Tracking The Real-Time Demand Of Digital Consumer (Podcast)

Our Founder & CEO Ganesh Subramanian had a podcast with Jane Singer, Founder of Inside Fashion , New York & Hong Kong, on the need of the hour for fashion brands and retailers, and sensing the real-time demand of digital consumers. Here is the full podcast (32 min) and its transcript below.

Transcript Of The Podcast On Real-Time Demand

Jane Singer:

Knowing what consumers want to buy, and when they want to buy it, is a challenge that all retailers face. For fashion-driven companies, that challenge is even greater since styles changed so quickly. Technology solutions continue to try and solve that problem through a wide range of data-driven applications, but most predictions lack accuracy and tend to be lagging indicators of real-time demand. I’m Jane Singer and welcome to A Seat at the Table. Today we’re speaking with Ganesh Subramanian, and founder of Stylumia, a market intelligence tech company. Using technology to gather and analyze market data is not new, what is new is Stylumia’s approach. The company has developed a proprietary demand ranking system (real-time demand) based on a wide range of traditional and nontraditional signals that it picks up from across the internet. Ganesh, thank you so much for joining us today.

Ganesh Subramanian:

Thanks, Jane for getting me on this podcast. These are very challenging times, but at the same time, these challenging times also provide great opportunities. And at Stylumia, where we are right now, we looked at holistically that how is fashion forecasting and prediction happening today? And this was back in 2015, we looked at and saw that it still continues to be largely a push-driven value chain, where we as experts predict what we think that customers will buy, and make products and push it through the chain. And globally, if I just have to give you a statistic, that in 2018, the world produced 150 billion garments. Out of that, 50 billion garments never sold a single piece, five zero.

Jane Singer:

Really?

Ganesh Subramanian:

Yes. And 50 billion garments sold at discount. Now, if you just look at the magnitude, close to two-thirds of what we make did not connect really with the customers the way we wanted to connect.

Then we looked at the fundamental behind, which means in spite of having all the solutions of today that happened in 2015, and then we said, let’s dissect this problem and say, “What are the various components?” The first component is, what idea gets into the funnel? Which is I call it the “what problem”. Are we putting ideas that have a low probability of winning? How do we ensure that we have a high probability of winning ideas getting into the design from it? Our fundamental insights that currently either it into fashion forecast, or even if it’s data-driven, we see that you supply signals to the right fashion trends. It’s like this, you can collect a lot of data today and say, “What is happening?” Now what is happening is a supply signal. So it’s like going through a mall and say, coming out and say, I saw a lot of red on the windows, which means red doesn’t trend. Now actually, that’s not so. A lot of red also does not sell well.

If you just go by what’s out there, then we are going to have the same era of supply. And we know two-thirds of fashion do not meet the consumer demand the way we wanted. Then how do we solve this problem? Before we said, we developed, what’s called a Demand Science® where we inverted the pyramid and say, “Instead of rethinking what customer want, let’s reverse the pyramid and then say, let’s look at what’s happening in the consumer side of the business. And let’s see what all they are looking for at various levels of the world.” From luxury to the designer, to fashion shows, to retail, to brand, to social media. And that hierarchy of what’s happening in the consumer side, and today consumers are ahead of brands, and we don’t try to catch up.

Ganesh Subramanian:

Now we curate all of this data by collecting them at an internet-scale, and like Google rank pages, we rank products. For the click of a few clicks, one can get what are the winning ideas from my source of inspiration in my country, in the category of choice, in the brand of choice, in the designer of choice, in the price point of choice, and in the time span. Now that’s what we call, we have a solution called a market intelligence solution. We renamed it recently called CIT, Consumer Intelligence Solution (powered by real-time demand science engine). That’s one of them. That solves the “what problem” so that you pick the winning ideas into the funnel.

Now after that, brands, what they do is that designers then put all of these inspirations and come out with new designs. Arguably so, and now the question is whether these new designs will do well? Which of these new designs will do better than others? Which means we have to style grade them. And then you put a bet on each of these styles, put X dollars on each of these styles.

