>> live from London, England. Q. Covering A Ws summat. London twenty nineteen Brought to you by Amazon Web services >> to the A. W s Summit in London's Excel Center home. Susanna Street and David is my co host today on the Cube. They mean so much to talk about here at the summit today to do with machine learning an A I and I'm really pleased to say that we have to really key people here to discuss this. But we've got some Tom Summerfield who is head off commerce, a foot asylum on also Richard Potter, who is the CEO of Peak. Now, you guys have really formed a partnership, haven't you? Foots asylum is a leisure wear really. Retailer started in bricks and mortar stores. Really moved online on Peak has been a pioneer for artificial intelligence systems really well to get together. What what comes? Sparked Really your demands ready for their services, Tom? >> Yeah, well, so way knew that we needed to be doing something with data on A and we didn't really know exactly what it would be way were interested in personalization, but then also in a bigger picture, like a wider digital transformation piece for the business where well established bricks, a martyr business but a fast grow in online business. And we're interested to know how we could harness the momentum of the stores to help the digital side of the business and also vice versa. On we thought data would be the key, and we ended up having a conversation with the guys at Peak, and that's exactly what we've been able to do. Actually, on the back of that deliver, we're delivering a hyper personal experience for our consumers Now. >> I was one of the statue that I notice when looking into what you be doing, a twenty percent increase in email revenue. So that's quite remarkable, really. So Richard, tell us you how you're able to do this. What kind of services that you lean on? T make those kind of result. >> It's a combination of a lot of things, really. You know, you obviously need people who know what they're doing from a returning a business perspective. Married with technical experts, data science algorithms, data, I think specifically how we've done is picks built a fairly unique A I system that becomes almost like the central brain within our customers businesses on off that algorithms help automate certain business processes and deliver tangible uplifts in business performance like the twenty eight percent up lift in sales here, Um, in order to do it. So it's quite a long journey, I suppose. The outlook we took when we started collaborating was was that if we could deliver that hyper personalized shopping experience, we were always going to be ableto show customers the right product at the right time. And if we were doing that that we would lead Tio Hi brand engagement, higher loyalty, higher on higher lifetime values of customers. And that's and that's what's shown to be the case in a silent example. >> Yeah, definitely that echo that. You know that the hypothesis hypothesis, wass. If you can show the right custom of the right product at the right time, then their purchase frequency average order Volumetrics all starts move positively and ultimately than affecting their long term engagement with our brand, which increases revenue on also delivers a more, you know, a frictionless consumer experience, hopefully for the customer, >> because I suppose your experience is the same. So many companies out there they're sitting on this huge pile of data, yet they don't know how to best optimize that data. When did you first realize, Richard that there was this kind of gap in the market for Pete to grow? >> Yeah. I think data and analytics have come on a bit of a journey away from common sense reporting Thio more advanced analytics. But when you get a I and machine learning what you're talking about, his algorithms being our self learning make predictions about things, and that actually fundamentally changes the way businesses can operate on DH. And in this case, a great example is you know, we're sending hyper personalized marketing communications, Teo every single foot silent customer. Um, they don't realize necessarily that they are tailored to them, but they just become more relevant. But it doesn't require a digital marketed to create every single one of those campaign or emails and even triggered the sending of those materials. The brain takes care of that. It can automate it. And what the marketer needs to do is feed it engaging content and set up digital campaigns. And then and then and then you're left with this capability where eyes saying you might be a market for this product. Let's let's send you something that might appeal to you on DH that just gives that gives a marketing team scale. And then, as we move into other use cases like in the supply chain for film and delivery of product the same thing that teams just get huge scale out of letting algorithms do those things for them. Andi, I suppose the realization for us that there was that gap in the market was just that you can see the out performance of certain cos you can see that Amazon attributes thirty five percent of their sales to their machine learning recommendation systems. I think Netflix says eighty five percent of all content is consumed >> in prison. It's Al Burns. Andi. Companies >> like that can harness machine