Interview with PredictivEye CMO Shadi Yazdan on using AI in Conversion Optimization

In this interview Darius spoke with PredictivEye CMO Shadi Yazdan on her interesting background of entrepreneurship and product launches and her current startup which is using a propriatery AI engine to increase conversions on websites.

TRANSCRIPT:

Darius 0:01 Welcome to the retail tech podcast. My name is Darius Vasefi managing partner of InfiniVentures labs, which is a startup studio and specializing in e commerce and SAS products and retail. This interview is on clubhouse and is being recorded. We'll be taking audience questions later on. Today I am speaking with Shadi Yazdan chief growth officer and Chief Marketing Officer of a company called predictive AI. That is helping retailers and online companies convert better using machine learning a unique style of machine learning and artificial intelligence. How are you today? Shady?

Shadi Yazdan 0:48 I'm good. Thank you My journey?

Darius 0:52 Yeah, yeah, so let's let's start from, yeah, maybe a little bit about your background. I'm always like, really happy to see female founders, or executives and startups, we don't see enough of them. So I'm really excited to see and learn about your background and how you got to predictive AI.

Shadi Yazdan 1:14 Awesome, more than happy to share. And also you forgot a female in the AI space. I'm still considered a unicorn in this world,

Darius 1:22 that's even more rare.

Shadi Yazdan 1:25 Yeah, still quite dominated by men. So yeah, I'm gonna start. At the beginning, my entrepreneurial journey actually officially started back in 2000. When I started my undergrad degree, my father sat me down, and he said, you know, you need to go to university and get a degree. I also don't care what degree you get, which is quite unheard of, for a Persian father who wants their kid usually to be a doctor, lawyer engineer. So he said, I don't care what degree you get, because you're actually there to meet people grow your network and learn life skills. And by the way, he said, I'm not going to be able to pay for your education, you have to find a way to pay for it. My father is all about teaching life lessons, when you're in the middle of trying to figure your life out, he just drops bombs on you. So this is kind of like his way. So at the same time, my sister was working retail and making almost nothing. And I thought to myself, how do I make the most amount of money work the least amount of time and still have time to complete my course work. And I was doing a double major at that time. And I looked at, you know, what I was good at and successfully launched, grew and sold a tutoring business, which I ran for four years, before I left for a once in a lifetime opportunity to work and live in Japan. And after I came back, I spent about 12 years working for the government of Ontario. So I live in, you know, Toronto, Canada province of Ontario started working for the government as an educator, facilitator and project manager. I always felt like there was something that was missing, but I just couldn't figure out what it was. And I didn't know what else to do. So I used to run a lot of side businesses, anything from selling products, to tutoring, just anything to get me excited about work. And I really loved being the one in charge planning, organizing, interacting with clients, one on one. And I got to be honest, I really fell trapped. In my nine to five job at getting an advancement or moving to a different area of the organization was quite challenging, because everything was really based on seniority. And to be honest, my heart wasn't really in it for me to even work extra hard to get what I needed in place to get to the next level. And my performance reviews often left me thinking there was seriously something wrong with me, because I was told over and over again that I questioned things too much, that I pushed boundaries, that I was too intimidating, too ambitious, and the list goes on, and on and on. So finally, one day, the voice inside me that kept telling me to leave God so loud, that I couldn't ignore it anymore. And I decided I had to do something. And the only thing I knew at the time was to get a master's degree outside of the world that I was in. So I decided to apply to an MBA program. And after a lot of research, I learned that if I did a regular MBA, I would have to specialize in one particular area. And I had no idea what I actually wanted to do and I really didn't want to get trapped again in a job or career that would make me unhappy. However, my research led me to the Executive MBA program with would allow me to learn about the different areas of running a business. This was perfect for me as I could learn, and actually see what excited me most. So it was kind of like a buffet for me to try out. I knew the program was going to be expensive. And I knew that I would have to have, you know, the that it was really important for me to find a way to do the programs somewhere that would be globally recognized because I wasn't really committed to living in Canada for the rest of my life. And unfortunately, the Canadian schools aren't really recognized around the world as the American schools are. So I knew I needed to get an executive degree from an American School one with a big name. So I decided to apply to my dream school, which was MIT, to the Executive MBA program. And I got accepted, and I was even offered a scholarship. But unfortunately, I could not afford the tuition, it was crazy. And with the conversion rates as a Canadian, it was even more expensive than I could fathom, my parents couldn't help me, I didn't have the type of income working for the government to be able to pay for it. So it was very heartbreaking for me not to move forward with the decision of doing my MIT degree.