Now the question is, this process also today’s fragmented a fair amount of collective judgment, which gets into it. There are a lot of brands who go to the wholesalers, look, buy or book orders, and then come back and say, “The aggregated orders reflect the demand.” The wholesalers who buy, also don’t sell very well. They also left a lot of stock, which means we are taking a lot of signals which they themselves may not be completely true. So the whole idea is we developed what is called, it’s called Stylumia Apollo. Stylumia Apollo is a prediction engine for predicting the demand for every single new product. And what’s the potential demand of them so that it automatically algorithmically creates all the new products. It can also predict the demand in terms of dollar value or unit volumes for those new products. The data is so much in the market is so dynamic, it’s very, very important that even human is not able to make informed decisions because there are so many moving parts. So that’s the how much problem.

And where we are progressively moving now, as we recently launched a solution called Store.Y, which helps you predictively distribute these products to the right stores by decoding the tastes of each of the stores. If it is offline, as a vendor offline picks up. Now, the whole point is right from getting the right ideas from data at internet scale using consumer intelligence to predicting the right quantity and putting it in the right place, so that’s what we have been solving over the last four years. And one more last thing is, which is in our lab is, how do we use generative adversarial networks to create winning design ideas? So that’s something that is not commercial. So that’s a bit of what has happened so far, Jane.

Jane Singer:

Getting back to the first thing that you were talking about, how are you able to gather this data from such a very large and diversified pool?

Ganesh Subramanian:

Yeah. Great question Jane. Basically, we collect public information across brands and retailers from their eCommerce websites. If it’s influencers, through the influencer handles and platforms like Instagram. And fashion shows from the public available fashion shows images, so after the fashion shows are over. So it’s not a question of just the information and the key thing is that, what do we do with that information? How do we find our demand, which is not public information? Therefore, we’ve developed a proprietary demand ranking engine. You call it a demand sensing engine. What it does is, it spots in all the products in these brands and retailers on a daily basis, collects a lot of demand signals. And our engine, what it does is on a biweekly basis, it ranks all products. For example, let’s say if you want to know winning products from Zara Spain, or H&M or a retailer in Japan, or it’s Nordstorm, or wherever brand. Whichever part of the world, whether it’s Farfetch, you get to know in few clicks. What are the highly probable winning products on these platforms?

Jane Singer:

What would you consider to be a demand signal? What would be an example of that?

Ganesh Subramanian:

See a demand signal is something that which consumers, which works for the consumer, and also it works for the retailer to a large extent, because if it doesn’t work for the consumer to a large extent, also doesn’t work for the retailer. Sometimes it could also be working for the retailer, for example, margin, but not working for the customer. But we pick signals, which are a combination of both. Some signals could be for example the number of views a product is getting, amount of visibility a product is getting. That’s an indication because no retailer will give you high visibility if a product doesn’t have it. And we also look not at one particular point in the life cycle, we look at all of this and time series.

Well, we have a complete evolution of every product over time relative to other products, but that’s the whole objective, this is one of it. We also look at the velocity of stockout, how fast a product is going out of stock, with an indication that the product is selling better. But having said that, you might ask, there are some products where we don’t buy depth, how do you figure out whether their depth is low, how do you still figure out? What we look at is we look at peak performance of the product, not the average performance of the product. So in effect, we try to figure out the velocity of sale of each of these products related to each other. There are other signals like customer ratings and reviews, and there are different signals, which could be different across various brands and retailers across the board.

Jane Singer:

Okay. So you’re collecting a wide range of data, and then ranking it to determine consumer demand or at least what people seem to be reacting to?

Ganesh Subramanian:

Yeah, absolutely. And because we do it across the waterfall, that if there is a client, there is a brand, we actually ask them, “Which are the countries you would like to travel and get inspired by? Which are the brands you look up to?” And if they want to look at what are the opportunity misses, what’s happening in the market right now, what’s working and what’s not working, then you get to see the reality of now. The solution gives you both what’s now, and also what’s going to happen. And you will also get the trend of view of all of this. It’s not about, which are the products we also get into fine details. Recently, we partnered with Pantone, where not only do you get to see winning products, you also get to see winning colors.

