Microsoft Accelerator Podcast with co-Founder & President of APEX Technologies, Tiger Yang
“We Know What You Will Buy in the Coming Months”
Source: Published by Microsoft Shanghai Accelerator, April 18th 2019
Transcript
Summer:
Hello fellow listeners. Welcome to a new episode of the China innovation decoded. In the age of AI and big data, and with those gradually turning into buzz words, sometimes it is rather difficult to make sense of it.
Today, we have invited Tiger Yang. Co-founder of APEX Technologies. Tiger and Jimmy, two co-founders at APEX Technologies, actually met eachother in their freshman year as lab partners for electrical engineering and computer sciences at UC Berkeley. They started developing their interest in enterprises on improving efficiency in the marketing field. And when they decided to come back to China, APEX Technologies was born.
Tiger:
Both myself and Jimmy, we feel that China is the right market to start the business. Because we focus on B2C enterprises in retail like Starbucks Coffee, Michael Kors as a fashion brand, Maserati as automotive. So we focus on B2C, and the Chinese market. And especially the consumption part of the China market is definitely growing very rapidly. That’s one, and two, you know, even if if the economy grows actually sort of slow, it slowed down in recent years. you know the need of using data, as well as AI to make sure that operations and marketing is more real-time, more personalized, to ensure that these enterprises can continue to grow their revenue and profit. The timing I think is very much ready for us to do this project. In north America there were some companies before us, that was probably in their series C or D funding phase, before we, at an early stage of us coming back, but we are the first in China to provide data-driven customer data platform (CDP) and using machine learning to do predictive analytics on marketing etc.
Summer:
Yeah, cause I know that, as you might have known that, Big Data is a buzz word and it has been thrown a lot in recent years. And there always has been this debate about not just collecting data but also making sense of it. So how do you do that at APEX?
Tiger:
Right, that’s a very important question. I think everyone talks about, you know, about ABC or ABCD etc. right? In 2015 or 2016 it’s Big Data, and then AI in 2017…
Summer:
And blockchain…
Tiger:
And blockchain etc., yes. And, now, both concepts, as well as, we do see that, the actual use cases, as well as the value it would bring to no matter if it’s basically customer or enterprise level, you know for us we focus on the scenarios surrounding customer data. Especially marketing. So, here’s how I feel about it. Ultimately, no matter what new technology it would be, you know, big data, AI, or blockchain, it would have to serve a business purpose and business value. To create real value, no matter if it’s for business kinds or customers.
That’s the key, or the fundamentals of a successful startup or business. And that’s where the real value basically arises. So, I think what we’re specialized on and how we’re differentiated from all these big data companies, is we’re focussed on a specific scenario, which is helping these B2C enterprises, especially mid to large enterprises, using their data. Using their own customer data. And make sense of this customer data and make sense of this customer data. From “if they have not collected this data” to “if they have not integrated this data together”, because nowadays the touch point with customers varies so much that they can’t recognize if it’s the same customer if they log in on mobile applications or an offline store. So, it’s important to break these data silo’s and ensure that enterprises have a holistic and unified view of their customer profile. So that, with that data they can use this to enable personalized marketing. And through the help of some machine learning algorithms, for instance, let me share a few cases because that might important to how how you understand…
Summer:
Yeah
Tiger:
For instance, as we mentioned, Maserati is one of clients in the automotive industries, so, you know, we have connected all their data points where they have customer data. Basically, their CRM or DMS etc., all these systems, WeChat etc. So, first of all we have all this data integrated into our CDP, our customer data platform. And then, what you wanna do is to predict who the customers are who are more likely to buy a Maserati within the next three to six months, as their average sales period. Otherwise if they would start calling these potential customers, they might end up calling the potential prospects who do not have an intend to purchase a car in the near future. So we use machine learning to predict the likelihood and possibility of potential customers purchasing a Maserati based on the Maserati owners in China. Meaning that they have around a 100,000 Maserati car owners that we can study all these features and trades of, all these data fields of these clients and customers who have already purchased a Maserati, and using that to predict their prospects in our system to who is more likely to purchase a Maserati. So that’s one example of increasing marketing efficiency.