So very quickly, as I was going through this problem solving, trying to figure out what do I do, my father sat me down, and I'm a child of refugee parents. That came to Canada in the 80s, when the war was happening between Iran and Iraq. And my father sat me down. And he said, first of all, stop feeling sorry for yourself. Second of all, you need to be grateful that you were in this country to begin with, because God knows what would our life be like if you were back in Iran? And third of all, if I can bring a family here with absolutely nothing in my pocket, you can figure out how to get a degree. And I was like, Okay, you know what? That's right. You're right. So I started working, trying to figure out very quickly, I came across a program in Canada, where I could get an American degree through a partnership with a Canadian school. And the program met all my criteria of being face to face traveling to the US to study with American classmates, which was really important for me, because I really wanted to do an MBA, not for the learning, but more for the network and the connection, because that's what I want it to build. So I'm proud to say that I got an American degree, paying the Canadian fee. And through the MBA program, I kept talking to different people to find out what they did and why they did it, and how they liked it. I was on a hunt to really find my dream job. And I didn't know what that was. And the majority of the people I spoke to, had no idea why did, they did what they did. And most of them talked about the fact that they were advised by career counselors, family members, and so on. So this got me thinking about the importance of not really doing just any job, I knew I would be in the same place as I was, at that time. And I didn't want to be miserable in my job. I wanted to do what I loved. And I wanted it to bring me lots of excitement. So I sat down and start to figure out what are my What are my values? What do I want in this ideal job? I may not know what it's called, but what are those ideals. And I came to this after I came across a book called designing your life by Bill Burnett, and Dave Evans. And this book was a game changer. For me, it gave me the tool to figure out that dream job. I knew that I wanted to work, I wanted work that would give me flexibility. autonomy, allow me to make tough decisions, and constantly learn and learn quickly. I found out very early on that learning, not being the one that knows all the answers was really, really important for me in any job that I did. So the book really also taught me to reflect on my life, what I had done in the past, and what brought me the most joy and made me lose track of time. So the importance of being in the flow. And this is where I went back to the businesses that the business that I ran in 2000. And all the side businesses that I ran and how much joy they actually brought me. And this is where I decided that the entrepreneurial world to create that ecosystem. And that job was really what I had to do. So I built a business right in my MBA program with three others in my MBA class, great business idea, but not the right team of people. It was a disaster. And this is where I quickly learned the importance of choosing the right people to join a business venture as co founders and not just blanketly approach someone and say, Hey, will you be my co founder? It's almost like just, you know, meeting someone on the street and saying, Hey, will you marry me doesn't work that way. You got to date people. So after this, I By running a number of other businesses and learned that I didn't want to be a solo founder. So I had all of these criterias in place for myself, and I was learning very quickly. So soon, I co founded a US based e commerce marketplace called shop Tara. And this is a marketplace of all natural, eco friendly personal care pet care products, which we ran for about a year and a half, great team, great product, we had lots of fun doing it. And very quickly, at the end of that one and a half year, I decided, where do I want to go from here? Do I want to leave and go to Jersey where the business was registered live there? Where do I want to do with my life, and I decided that take, you know, exiting the business was the right choice for me. Because at that time, it wasn't going and living in Jersey was not really something that I wanted to do. So very quickly, another one of my MBA classmates who was on the same study team, as I was, approached me about this AI company that he was building. Predictive AI is an AI company that predicts customer behavior with 90% accuracy working with midsize e commerce companies, from retail to travel agency to those selling products and services online. So by predicting customer behavior, we were able to massively increase that conversion rate. So I not knowing a lot about the AI space started digging deep learning everything in anything that I could. But at that time, we were targeting insurance companies, healthcare professionals, Telecom, and retail, and then the pandemic hit. So today, we've had to pivot and we've had to pivot a number of times, and we target the midsize e commerce space, believing that that is an industry that really needs us just based on the pandemic and everything that's happening around the world. That's proving to be a bit more challenging than we originally thought. So today, in addition to running predictive AI, as you are fully aware, I'm also quite active on clubhouse. You know, I'm just here sharing my experience in the startup world giving marketing tips, talking about the importance of taking care of your wellness as you're building your business, because I went through some terrible burnout as I was building businesses trying to figure out what I need to do next. And I do all of this as a way of supporting others on their journey as well, because that's really important for me. And I'm also very involved in my community. I live in downtown Toronto, I volunteer a lot of my time. I'm a volunteer advisor with a number of accelerators and incubators in the city as a way of giving giving back to that startup ecosystem. So that essentially, as quickly as I could articulate is my background and how I ended up where I am today.