Jane Singer:

So how are you doing that? In other words, how are you identifying the winning colors?

Ganesh Subramanian:

Yeah, very interesting. “Winning” is very important. Right now that’s Stylumia’s fundamental key difference, one of the key differentiators. So now that’ve be ranked on the products, now we have said top 10% of products and we have bottom 10% of the products, we have a lot of mediocre. From the winning products, using our computer vision engine, we extract the fashion from all the images first, and then we extract the pixels of colors from those images. We don’t go by text, because in fashion you cannot narrate attributes. A lot of people go by pink as a color. There are so many hues of color, which pink is working well? Absolutely. Therefore, what you do is we segment it, remove the color element from there, and we convert that into an RGB and then we pass on that to Pantone, get the Pantone color with that collaboration. For, we are able to present winning Pantone colors, which are very actionable along with the silhouette, along with the design elements, ready to adapt by our customers.

Jane Singer:

Okay. That’s really interesting. Looking at all the different information that you’re presenting and because it’s done quite scientifically, right? Using data and using algorithms and so forth, roughly what would you say the margin of error is?

Ganesh Subramanian:

Let me just give you some case studies where we have worked with different groups across the world, where we’ve improved full-price sell-through by an absolute 7-8% in just two to three seasons. Now, what that means is that if somebody is having a full-price sell-through… Now, this is also with adoption in a part of the range from let’s say from 50%, we moved them to 58 to 60%. But over a period of time, what happens is that… And In fashion, the margin itself is 10%. Now, if you increase your full-price sell-through by 10%, now you’ll save straight away a 5% margin, which is 50% of your profitability. In other words, you increase your profitability by 50%, which I’m talking about in three seasons of adoption.

I think the key point is none of this will work if you don’t adopt it. And this is a different way of working. And I think over the last three, four years, we’ve been working with our clients to one, buy into the philosophy, second is to adopt. And once people start seeing the results, it automatically starts rolling, because of the lead time involved it takes some time to see the result. But having said that, yeah, having said that it’s very important to try unless you’ll never see. Not really that, now products made with Stylumia’s intelligence, If you make products Stylumia’s intelligence, we have seen cases where we are able to improve up to 60% lift in revenue velocity. And there are extreme cases where we have improved by 10 X, but I’m just saying, if I just overall look at the population, 60% lift in revenue velocity, that’s the sales improvement, reduction of inventory by anywhere between 30 to 40%, which means if you’re sitting with four months of inventory straight away, you will shave away 30 to 45 days of stock.

Because products are selling very well. And more you start using this than you can, all of these metrics can start coming down. The whole idea is to improve from the current baseline. This is just from the market intelligence solution and maybe we’ll come to the prediction of demand. There’s a lot of leverage there too.

Jane Singer:

Right now, there are certain data based on the tools that we have available that enable people like yourselves to capture consumer data. What do you see that’s under development that will enhance our ability to gain even better data?

Ganesh Subramanian:

Absolutely. That’s a great question, Jane. Most of us today are collecting a lot of information within the spectrum of the fashion, but if you just look at how is fashion consumption triggered? What are the sources or causes of fashion consumption? I think fashion consumption is also driven by what’s happening around in the environment, what’s happening in the social space, and that’s just not limited to the segment of fashion. For one, I see the breadth and width of data sources need to enhance completely. For example, let’s say the weather forecast data, I’m just giving, for example, in 2018, entire Europe placed a huge over-bet on jackets. All of them underestimated global warming. So just imagine this, all of us making the same mistake, we are following each other. He’s making this jacket, that brand is making this, and all of us are making and putting… But what we all missed is global warming and temperature changes, which people don’t need as many jackets. And also the quality of jackets they want lighter jackets.