learning to such a great degree. How does how do you know howto other businesses do it? Who can't access that talent pool of Silicon Valley or along the global? You know, the global talent leaders in tech and that's that's what we have. The insight that is Peak Way could create a company that gave our system is the that technology and that capability Teo deliver that same kind results that the Amazon and Netflix >> So before the Internet Yeah. Brand's had all the power you could price however you wanted if you overprice, nobody even even knew. And the Internet was sort of like the revenge of the consumer. Aye, aye. And data. How gives the brands the ability to learn more about its customers. But you have to be somewhat careful, don't you? Because your privacy concerns, obviously. Gpr etcetera. So you have to have a value proposition for the customer, as you were saying, which they may not even know that machine is providing these offers. Yeah, but they get value out of it. So how do you guys think about that in terms of experience for the customer? And how do you draw that balance? >> I think from my angle that Richard touch on a couple of bits there to do it scale first and foremost across the entire all on on Thai network of consumers is killer element to it. But to deliver that personal experience, I think consumers nowadays are so they're more expectant of this. Really? We would have considered it innovation a couple of years ago, but now Actually, it's expected, I think, from the consumer. So it's actually in the name ofthe You have to move forward to stand still. So but way Think we're We're right at the front of this at the moment. And we're really looking now how we optimize the journey for the consumer so that actually we know if we're from Simpson transactional data that we have in a little bit of over behavioral data that, you know, we're really conscious of the whole GDP, our peace and stuff, and that's really, really relevant and super important. Andi, I'm pleased to say that you know, we have that backed by a peek. It's completely on lock down from that perspective as >> well. Where do the data's where the data source of comfort. You mentioned some transaction data. Where is the other data come from? Using show social data and behavioral data? Where does that come >> with these elements of social data? Some of it is a little bit black box, so you can always access it. And that's the GPR piece there. And rightly so. Actually, in some cases we have a loyalty scheme which allows us to understand our Kashima's better in our bricks and mortar retail, which is really cool that we've got some of that transactional data on a customer level from the store's way know that some people in our sector maybe don't have that, so that so that allows us to complete sort of single customer view, which then we can aggregate in peaks brain, then transaction data on the website in the app and bits off browsing, you know, just within our own network. But you know where customers potentially being and reactive of somethin, a piece of content on journeys within the website, That's that's how we build that view. >> Do you think this is the way that more bricks and more two stores Khun survive? Because so many are closing in high streets up down that you can in other countries, because simply they're not really delivering what the customer wants? >> Yeah, I think so. Rich Now, both feel quite strongly now that wear something to this now a little bit. It's a really As as our relationship for the two businesses has evolved, it's become clearer and clearer that actually we've armed with this. You know this data, our fingertips, we can actually breathe fresh life into the stores, and it's in the eye of proper true omnichannel retailing way. Don't mind where the cost consumer spends the money. We just need to be always on in a connected environment. So that way said before pushing the right product at the right time. And when that when they're in market, we turn up the mark the message a little bit. But then understanding when they're not in market and maybe to back off him and maybe we warn them what with a little bit of a different type of message then and actually we're trapped would want to challenge ourselves to send but less better marketing communications to our consumers. But absolutely that store piece is now, so we tail back. Our store opening strategy is a business to focus more on the digital side of things, but now it's possible that way might open some more stores now, but it will be with a more reform strategy of wet, wet where, why we need to do that? >> Isn't this ironic? The brick and mortar marketplaces getting disrupted by online retailers, obviously Amazons, that big whale in the marketplace and your answer to that is to use Amazon's cloud services and artificial intelligence to pave the way for your future. Yeah, I mean, that's astounding when you think about >> coming. >> Yeah, sort of unified commerce approach, Tio. You know, there's a place in the world for shops. It's like it's not Romance isn't completely dead and going shopping Friends out, you know so on. Actually, yeah, we're using honest in the eight of us, but we'LL hire our friends at Peak. Yeah, it's it's some irony there. I think it's really cool. >> And that decision