Darius 12:46 Wow, what a fascinating life. That's amazing. Well, first of all, congrats on having a wonderful father that actually takes the time to talk to you. And not, you know, dictate things. So that's very good. And bringing the family here when, you know, things were really difficult. So I think he's probably the first entrepreneur in the family without realizing it. That's all about taking risks. That's what entrepreneurship is. 100% Yeah, so. So I think, you know, we've talked a few times about predictive AI. And I just wanted to probably, you know, quickly just touch on some of the key differentiators because you know, you know, we hear AI in pretty much every marketing pitch these days, especially in retail companies. And it's fine. We all using AI in some way form of or shape. But it differs a lot. And that difference is actually very important when you get to AI. So I'd like to maybe spend a few minutes just talking about the differentiating factors and why these differences that predictive AI have are so important.

Shadi Yazdan 14:10 Great, yes. To get to that, I just have to really take a little bit of a step back and talk about why this company was born and where the idea actually came from to begin with. Because I think that that's going to answer a lot of the questions in terms of what sets us apart from what is currently in the market. And I'm also quite frustrated with all these companies that put the name AI to their name. And really all they're doing is data mining. And using static AI rule based and just not really it's not the way that prediction is done and the way that AI models can be utilized. There's so many different types of models that are currently in the market. So to give you a little bit of a background, the whole idea of predictive AI was born out of my co founder who's actually the genius And the developer behind building our AI model, he was working for one of the biggest consumer insights, insight companies in Canada. He was one of the executive people in that company. And he was frustrated with a survey based and AV test techniques that were being used and still are being used to understand customer behavior. So unfortunately, as you know, these techniques don't really provide agility to adapt to change. And we can forget that they often take three or four months to produce any results. And by that time, those results are already useless, outdated, too old, and you got to go back and do all those survey based and in a B testing again, so they're not really and they take a lot of resource time, manpower to run as well. So my co founder knew that there had to be a better way to do this. So he actually approached his CEO and pitched his idea of developing AI models that predict customer behavior in real time, his CEO didn't really take to the idea so positively, and my co founder decided, you know, what, I'm just I'm really committed to this. And I know that this is where the future is going. And this is what I need to do. So he decided to build the models himself on the side as he was working full time. So you know, funded all by himself, he's a data scientist himself, but hired a number of other people to help him. And once the technology was developed, he left his job and brought me on board to help with the business side of things. At that time, I had already made a reputation for myself as someone who was able to successfully take a business and bring it to life. And he's not very the type of person that wants to be, you know, in front of the customers and talk to them. So he found that our skill sets really complemented one another effectively. So that was one of the reasons why I came on board. So now going in depth about how this technology actually works, and what it does that it's currently not offered in the market. I'm going to go into the detail on that. I just wanted to take a pause and see if there's any questions you have before I start talking about that.

Darius 17:19 No, I think that's very interesting. The way that you know, you are you connected with the founder or the your co founder and used your experiences in the past about a picking the right founding team is, is very important and is valuable. So I'm glad you touched on that.