Now the question therefore is, I think we just need to bring width and depth of data. That’s definitely one area. I call them alternate data deployment along with fashion to see the future. And lots of micro-moments are happening. I think more and more I see that we just need to get into reflexive fashion because what’s happening is that we are getting into uncertain times. We’re not prepared for times like this. As an industry we sitting with very rigid, very long supply chains, even those who are crushing it, they’re crushing it incrementally. Now the question is, I think we really need to see technologies to bring the supply chain shorter and shorter, to an extent, and say, can we all produce what is exactly needed?

It’s a long way. We are not in a made-to-order situation. But the question is constantly, a lot of people are working here to say that, how do we crash the entire fashion supply chain? And we know through the last few months, supply chains have crashed. Those who had two collections in a year or four seasons in a year now thinking about six to eight seasons and they are moving towards 12 seasons. The old terminologies of seasons are crushing completely. Digitization of design is happening. How can we stop making samples and start taking orders like 3D, as you know is coming up, how can you predict using a 3D design directly using Stylumia’s prediction tool for example?

Jane Singer:

Can you?

Ganesh Subramanian:

Yeah, we can predict just with a 3D design. Why do you need to make a sample? We’ll tell you good, bad, ugly with a 3D design.

Jane Singer:

In other words, what you’re proposing is to be able to use the data from Stylumia, to go directly to a 3D sample in order to shorten your lead time and also to cut out a lot of the costs on sampling.

Ganesh Subramanian:

Absolutely. Why sampling? And the entire even B2B transactions in terms of whether this will work well or not, We sample for a lot of people to get validations. We show it to a set of people. All of that can be digitized.

Jane Singer:

You’re able to tie that in with what you’re doing on the front end on the data side of things with Stylumia.

Ganesh Subramanian:

Yes.

Jane Singer:

Oh, very interesting. Yeah. It’s inching us closer to a supply chain that is more of an on-demand supply chain, as opposed to what you were talking about at the beginning where people are trying to predict well in advance of when they would actually sell the product.

Ganesh Subramanian:

Absolutely.

Jane Singer:

And I think that that’s sort of been a missing link for many companies.

Ganesh Subramanian:

Yeah. And Jane, I just want to maybe elaborate a bit more in terms of just not about product intelligence. So to win in DTC, you brought the right context. To win in DTC, let’s imagine, we have most of the manufacturers, let’s say if they want to go to DTC or most of the existing brands for them DTC what’s of the channels, so that become now a prime channel. What’s the difference between a DTC channel having worked leading one of the largest fashion, online retailers of India, leading from a startup to a billion-dollar-plus company? I have the experience also of seeing both offline and online worlds. In DTC Business, what’s very important is that it’s just not your product, your product, relevant content, in the right way it’s very important.

Ganesh Subramanian:

And you just need to be digitally pleasing. Very, very important that you have to therefore get one consumer awareness. Now, consumer awareness building through digital mediums is a separate subject by itself. For a moment, let’s say you build that. Second, is when the customer comes, you just need to make your proposition very, very digitally appealing. Now digitally appealing is photography, its imagery, its content. We sometimes take content for granted, for example, using our intelligence, not only do we know winning products, but we also bring winning texts.

Jane Singer:

Oh, really? Interesting.

Ganesh Subramanian:

Absolutely. Therefore, it’s not about winning products, it’s about winning shoots. The kind of photography, which is working very well. For example, using one of our tools, we realized that if the model is facing the audience, those products were not doing well. If the model is tilting, his/her head away from the audience, away from the consumer on the screen, it was doing better.

Now, this is after heat mapping, and we figured out that this is the pattern we saw across winning products in select brands and retailers. So it’s winning photo shoots, it’s also winning text. See, digitally what happens, people either search through Google or any search platform, be it Amazon, they search. When they search, it’s very important that the search words and search words are very dynamic. The consumer problems of boards, many a time are very different from how brands call their descriptions. And how do you align this consumer language with the brand’s language? Now, this is also something that we provide. In fact, there are large retailers in the US using our platform to augment their textual data, master data. So your master data and image or your salesmen.

Jane Singer:

So their master data would be the data that they’re collecting directly themselves, and then they would be using your data to compare and contrast to what they’re collecting?