Summer:
Do you do that in China as a whole, or do you segment it into different areas about like the purchasing behaviors, and then I guess for countries outside of China since you’ve expanded to other countries as well. So you’ve also… I guess it’s different from example Chinese general markets and US markets in terms of buying a Maserati.
Tiger:
Absolutely, absolutely, yes, for this client we mainly focus on helping them on the Chinese market first. But definitely, the region, or, there are a lot of factors that would impact or attribute to if a potential customer would purchase a car or not. For instance, there are some data fields that by, you know, basically human instinct might not be that important as we think would impact anyone’s decision. But you know, we would look at everything such as gender, female… because one of Maserati's largest customer segments is actually housewives. Housewives is actually one of the largest purchasing customer groups of Maserati
Summer:
Hmm. I didn’t know that before, haha!
Tiger:
Yeah, so besides all this, what we call basically, wealth inheritance. That is the second largest customer segments. So we look at all of this, not only including gender, but also which type of companies this person would come from. Whether it’s a global multinational company in China, or a state-owned company etc., all those fields would have an impact on whether this person is more likely to purchase a Maserati or not and what we do is, we train the machine not only algorithms that decide for itself. Basically using all these data especially, we will tell the machines and the algorithms that we have, in the training set of customers, who is a Maserati owner, and who is a # and after that they (the AI) will assign # in determining as what would be a score of each person, their likelihood from zero to a thousand of purchasing a Maserati. And when the call center would make phone calls to invite the customer, to arrange test drives, you know, then it would be basically targeting customer with higher scores first because that would increase the efficiency. And the sales representative would look at the people with a very apparent change in their scores, because they might have test driven yesterday or they might have just requested a quotation online etc. Those would definitely show their intention of purchasing a car.
But that’s one example. There are plenty of examples, for example we help airlines predict what would be the next destination of the customer. So you know, when they do marketing they would not do it based on where they just went to. So for instance, if this person went to Bali or Thailand, you don’t want to advertise or market, you know, Thailand for the next three months because they might not go there short term. If it’s for vacation purposes you should pick out something else based on the data. So with the prediction of the next destination. So for Maserati and airlines we have achieved 95% to 98% of accuracy rate of prediction.
Summer:
Oh, wow.
Tiger:
That’s actually pretty high, and using these models with data-driven, and all these models, you know basically as I mentioned we’re helping a Chinese airline increase more than 10 times of marketing conversion rate of purchasing a particular air ticket, based on our prediction. So, there are a lot of examples like that. As I mentioned Starbucks Coffee, you know if you want a latte or a, you know any kind of coffee, or if you want a snack as a kind of combination etc., all these things can use personal recommendation models to basically make everyone’s experience and marketing experience and customer experience personalized and different. And that’s how we mainly use big data and AI to both enhance the customer experience, as well as help enterprises increase their marketing efficiency. Reduce their marketing budget, but more importantly, even if the budget is set, you want to increase the conversion rate, the loyalty, retention rate, and if the client for some reason might be lost, you want to hopefully basically identify identify that before they are lost and try to make sure that this customer doesn’t go away. Et cetera.
Summer:
Is there a particular industry that you find most challenging to tap in or collect data from?
Tiger:
Right, so supposedly, it’s actually easy to understand, you know, why the industries if they have more online touch points with clients, it would be easier because all the data is online and you can just use coding to get all this data to come in. The industries that focus more on offline would be the more challenging industries, but you see that actually, you know, retail as being one of our largest kind of, what we call new retail, especially these last two or three years, these kinds really are already ready to… they already understand , even without us telling them, how important their data is and how important AI is. You know, we’re covering as I mentioned not only restaurants or food & beverage industries like Starbucks or ???, another client of us, as well as fashion apparel like Michael Kors etc., and you know, big super markets and eh, like Walmart. So basically, you know, we see the trend of traditional, even offline focussed retail industries, as well as one of the industries not even on the list that I mentioned is the real estate industry. We have met so many real estate companies, especially in Q4 last year and Q1 this year. They are also passionate in basically digital transformation data-driven, AI-driven decision making. You know, if they have a piece of land and they want to build a project, they want to figure out how many two-bedrooms, and three-bedrooms, and four-bedrooms there should be. Based on customer data of past experiences, of you know, constructions in the same city.