Shadi Yazdan 17:45 Absolutely. So the way that this technology really works, I'll take it back to talk about some of the popular way companies that are out there, like digital marketing type companies. So I would say like Google, and other marketing agencies are really good at help helping you bring customers onto your ecommerce site. But really conversion and customer experience and personalization is up to you. And they've done a fantastic job of convincing you that you need to spend money to keep bringing customers onto your website, bringing people you know, they're not really customers, I shouldn't even call them customers, but bringing traffic onto your website. However, as I mentioned, what about conversion, not a lot of people are talking about this. So our AI technology that we've developed our proprietary AI models, which we've named pX 360, picks up where Google and other marketing agencies leave off. So for example, once the visitor is brought to your ecommerce site, our technology gets to work to really understand their intent, and lead them to complete their purchase. And this is all happening in real time. And no one is offering this real time analytics, everybody makes a prediction once the sale has already been completed. Other companies out there that say they predict customer behavior, need access to your CRM data, historical data. And based on the data that they collect, they make a prediction. So they're predicting as I said, once the transaction has already taken place. So let me give you a little story. Let's envision 1000s of customers visiting your e commerce site right now. What we do what pX 360 is able to do is we do we conduct millions and millions and millions of experimentation on these visitors individually in seconds in real time, to understand their behavior, and most importantly, their actual intention. So for example, are they looking for a particular product? Are they looking to buy something? Or are they just looking for a dealer discount? Are they just researching? Are they just window shopping? What is their intent. And the goal of px 360 is to understand the visitors intent and fulfill it and fulfill it like it's never been fulfilled before. We like to say oftentimes, we know what that online visitor wants before they even know what they want. And during that prediction stage, every time the AI models predict successfully, the models are rewarded. And if a mistake is made, they learn and improve quickly. And this makes the model smarter and smarter as it continues learning. And the learning process, oftentimes, depending on your environment, can take anywhere between two to three weeks. All of this, again, is happening in real time, in seconds. And once the models figure out the customer intent, it gets to work to lead them to make a purchase, which leads to you increasing your conversion rate. So the technology is extremely unique with nothing like it in the market. Because the current AI models, which we've just started this discussion with where everybody says they're using AI, the current ones on the market follow a pre assigned rule that's given to them. For example, if customer does a model will do X, all of this is already pre determined. And as we know, people and the market don't follow any particular rule, things are changing people's needs and wants are changing every second of every day, one minute, I go online, I need to do some shopping therapy, and the next minute, I no longer need that particular product, whatever it may be. So the goal of our technology is as soon as you land on the website, it starts to, you know, give you that customer experience that you need the incentives and so forth, to help convert you to become a paid customer. Here is the power of ps3 60. It's always learning and growing with the human. So I'm sure you're wondering how we do all of this. And I'm more than happy to explain the strategies that we take play that the strategies that we implement to actually make this happen. But I want to take a pause here to see if there's any questions or anything that you want to jump in with.

Darius 22:28 Yeah, so so that's very interesting. And I definitely would love to learn more. Let's see if we can touch on some of these details as we talk with Bharat.

Unknown Speaker 22:42 I can use it. Sherry, welcome. Hi. Thank you for expanding great to I actually it's really interesting. So I'm just curious, like how what components that you consider, like how do you get like a, for example, if I'm looking from looking on amazon.com? For example, here? How exactly, you promote the products or something like that. So on a on a technical level, is that okay, is my question?

Darius 23:19 Sorry? If Yeah, that's fine. That's that's Shadi can reply.

Shadi Yazdan 23:25 Yeah. So it actually takes me into explaining my strategies to you what strategies we actually employ. Before I go into the deep into the strategies and explain that to you, I actually tried not to use very technical jargons as much as possible, because we do have people in the audience that may not understand fully the technical capabilities of the way that AI models work. And I am not a technical co founder either. So I will explain it to you in a way that would hopefully resonate. And please, again, feel free to ask any question if I'm not clear. So essentially, as soon as that as soon as that visitor lands on that ecommerce site, based on what they're looking at how much time they're staying, staying on that website, how they're interacting with that website alone, we start making the prediction. So let me go through and tell you the strategies that we use to increase that conversion rate. Now there's three separate strategies that we utilize. However, in some cases, some companies may only need to utilize one of these strategies, or two or all three. Again, it really depends on your ecommerce company, and what you need. The first strategy is called real time digital experience optimization. To explain this really, I'm going to use an example from an online travel company, which majority of us have experience with and reviews. So in this example, a travel company let's say has hundreds of different combination of content image video call to action buttons, and the Want to show these combinations through their main banner, but they're really not sure which combination is going to resonate best with their particular segment. So we create millions of different combination of content, image and video call to action buttons all rearrange, and we put each one in real time in front of the online visitor. And through this experience alone, we found, for example, let's say three images, with particular combination to be the winners out of millions, which combined brought 31% more conversion. For example, out of these three banners banner, one combination, during the experiment resonated with 80% of the customers, which led them to click on that banner and make a purchase. So for example, if 10 visitors saw banner, one combination, eight of them will convert, and banner two and three together out of the three examples that I gave you resonated with 20% of the customers together, all of these three banners resulted in 31% jump in the conversion rate for this travel for this made believe travel company. So essentially, what we're doing is we are running millions and millions and millions of micro changes and tests to put that banner in front of that particular visitor to see what is going to resonate with. And this is being done in real time, very quickly in seconds to see what is going to be most optimal for your particular segment. That's one strategy. Any question about this strategy before I move on? And explain the second and third strategy?