Ganesh Subramanian:

Yeah, absolutely. For example, should I call it a mini dress or should I call it a short dress? If a consumer is searching short dress, and you have called it a mini or midi dress, your product is not showing up.

Jane Singer:

That’s a good, very good point.

Ganesh Subramanian:

Now, the point is about taking care of consumers’ interaction in the digital medium through text, the perception of imagery. Of course, getting the products, right. You can actually spoil a great product through a relatively bad shoot and poor descriptions.

Jane Singer:

I think that’s a very interesting point that you bring up because I have noticed that when I look at descriptions of products, if you’re looking at let’s say eBay or whatever, and people it’s just individual people, their way of describing a product is actually a lot more engaging than when you’re looking at products on a corporate website. When I say a corporate website, I mean a big brand’s website, where the description is a bit dry. And oftentimes doesn’t include it actually surprisingly as much product detail as what you would get when you’re looking at someone who is selling on, I’m just using eBay as an example, I’m sure it’s the same on other similar sites. So it’s interesting that you bring that up because I think you’re right, how you describe the product and all the detail you give is very, very important in terms of making the final sale.

Ganesh Subramanian:

Absolutely. It’s also making you relevant. I think in digital DTC there are two things very important, and if you actually read how all these algorithms work, whether you want to sell on a multi-brand retailer like be it the Zalando, Germany be it or Amazon or any retailer for that matter, you need to be discoverable. What the retailer is looking for always is two parameters, relevance, and performance. How relevant are you for the consumer real-time? Now, I think we all need to ask very deeply how relevant is my proposition at this point in time? Now, imagine if you had a one-year calendar how do you answer this question, am I relevant today? Am I relevant right now? But that’s one question. Second, is my imagery relevant? Are my text relevant? Is my pricing relevant? So relevant in various key dimensions the consumer is looking for, and they are going to rank all products based on how relevant are you.

And second is how performing are you? In digital, you improve your business if you get more visibility, simple as that. As long as you’re performing, you get more and more impressions, more chances that people will click you. You go to PDP, you go to cart and you get converted. Therefore, if somebody is starting today, this is a message for people who want to accelerate their digital business. My few inputs on this is you just need to… Digital is not like scaling linear, it’s like a flywheel. How do you run this digital flywheel? A digital flywheel is like a huge iron rim, which you need to now push and start rotating. Initially, you need a lot of inertia, which means you have to break the inertia and move it. Which means what? You don’t spend your marketing dollars, for example, in a linear way if you want to get traction. If you want to get traction, you need to go through a nonlinear route and say, “Okay, my marketing budget for this year is 8% average.” So you start with 20, 30%, and slide it down and create the momentum.

Jane Singer:

Very interesting point, yeah.

Ganesh Subramanian:

A lot of people do is, they do the average bit and then say like, I used to spend marketing budget of 7% even, that’s okay when you are flywheel already running offline retail. In online retail, a lot of people don’t know you. Now in that market, if you have to create the initial momentum, you just need to start nonlinear. For example, I’ll give you this. This will be very counterintuitive. Many people won’t even believe that this is the right thing to do, but let me share with you. What happens in offline retail? You start with full-price sales, and then you start discounting towards the end of the season.

Jane Singer:

Yes, of course.

Ganesh Subramanian:

Yes. Now, it works the reverse here. If you want momentum online, you start with a 30% discount and come back to an 8% or 10% discount. It’s very counterintuitive, but it works. But the question here is that… No, the discount is just one lever, I told you, marketing is another lever. I’m not suggesting that you should start at the discount, but just to give you a perspective that you need to create and move this flywheel in an extremely competitive market. But the moment you get the eyeballs, and of course, you have relevance and performance, it starts to pay you exponentially. In other words, if you don’t do well, you continue not to do well. In fact, you go spiral down to poor performance.

Jane Singer:

What you brought up is a very interesting observation. And I think it’s probably, and I don’t want to say it’s the secret sauce, but I think it’s a good part of the recipe.

