Episode 30 Codurance talks: What is the role of machine learning in optimising e-Commerce experiences?

In this podcast José Enrique Rodríguez Huerta, Managing Director at Codurance Spain will chat with Sergi Ortiz, CEO, Co-Founder at Shimoku about how Data-driven technology infrastructures can create a huge increase in value on e-commerce platforms.

Sergi is a graduate in physics from Universitat Autónoma de Barcelona and Freie Universität Berlin. He also has a Master’s degree electronic engineering from the  Universitat Politécnica Catalunya, a Postgraduate in Data Science and Big Data from Universitat de Barcelona and a MBA-EAE from business school.

Machine learning allows us to make better predictions, and increase the business value chain through the best user experience. With more comprehensive and accurate information, organisations can make data-backed decisions that ultimately lead to better products and services.

Technologies like machine learning have the ability to help you become omnipresent, they run 24×7, and deliver the perfect shopping experiences that customers demand in today’s day and age.

José Enrique Rodríguez Huerta: Hello, hello, hello, and welcome to another episode of Codurance talks, the podcast where we talk about all things technology, software development and craftsmanship.  

And today we're going to be mixing a couple of topics in particular, we'll be talking about how we can use machine learning to optimise online e-commerce experiences, something that if we look back a few years was kind of reserved only for big brands like Amazon, and so on, and with the let's call it democratisation of this kind of technology with frameworks like TensorFlow, or providers like AWS, and Azure, and so on giving us components that we can use out of the box has become a lot more accessible to, let's say, almost any business.

And to that we have with us today Sergio Ortiz, he's the CEO and co founder of Shimoku, a company that is helping e-commerce businesses make better decisions aided by the power of machine learning. 

Sergi Ortiz: Jose, thanks for inviting me.

JRH: It's great to have you here. And, you know, when we were preparing this thing, I was quite a quite interested because you have gone through a journey in your career before we you know, Shimoku? No, because you started with physics, right? You started, you have a degree in physics, and then, you know, move to electronics, and then, you know, big data that a science and an MBA. So that for me was kind of fascinating.

S.O: Yes, yes. It sounds like I don't know what to do right now! 

JRH: On the contrary, I mean, I think it actually provides, you know, a different point of view, not just the, you know, technical challenges, but also the business challenges that technology is trying to solve, right. And this is one of the reasons I feel we a great episode today, because, again, it's it's something that is pretty interesting is that a overlap now between the technology itself, which is something cool, but at the same time solving a business problem now, and, again, when we, when we think about this stuff, like we talked about machine learning, all the times in the in this podcast and stuff, but I guess what people don't realise is how much we are influenced by this technology, you know, even when we don't buy online, right? Like even even if we just go to a store that is somehow being fed a lot of the times into machine learning models, to then you know, make decisions and all this stuff. So just to give a bit of a an overview, you know, what are some of those ways in which you think, you know, we're being influenced, let's say, as consumers, or as, you know, people who buy online?

S.O: Yeah, and this is, this is a very good point. I mean, certainly, we are getting in a new era in which basically, algorithms are everywhere. And this not only influences our work in the future, because many jobs are going to get lost. Of course, some others are going to be created, but the mass destruction of works is going to be remarkable. But beyond that, it also influences our our daily life, right? Because things that before require you to think in the process and to optimise the process. I don't know for instance, we can think in, in a taxi driver, right. So these things are these kind of jobs are probably not going to exist in 10 years in five in China, and probably more or less in the same time in the States. And in 10-15 in Europe. From this automatization of a specific task, it also comes another automatization of other tasks that maybe we are not so aware of, like when we choose a movie, for instance, right? Now we have Netflix that well we all scroll through during hours in Netflix, right? But many times we choose one of the of the recommendations we have for us. And if we think of it like before, like 20 years ago, we went into into a movie store, and we spend there some time with our family just deciding discussing, and it was not so influenced, right. And of course the information they have of us was much more reduced. I mean, if the owner of the store knew as they could suggest, basically in our experience, but it was much more hand-crafted experience. What do we have nowadays is quite different actually is and it's also problematic. How many trends goes into the direction of and really thinking in the ethics and the ethics of these AI that really can influence our life, unconstrained this life, because, and this is listen to stories every month of this happening, right? Like, black person on the stage, don't get alone, just because it's a black woman, right? So the way it's influencing our decisions, the way it's influencing our job, and the way you really it influences a, where we can arrive in our life. And it's already strong. And the thing is that it's going to get stronger and stronger during the following year. So we are just in the beginning of, of this.