Unknown Speaker 26:44 I think it is good to have follow up question on this, for example, do you also gather the information from the another applications as well like let's say if I'm searching on expedia.com. And that like one one of the behavior that I'm looking, so when I go to another application? So usually the information will be shared across right? And do you guys also gather such information,

Shadi Yazdan 27:18 we actually run our analytics on your particular website. So if we were to run analytics on Expedia, then we would have to be a client of Expedia. So that's completely it's a different world, right. But so we can run and we run our analytics on your particular ecommerce site, because of your particular conversion rate on your particular site. So what you're writing is more like data mining from other providers that are out there, we are not, we don't do any data mining, we don't use any third party data at all to run our analytics, all of our analytics, we run on your particular customers, because the behavior of your customers are going to be different than any other customer of any other website.

Darius 28:10 So I guess, if you are already borrowed, if you're already using third party data in and bringing that data into your own systems, then I guess predictive, I will access that data is that accurate, shoddy,

Shadi Yazdan 28:28 anything we predict based on your own environment, and we predict your customers behavior. So we're not looking at the third party data and how your customers are interacting with Expedia. We're only looking and predicting how customers are interacting with your particular site with your particular products and services that you're offering. This makes sense. Yeah. Thank you. And thank you so much.

Darius 28:56 So one question that I have Shadi, as we talk about your first strategy is, to me it sounds similar to multivariate testing. Is Is it the same what's the difference if it's not actually the same?

Shadi Yazdan 29:13 It's not at all the same as multi variant testing, because multi variant testing gives you certain number of variants that you're testing against one another. We are doing field millions, and sometimes billions of test all at the same time. Whereas multi variant testing has a cap, we don't have a cap. So we're running millions and millions of testing all at the same time in seconds in real time to give you that prediction that you need, which is 100% way more effective than your traditional AV testing and multi variant testing. This kind of testing that we do is currently not in the market, and we're in the process of putting a patent in for a technology that's able to do this.

Darius 29:59 Okay, great. Let's go to the next strategies.

Shadi Yazdan 30:02 Absolutely. So the second strategy that we use is called real time personalized messaging. So as I mentioned earlier, our technology's always trying to understand your customers intent. There are three different types of people that often go online to shop, there's that window shopper who's just researching, there's a returning customer that knows exactly what they want. And they come on to your website, and they buy it. And then there's that hesitant shopper, which we call the golden goose, they need some help, they're unsure, gotta hold their hand through the customer journey, they've had a shitty day, I hope I can say that word. And they just want to come online and they want retail therapy, or they just want to buy something. So in this scenario, with hesitant a shopper once the technology and our AI model has the capability of identifying who is the hesitant shopper, once it identifies the customer as being hesitant and making a purchase during the shopping stage, you'll start to engage with them. So pX 360, knows exactly the right time down to the millisecond, it should engage with a visitor to help them convert to become a paid customer. So for example, it could do this by offer them particular incentive or discount that you've, you know, agreed to give to the customer, any which way that you want to, you know, give an offering to that customer. So now, instead of you giving site wide discount, you're offering discounts based on the particular customers behavior in real time. So you and I could be on the same site, look at the same product, but our behavior is going to be different. And the AI models pX 360, will determine what incentive each of one, each of us should get this strategy alone, on average, is able to boost your ecommerce sales by 15%. That's what we've noticed. Any questions about this strategy?

Darius 32:12 So that's a that's a pretty significant increase in conversion. I'm probably like, not as I'm trying to trying to understand, like, really the the detailed, informative differentiation between the first strategy and the second strategy. If you can maybe go into that, because they kind of seem a little close or similar? Or if I'm probably I'm not understanding it correctly.

Shadi Yazdan 32:45 The first strategy? Good question. So the first strategy, what it's really doing is enhancing that customer experience. So putting in front of you something that's going to resonate with you most. That's really, and that's kind of like very similar to what Amazon does. Amazon makes micro changes on their website that we're not even aware of that are happening just based purely on the way that we are interacting and behaving on the website and the products that we're clicking on. So that's what the first strategy does. The second strategy, the AI models are sitting in the background and watching your behavior. And when they determine that you're a hesitant shopper, and you're not sure about what you want, or you need some help, then they incentivize you by it could be a pop up that says, hey, if you make a purchase in the next hour, you will get 10% off. Or it could be a chatbot if that's what you want. Because we have a client that wanted us to embed a chat bot, another client wanted a pop up, depending on what you want. Because currently, what we see on websites when we go visit them is we see a blanket, you know, you get 20% sideway discount. Well, what if you would give discounts if that's what you're doing to people based on their behavior. So Not everyone gets a discount. It's only those that are hesitant and unsure. Or maybe they added an item to their shopping cart, but didn't go through that whole shopping stage. That's really where this strategy comes into play. I hope I answered your question.