Ganesh Subramanian on Real-Time Demand:

See where I’m coming from is if you just go into all of this, you’ve been not talking about various dimensions, and I would sum it all and say that to win DTC, you just need to have “real-time relevance”, and “real-time performance”. Now to be real-time relevant, I think each one of us needs to ask how close to real-time are we in everything that we do? And product supply chain is just one of the elements. Real-time on texts, real-time on imagery, real-time on marketing, real-time on pricing, all of that, and real-time trend-spotting.

Now, if you just put it all together, lots of today’s processes are nowhere close to real-time. They are batched. What we think in autumn 21, will do well. And there is a report which is extremely static and that people are going through and making big decisions. Between now and 2021, autumn, there is so much which will change. I think the question is we all need to think about how close to real-time are we going to go? And it purely depends on the strategy of the brand that they can go closer, unless you have built iconic brands that anything you do, you set the trend, even they are facing challenges today.

Jane Singer:

Oh, absolutely. I don’t think anybody has an immunity card in this particular market. And I’m not saying specifically with regard to the pandemic, I think that even before the pandemic. Even as you were saying, the mightiest of brands can not be guaranteed, that people are going to buy anything that they put out there. They also have to be quite reactive to the market.

Ganesh Subramanian:

Oh, absolutely. Absolutely.

Jane Singer:

So I’m interested then just as a closing thought, what do you see as the next step? What would be the next step? What are some of the things that we can look forward to in the next year or two years or three years?

Time is going to be a huge differentiator. When I say time is a differentiator is “timing”. We can all buy revenue, we can all buy money. Money can always come back, time will never come back.

Jane Singer:

That’s very true.

Ganesh Subramanian:

Three years back, four years back, those who, or five years back, if somebody took on an interest in digital, they developed digital really very well. Today they are facing less hardship compared to those who are waking up in a crisis. So for me, I think the big shift is going to be timing as leverage, timing as a differentiator, just not about the product. Some people keep on asking us and say, “What’s the white space in the market?” I said, “You do everything that you do, just change the timing. You have a differentiator, which can’t be copied easily.”

Jane Singer:

Yeah, it’s interesting because I think that people are looking at everything, but that.

Ganesh Subramanian:

Absolutely. People are saying that “Tell me what will happen in 2022?” I think, is that the right question to ask right now?

Jane Singer:

Yeah. I mean, nobody really can predict the future.

Ganesh Subramanian:

I see that as a trend and I see those who catch timing as their differentiator, apart from everything else we’ll win in the time to come. I see that in the industry, we can all collaborate. I think like in the technology world, there is so much integration, which is happening. In the technology world, so many tools talk to each other. I see in the fashion that a lot of discrete tools are available, they all don’t talk to each other today. Very little integration. I think ecosystem partnership is a big moment I see. 3D companies working with intelligence companies, intelligence companies working with ERPs, all of this coming together and to solve a holistic problem. We can actually multiply all the benefits to the customers by bringing it all together. It’s all happening from manufacturing to design tools, to intelligence solutions like this to supply chain, so I see integration happening.

The last one is skill upgrade is something that I see as very, very critical.

Jane Singer:

Which areas do you see a talent gap?

Ganesh Subramanian:

We’re not used to technology adoption at a fast pace. How do we adopt technology from the time it comes, an experimental culture? And a DTC business is full of experiments. If I have to give a technological framework, it’s called the AB test. How can we make a fashion industry, which is used to AB test? Not just in the front end of the website, can we do an AB test in design? Can we do an AB test in manufacturing? Can we do an AB test? Now AB test instant validation of various things which are available? Yes, when you do a lot of experiments, 20% of it will work. Now, if you go with that, I think we will all see a huge acceleration of what versions of businesses changing transforming. The cost of the experiment today is very, very low compared to the kind of wastage we are all having.

Jane Singer:

Ganesh, thank you so much and really appreciate your time and joining us today and wish you tremendous success.

Ganesh Subramanian:

Absolutely. Jane, thanks for inviting me, and it was great sharing my few bits with the audience today, and hope it adds some value.