that this year the demand is down by 50%. And I think a lot of smart users are using both data sets; of course, current data is a lot noisier. So you have to actually be careful in throwing out things that are noisy. But then you still have to use the current use data. I think this is where probably a lot of newer machine learning techniques help understand how to mix the two or how to give more emphasis or influence to certain factors in the data more than others. But people are using all kind of data cautiously. Lawrie: Yeah. And the thing about machine learning is it can help you to weigh different signals in different ways. But perhaps also help you to understand, to get some understanding. Because those weird, you're quite right actually, the outliers are still telling you something. It's a weird reading but what is it telling you? There's something in there. The difficulty is if it's very extreme is to, you don't want it to affect everything else. But there is, you're right, there is some intelligence in there. You probably shouldn't just throw it away. Denman: I think we have time for at least one more question, maybe two. We are running out of time, we have a few minutes left. This research obviously was focused on apparel. What's to become of the apparel segment? Are the shakeouts over and is this trend going towards subscription boxes and that kind of stuff? Is there going to be some long-term stickiness there? Lawrie: That's a really good question. I actually think that the problem with apparel is that people enjoy the shopping experience I think. I think we'll go back to an experience based differentiation. There's some things they hate. We know they hate the changing rooms. We know one of the reasons that they tend to do so many returns is because they buy all the sizes online and try them on at home. And in fact I have a friend who has teenage daughters and they have kind of parties doing this where they buy everything, try it all on in the week and then send it back, which of course is really troublesome for the apparel brands and for the apparel retailers. For the cost of managing all that returned inventory they could perhaps have better changing rooms. But I sort of think the future of all of this will be something a little bit perhaps like Disney where there's a cast around you, and you or your child at Disney is the center of this experience. And all the transactions are happening from people in character in your story. So I anticipate in future there will be a kind of blended digital or physical experience in which you as your snowflake self will be the center of everything. And I think the store will be part of that. I don't think it's going to be just online. Kulkarni: I agree with what you said. I think brands, actually, unfortunately, nonessential retail, has difficulty maintaining a continuous relationship. Think about how many times if we go to a store or think of a brand when we buy our apparel or footwear. So being relevant in the customer's mind is difficult. So they have to actually think up engagement morals. I think Tim's question was the description right? I know Nike and others have actually started experimenting with some of those. But those are already signs of actually how do I improve the engagement with the customer? A lot of these brands have started to talk about, talking of Nike again, things like actually enticing people with their apps. Trying to connect other people within the community and trying to understand their fitness goals versus your fitness goals, then creating a brand experience just as what George talked about, right? I think that's where things are going. A store is just a part of the ecosystem of a broader relationship. I think the challenge is how do we build and nurture that relationship in a world that has so much clutter with all the brands trying to do and compete for the same attention? Lawrie: It's not about selling stuff, because the customer thinks they can sit at home with their iPad and buy stuff. So it's got to be about exactly what you said, the experience. So Nike's whole thing is it's not the shoe, it's the running experience. Actually, if you're a retailer, you have to provide something even more, so you've got to provide parties, lessons, something like that, a community, community content. So the customer is deep, deep bound into them as an experience. You're exactly right about that. Denman: Thanks George and Yogesh. We are running out of time, so if there are any other questions or anyone wants to get in touch with George or Yogesh, you can reach out to them directly or you can reach out to RIS and we'll put everyone in contact. But for now, thanks George and Yogesh for spending some time with us, and a big thank you for Antuit.AI for sponsoring the webinar. For now, thank you everyone for listening, and we'll talk to you again.