Darius 34:12 Yeah, so it's, it's, it's more like a personalized offers, based on not based on like a predictable, predictive, like formula. But based on what the AI agent is learning about you.

Shadi Yazdan 34:32 While the AI agent is learning and predicting what's going to resonate with you most Right, so the first strategy is really all about enhancing that customer experience. The second one is personalizing the way that personalization should be done. Okay, so many people talk about personalization, but things are not personalized. When you and I go on ecommerce site. I never feel like things are personalized to meet my need. I just go on the site and I'm getting 20% off just like everybody else. But how great would it be if I was offered a particular incentive purely because of my behavior. And I'll give you an example that actually with one of our clients that we were able to embed those of us that live in Canada, we shop online, US companies all the time, what really frustrates us is not so much the pricing and the exchange rate. But what really frustrates us it was really tough for us living in Canada is paying the ridiculous shipping fees that we often have to pay. So if a technology determines that oh shoddy is behavior based on what she's looking at, and what she's clicking on, and how she's interacting on this website, she keeps adding items to her cart, because she wants to get to a point where her shipping is going to be free, then this is something that resonates with Shadi, then I should be giving her free shipping, because then she's going to buy No matter how much item she adds into her card she's going to buy if I give her free shipping, something like that. I hope that makes sense.

Darius 36:00 Yes, definitely. Thank you.

Shadi Yazdan 36:02 So then again, as a customer, I feel like wow, you know exactly what I want, you're giving me exactly what I need. And my chances of going back to that ecommerce site over and over again, is much greater, and that customer loyalty is going to be built because first of all, you treated me like Queen because of those banner changes and the way that you made those micro changes on a website. I really love the way that this site is resonating. And now you're giving me a incentive based on my particular behavior. This thing is completely personalized for me, why would I ever want to go to to the competitor, I'm staying with you forever, because you've already made me fall in love. Which brings me to strategy number three, recommendation in real time. Everybody talks about recommendation. But they're only making a recommendation once the purchase has already been made. Once the person has already made a product, you know, purchase the product. I was actually talking to someone at Cole Han, which is a shoe company. I think they're owned by Nike, if I'm not mistaken. And one of the challenges that they were saying they're currently having is that when someone makes a purchase of shoes online, within a few days, they get an email with the exact same shoes and the same color in case they want to purchase it again, why would I want to purchase the exact same shoe, exact same color, exact same style when I just made that purchase. So this strategy is something they're quite excited about. And we're having talked with them. So sensor technology is continuously learning and making millions of predictions every second on the individual visitor, we know exactly what they want. So for example, let's say the anonymous visitor is looking for a backpack or shoes or whatever it may be. px 360, has already figured out this particular visitors behavior and puts in front of them options of bags or shoes that resonate with them most in real time. If you're looking at this bag, or this shoes, or whatever it is, Hey, what about these other options? Have you considered them, the color is different, the mate is different, whatever else it may be, but it's gonna resonate with you based on your particular behavior. So again, we're recommending in real time, where everyone else makes recommendation after the purchase has been made. They're using historical data, we're using real time data. And that in itself is a huge value proposition right here. And just to take this a step further, you can use the data that's been gathered here and use it in your offline marketing as well. So now, when you're sending email to your potential customers, you're sending them email about products that you've already determined is going to resonate with them. So every person on your mailing list is getting a completely different product recommendation in the emails that they're receiving. How incredible is that? And that my friend, is the third strategy that we are able to implement.

Darius 39:10 Okay, that's very interesting. So I noticed that Bharat is works at Amazon. And I used to also work in the prime team. So Bharat, what do you what are your thoughts as far as how this differs or potentially differs from what Amazon does? On the I'm assuming? Are you working on the marketplace? team?

Unknown Speaker 39:34 I'm not really. I work on a customer success team. So I have no insights on the marketplace side.

Darius 39:42 Okay, all right. Good. Yes. Well, they do. I mean, they they are pro Amazon is probably one of the most advanced customer experience, you know, I guess applications and websites. There are a ton of things happening on the background that we Don't even know. So. So we're always trying to learn more and you know, get get better. Of course, it's, you know, we're not at 100%, we're not never going to reach perfection, but we're all striving to get there. So

Unknown Speaker 40:19 to do so on the apple event that happened yesterday, because they wanted to integrate all the apps with the Apple ID so that applications won't share the data. So I was curious on that part, actually.