Summer:
You mean before they start building something? They want to know.
Tiger:
Right, because, you know, personalized marketing is definitely one of the largest use cases of our product and platforms and solutions but, you know, besides personalized marketing, we also help, since we already have integrated all this customer data for our enterprises, their internal decision making can be data-driven based on the customer data, customer centric view data. As I mentioned, these automotive manufacturers not only want to increase their marketing but also want to see if they’re going to make a new car, what would be the colour or the exterior or interior of that car and what the customer actually cares about.
And so its the same thing with real estate. There is a concept called C2M, customer to manufacturer. That is also rather a buzz word or — using customer data can not only increase marketing efficiency, real-time decision making based on the data, internal decision making, but also, to manufacturing, the supply side of it and as I mentioned, the real estate example is a C2M, based on customer needs, how many bedrooms they’d need or how large the kitchen should be.
Summer:
That’s true and that is interesting, but I was thinking, going back to what you were saying about, for example, the colors of the Maserati — that’s rather specific, because many people like different colors so why would that be an important factor to focus on?
Tiger:
Right. So, that’s one of the products that we actually have. As I’ve mentioned, we have a Customer Data Platform (CPD) product, and we use machine learning to predict customer behavior basically whether they would be more likely to purchase the product or not. We do have a third product, based on natural language processing technology, NLP technology. What we did for one of the electronic car companies — after their new car is on the market and people start to get a feeling of how the car is, we start collecting all this customer feedback for them. These could be comments and feedback they leave anywhere, so long as its on the public internet. You know, what are the customers talking about — are they talking about the battery and how long the battery can last before they have to refill the battery or the color, as I’ve mentioned, the color of the car — is it ugly or pretty, as well as a lot of other stuff. What we do using technology is to make sure that the decision making level, management level people, can get all this customer feedback real time, they can real time look at all this customer feedback, no matter if its internally in their system or anywhere on the internet. So using that feedback, they can think about, in the next car model, what they should be working on. Ultimately, the goal is everyone could, potentially, be delivered to a different car or that customization can be maximized.
Summer:
So if many people are mentioning the colors, that might be a trend they can focus on for the next car release?
Tiger:
Exactly. And its not only about the product, the services are also important and sometimes it could be a security issue that they have to focus on or a service that they can immediately change. Service mainly because we have clients like Didi Chuxing, as I’ve mentioned. So, if its a service problem that people are complaining about, a particular service issue, then they should actually know that immediately and try and see if they can change the service to enhance the customer experience, etc.
Summer:
I see, I see. So how do you differentiate in this large pool of companies that are also focused on using AI and Big Data. How is your company different from, for example, Facebook’s efforts to tap into big data?
Tiger:
I think that from the larger picture, there would be a few types of companies and how every company differentiates would be as following:
Some of the companies would probably be building the infrastructure of big data or even cloud computing like some of these startups that have incubated in the Microsoft Reactor. So they are focused on the infrastructure part, and we’re not — we’re focused on the application of AI and big data. At the application part there are different specialties — we focus on, especially, marketing and the customer side of it. There are companies that focus on anti-fraud and risk management, fintech companies, etc. There are definitely companies that focus on using AI for other purposes and applications and we focus on, as I’ve mentioned, marketing etc.
So, that’s one aspect of segmentation — the second is “What kind of clients you would serve?”. As I’ve mentioned, we serve mid to large B2C enterprises, especially coming from a traditional industry but they really want to do digital transformation that would help them to go to the next level and transform and face the competition of internet companies etc. There are some companies that might focus on internet companies, and small to medium size companies etc. So we feel its different, its actually very different at those sizes.