J.R.H: So you mentioned recommendations, that's definitely one way which we, you know, we see it all the time now. But it definitely goes beyond that, right? Like it gone to the extreme of, you know, with that information about you, basically, a can predict no a, what your behaviour is going to be right or a when to maybe suggest for you to buy something or when to is a better moment to get in touch. But this is, this is something that we're not very aware of, ... I think it was for diapers, and that kind of stuff now, and the name of the person's daughter, and finding out that the daughter was pregnant just because of her habits now. So there's this ethical kind of aspect of it. But there's also, you know, the ability to some extent to predict the future.... for the business, at least this is this is pretty interesting...

S.O: Exactly! Yeah, even going further. I mean, it's really driving our lives, meaning if we think of all these a thing that happened with Facebook and elections, and I mean, I, if we think like FBI 30 years ago, they check what books people read on the public libraries, but now this is they can go far beyond right. And even if we think in our democracies, they are in risk, like it's very easy to get rid of politicians that are a problem for the elites. And yeah, it's really not the layers of society in all the layers of our daily life nowadays already.

J.R.H: So, when it comes to e-commerce is right, which is what we will be focusing on today, what are some of the examples that that you can give of this? we talked about recommendations or what else do you see that.

S.O: For the commerce the dream is like you are in a physical store, this has been always the dream of the commerce right, you have someone giving you the surveys, you recognise that person that person recognise you and that person tries to give you the best possible service in the moment they commerce born in the in the early 90s. Now we have still an example as Amazon so it was far more like buying those all style of buying through phone calls with a brochure right. And the kind of things that are happening in e-commerce the kind of things that are showing up have to do in particular with this, with this idea of rich the difference between purchasing through a brochure making a phone call and purchasing in a store where you feel you have someone really taking care of you. 

J.R.H: You really personalise you know almost tailored experiences for you now as a consumer.

S.O: Exactly, exactly. So, basically, we will go with recommender systems, but when I say we I mean the market, the industry, right? But we also go with things like forecast, churn, forecast, the engagement of the users forecast, what problems are going to be solved and in which amounts, all these sorts of things are already happening. And the idea is to really improve this customer experience, which is the ultimate goal, right?

J.R.H: Well, and it's not just for the customer, I can see one of the assumptions you mentioned, how this can be very valuable for the business. Now if the if you know that this particular product is going to be selling more or you know, you can remove forecast, a you know when to you know, get more of that product now or or ultimately determine You know, if you need to stop buying and so and save money now also or basically not to have that opportunity cost of people having demand but not having the product itself.

S.O: Yes, and and it also has a strong influence in the people working in the commerce, I mean, in the people, which is under the hood, for instance, and marketing teams, which before they work with these long CAC cells of the customers trying to find out some clothes and some patterns, right, these people now can really work in things that are more for humans more a in the imagination side more in the creative side, right. And all these a task that we can understand is driving right, or 30 years of operations, based in, in some outcomes, these things are, are getting more and more automatization by machine learning. And this is influencing us as customers, but it's influencing us as workers of the sector.

J.R.H: Yeah, so what what do you think is necessary for you know, a business to to be able to, let's say, innovate in this in the right, when I say innovate, I'm not saying like the being the ones, you know, in the bleeding at the bleeding edge of technology, I'm just saying, you know, if you're bringing in new technology to enable things that you're you weren't doing in the past, that's also innovating, within the context of your reader, what do you think are the say, prerequisites now, to some extent that that you find are needed for this?

S.O: Yes, regarding the requisites, there are by one side, the technical right, like you have to have good data, you have to have an hour's history of data, so that the the algorithm can really learn something. But there is also one thing that we the time, I put more and more importance, and that is never, never thought or I never say this in in blog post and so on, which is basically like, understand what is the problem you want to solve meaning much machine learning, artificial intelligence, in general, is not making any miracle basically, is taking a task that can be done by a human. And it's making that task in a predictable way, meaning is not going to have bad days, is not going to have holidays, it's going to do it 24/7 with the same performance always. So for me, it's very important that we understand which are the task that our people in the company are doing, which of them can be automatized. Until we automate also need this other layer of 'okay, do we have the data do we have the IT infrastructure that allows allows us to run a machine learning curve learning so that the task doesn't need to be done by a person anymore, or at least supervised by the person, not carried on?'

J.R.H: So the technology itself is one of those, do you feel there are other areas that are also, you know, need to be in place? Because it seems interesting, though, because you're mentioning that, you know, the first thing is you need to be clear about what problem you're trying to solve. What is it that you're trying to, you know, do predictively and so on. But a lot of the times organisations, you know, they are messy, you know, it's difficult to get that alignment know that a alignment between the technology and the product, we should, because sometimes it's not even very clear what it is there... 

To be continued....