Darius 40:34 Yeah, I don't know. I don't know. I haven't heard. Are you talking about the apple event yesterday? Yes, yes. Okay. You know, I haven't caught up with the announcements yet. I'll need some time to, to go in there and talk about it. Actually, I have a room on Friday mornings, Pacific time. 730, where we talk about what happened in retail over the past week. So I'll probably do some research on on the apple announcements and going to some of that on Friday. If If, if you want to, if you have time to join. So. But that's a good question. Yeah. So I mean, this, this whole idea of privacy is also really important. And actually, you know, so shoddy, how do you if if, if you don't take any cross app data from other apps, that makes your solution even more powerful, because it's not dependent on other data from other apps,

Shadi Yazdan 41:39 you got it, my friend. That's why we're poised in a very good time, we actually knew this was going to happen. And this is not new. And for anyone that's surprised by this, they haven't been paying a lot of attention to what's been going on. This is long coming, you know, and we knew that this was going to happen. So one of the most. For us what we did, when we were developing our AI models, we wanted to make sure that we were developing a company that first was not a data company, we don't collect any data, all data belongs to the end user. We actually hold the date, when we're running our analytics, we hold the data for 90 days after that, we delete it. And if the client wants to keep it, they can keep it It's useless to us any data older than 90 days is useless to us. The other thing that we took into consideration was the privacy laws that we knew were coming down. And we wanted to make sure that we were not dependent on any pixels, any third party, and we didn't have to worry about, you know, predicting customer behavior based on any of those third party apps or anything like that. And that was really crucial to the building of this of this technology. And that's why I say we are really poised Well, right now, for any company, not just ecommerce that's looking to use these AI models to be able to more effectively predict their customer behavior.

Darius 43:11 Okay, thank you. So just to do a quick reset, I'm speaking with Shadi as done on the official CEO, and cmo and business development lead at a company called breathe predictive AI co founder. And we are going to go through for another like 10 minutes. And then I'm going to have to jump off and go to my meetings. But what I wanted to ask you about Shadi is that so what is it typical implementation? If somebody is interested in trying out predictive AI? What's the process? Do you have like pilot programs that they can try the software for? You know, like to maybe a few months?

Shadi Yazdan 43:57 Absolutely. But first, I wanted to go back to the unofficial SEO because some people might be wondering what's going on there, I'm being pushed into the role. And I'm not 100% convinced that that's the role I want to take. But I'm actually falling into that role naturally just based on what I am doing and all the business development and and the clientele that I'm talking. So that's really where that where that title comes from. But I am I've been the CMO and now the CRL. For for predictive. I

Darius 44:29 will confess, I'll confess that I'm the culprit, one of the culprits of that so so yeah,

Shadi Yazdan 44:35 you're the one that's Yes, constantly telling me I should really step into the role. And and you're not just you're not the only one. I have heard it from some of our

Darius 44:44 I'm not surprised. Wow.

Shadi Yazdan 44:45 And I'm just you know, going back and forth, back and forth with that. So in terms of integration and how that would look like first and foremost we work with midsize ecommerce companies. So for example, If you have a less than 20,000 customers already that are buying from you this technology isn't going to work for you. That's, that's first and foremost. Now, how many traffic do you have on your website, you could have 60,000 100,000 traffic. But if you only have 20,000, at minimum customers, then we can work with you. But we oftentimes take the average the last six months average of the customers that you've had the transactions that you've made, to see if you have enough customers, for us to be even able to run our analytics and do that prediction if, if it comes to that point. So that's number one. The second thing that we look at is we look at your environment. Where is your e commerce? What tool are you using? Are you on WooCommerce? Are you on big commerce, all Shopify, we can work with majority of e commerce platforms that are currently out there. I say majority, because there are some that I've learned about that I had no idea existed, like yahoo shopping, that's like a technology over a decade old, that will not work really effectively with this technology. So what we have to do is we have to take the client, migrate them, bring them into one of the other current ecommerce platforms, and be able to then run the analytics from there. So we do take about a week just to understand your environment, make sure you have the customers that you need to have in place. And once all of that is in place, we charge onboarding fee. And then we also charge monthly fee for technology. So give you just a quick little what that price would look like. It would be if you have anywhere between, let's say 15,000. And we've been able to work with customers that have 15,000 customers, even though I say 20,000, because I like to be more conservative. So anywhere between 15,000 to 20,000 customers that if you have that on your website, that's those are the customers that are making transaction, we charge you a very small fee of $700 onboarding fee. And then every month, it's about $1,000. To use this technology. Why such a low price point, because we want to prove to you that this technology actually works. And we don't want the price to be a hurdle. And we're still a startup as well. So it gives us the case studies we need it helps you get the results you want. But we actually don't charge you the fee until the end of the month. So we most companies charge you beginning of the month, we charge you end of the month, because we want to give you a month to try out this technology to see the results for yourself before the end of the month. If you don't see any change, no, nothing has happened. Your conversion hasn't gone up. You can pull out and say I'm not interested anymore. You haven't lost a penny. And we're offering that service right now as well.