that you made obviously made made lightly. But you saw the advantages of working with the clouds outweighing the potential trade offs of competition. >> Yeah, I mean, that's not that was never really, really no, I'm certainly not know. I think this is something that is happening, that data, and on harnessing it in a in a safe, responsible, effective way, I believe, is the future of all commerce. So >> that as far as security is concerned because, of course, we have had data breaches. Yeah, customers, credit card details, access. How do you ensure that it's as secure as possible in the way that you you you choose the services. I think >> that come that just comes down to best practice infrastructure on the way we look at it, a peak is there's no bear tools in the world to do that, then the same technologies that Amazon themselves use. It's to do with how you configure those services until ls to make it secure. You know, if you have an unsecure open database on a public network, of course that's not secure. But you could have the same thing in your own infrastructure, and it wouldn't be secure. So I think the way we look at it is exactly the same thing on actually, being in the Amazon plan for us gives us a greater comfort, particularly in terms of co location of data centres, like making sure that our application fails over into different locations. It gives us infrastructure we couldn't afford otherwise, and then on top of that, we get all these extra pieces of technology that can make us even more secure than we could do. Otherwise we'd have to wait, have to employ an army of infrastructure engineers, and we don't have to do that because we run on Yes, >> okay, so we were able to eliminate all that heavy lifting. Same goes. You've got this corpus of data. I'm interested in how long it took to get through. A POC trained the models how much data science was involved. How much of a heavy lift was that? Yeah, well, I think for us >> we better be pretty rapid. Actually, we start working together in January last year, so we're only just sort of year into that. >> And in that faith in that entire >> sofa length of of our relationship, we've gone from high for personalizing digital campaigns to recommendation systems on a website to now optimizing customer acquisition on social media and then finally into the supply chain and optimizing demand. And so on and on. I think there's a lot of reasons why we've been able to do it quickly, but that's fundamental to the technologies that that peak is built. There's two. There's two sides to it. Our technologies cut out a lot of the friction so way didn't run a proof of concept. We were able to just pick it up, run with it and deliver value. And that's to do with I think, the product that peak is built. But then you obviously need a a customer who's who's going on a transformation journey and is hungry to make that make that stick in London on. Then when the two come together, >> I think that it's an interesting point that, though, because while suite for asylum, we always I always say it's that we're not. We're not massive, but we're not tiny, but it's the sort place you Khun turn upon a Monday and say, I've had an idea about something and we're not doing it by Friday. That's That's a nice, agile culture. It can create some drama as well. Possibly. I think it's really straightforward to get straight into it. And I think this is where some of the bigger, um, sleepier high street retailers that Amar fixed in a in a brick Samara world need to not be too afraid to come out and start embracing it. Because I think some of them are trying now. I think it might be a little bit late for some now, but it's it was just it was just wasn't that hard really to get going here >> and you've seen the business results. Can you share any measurements or quantification. We've >> got a really a really good one that we're just talking about at the moment. Actually, Way were able to use segmentation tools within within the peak brain, too to use them on social than Teo. Create lookalike audiences. So Facebook custom tools, right? We'LL help you create audiences that it thinks will be wrapped pirates complex algorithms itself. But we almost took a leap ahead of their algorithms by fire, our algorithms uploading our own segments to create a more sophisticated lookalike audience. We produced a row US results or return on that spend People are not familiar with that of eight thousand four hundred percent which we we would normally be happy as a business. We've sort of seven, eight hundred percent. If you're running that that we've say on AdWords campaign or something like that, that's quite efficient campaign. So it's at zero. We were a bit like it felt like it's a mistake that, you >> know, that is >> not the right Yeah, but not so that's super cool. And that's really that's really opened our eyes to the potential of punishing that the, you know, our sort of piquet I brain to then bring it onto Social on. Do more outward. Advertise on there. >> So moving the goal post meant that your teeth have really high school. Thank you. Thank you very much for telling us all about that time someone feels on Richard for so thank you for joining me and David Auntie here at the age of Lou s summit in London. Merchant to come on the King.