The third aspect I think is very important is “What industries you would focus on?” because these companies would want to work you because, for instance, we have more experience and more case studies in automotive industry or in the retail industry. We have done similar cases and our models and our machine learning algorithms are more accurate because we have ran similar scenarios in the past for this particular vertical. We believe that in the B2C field, you would have to do one industry after the other — basically, you have to focus on a particular set of industries, and that would actually increase your know-how for that industry and that’s why these industry-level companies would select you as their partner.
The fourth aspect to it would be, I think, that big data and AI doesn’t limit our scope, because we’re solving a particular scenario’s problem. The list of technologies being used could go on to blockchain, etc. The technology has always been evolving, there is gonna probably be A,B,C,D,E in the future. So thats always gonna be evolving, but the problems we are solving might be similar and you’re working with a similar set of clients that you can provide with more technology and more solution and help them solve more problems.
Summer:
What do you think of the future of AI and big data? We are learning about past data stats and predict what people will choose or behave in the future but do you think that will help people to be more innovative or that it would prevent them to be more innovative in the future?
Tiger:
Its an interesting question. First of all, I would mention that, for instance, especially in marketing, we have used data-driven, AI-driven decision making to help achieve whats the core to marketing, which is at the right place and timing, to the right person, deliver the right message. That’s the core to marketing and using data, using AI is definitely gonna increase the efficiency, but it does not replace the creativity part of it, it does not replace the innovation part of it.
That part of it is still gonna largely rely on humanity because that would actually compress to around 30 to 50 percent of the entire marketing efficiency, and that’s not gonna be replaced in the near future. At repetitive work, as well as predictions, especially when there is a lot of data fields that can be collected, AI can do much better work than a human being, that’s for sure, but I think it would not replace human beings on the creativity part, etc. It will actually, in my opinion, help human beings focus on the more important work of innovation and creativity, etc. That’s my view on it. It definitely should be a help to humans, it will reduce the amount of repetitive work that humans needed to be doing in the past.
For instance, given the example of Didi Chuxing, they used to do all this categorization by pure human labor, but when the data increased ten times, and data will definitely continue to increase — right now we don’t have to do testing on samples, its all entire volume of data on customers — using AI is something inevitable and its something that is definitely in the trend and its helped all these humans beings with whatever repetitive work AI can do it can even do it better than human beings can, with more time saved and more energy saved.
Summer:
I totally agree. I do know our time is running out but I do want to ask you one question — could you give some advice to the many people like us that have studied abroad and want to come back to China, to start a startup. Do you have any general advice, or some things that you have found challenging in the past that you want to tell everyone?
Tiger:
Very interesting question. I believe, yes, there definitely would be a lot of people who studied abroad and want to come back to China and I believe the common reason why they would want to come back to China, especially to start a business or a startup is they think the Chinese market has potential, a very vast potential in the future. Some advice, I would think that its definitely a very interesting experience after spending ten plus years abroad and coming back to China. There are things that you would probably not think of, but you would definitely have to get accustomed to the Chinese market because every market is different and probably spending some time abroad does not help — it helps things otherwise, it does not help in the way of familiarization with the Chinese market. There’s definitely differences, as we ourselves encounter when we’re serving multinational global company clients- Its different, the way that people think, their mindsets and the way they do business is different. I would say that when they come back to China, they would have to focus on, that is, you have to get back to the localized thinking, the mindset of the Chinese market. I think there is only one way to succeed in China, and that’s the Chinese way of succeeding in China. There is no two ways or “I can take advantage of the global experience” — that is a very interesting experience, but remember, if you’re serving Chinese customers and Chinese enterprises, think in the same way as they are.
Summer:
Thanks so much, thank you for being on the show.
Tiger:
Thank you.
Feel free to scan the WeChat QR code below if you’d like to subscribe to the Microsoft Accelerator podcast!