Darius 47:52 Okay, so that's good. I think it's very important to be able to let you know, potential customers test and verify as easy as possible. So that sounds like you've got a good, good program there. We'll take another question mere welcome.

Unknown Speaker 48:11 Thank you. Very nice to amin, this lady I'm a really nice conversation going on. And also good conversion by Shadi. Okay. I will also in this base of recommendations. And I've been working for a company for naval data analytics. So I've been working on this space where we're trying to create some personalized for a client of ours do some better personalization. So wanted to know how predictive AI is doing their personalization, like Do they have different use cases where some kind of recommendation would need a session based window, whereas later on, you might just need a particular user based recommender? So your thoughts on that? Do you have different? Do you have different use cases? I rotating all those use cases separately? Or how is that working out for you? So

Shadi Yazdan 49:12 the three strategies that I went through and one of them was that recommendation in real time, every single one of those we have use cases of the beta customers that we've worked with and what we've been able to achieve. And that's where I was able to give you those numbers in terms of how much we were able to increase their ecommerce sales by and how much we were able to increase their even even their, their conversion rates as well. So I'm not sure if that's what you're asking. But for every single strategy, those are the types of use cases that we have. They're currently not published those use cases, as they're almost in the completion phase. So once that's done, then we can make them public.

Unknown Speaker 49:56 not asking about the different use cases or asking about like There are 10 different kinds of recommender right in the market which will be used for different use cases. Suppose. So there is one thing which YouTube uses, it uses a session based recommender. So, during my session if I'm looking at some videos related to one sports, and then I just scroll down, and I get recommended videos and the similar category, which is more f1 sports, right? So, so if I'm browsing for a Nike issue, then I go scroll down, and I see a bit of attributes. And then I see same kind of recommendations. So that's one type of recommendation. And then you have memory based recommender, like, if I'm a returning user, I use your attributes. And also there are hybrid models, which will use sessions plus your previous experience on the platform. So So def predictive is like taking care of all of this or how does, like, How are your algorithms defined? Yeah,

Shadi Yazdan 51:03 good question. So currently, we do recommendation in real time, based on what the particular person is looking at on that ecommerce site. So I use the example of the backpack. If you're looking at a backpack, then we're able to make recommendation of other types of backpacks that could resonate with you most. So that's how we're doing our recommendation in real time. This technology currently is ready to be implemented by the e commerce companies based on the area of needs that we've seen. And those were the strategies I follow. Beyond that, how would the technology work, we were actually talking to a number of potential clients at the moment to modify these AI models to fit their particular environment. So based on what the client says they need in terms of that recommendation, we can modify the models to do what they needed to do in any environment that they potentially currently have. But at the moment, the AI models are ready to go out for that ecommerce space. I hope that answers your question. And thank you so much.

Darius 52:13 Yeah, it sounds like it is session based mirror, if that's what you're referring to stay in the corner looking for your next session. Right, right. So well, thank you so much. We're coming up on time and I have a hard stop. So I'm sure Shadi is extremely busy as well. So thank you so much, Shadi for taking the time to tell us about the predictive AI and your fascinating background. I am going to put this up on the website in the next few days. So if anybody wants to didn't catch up the you know the, the the full interview, if you go to retail tech podcast.com you will be able to see listen to the entire interview. Again, thanks a lot. Shadi, Bharat and Mira for coming up and asking questions. Thank you so much for having me. I really appreciate it. Thank you Have a wonderful day, everyone. You too. Bye bye. Thank you.

Interview with PredictivEye CMO Shadi Yazdan on using AI in Conversion Optimization

Related Interviews

sponsors