Uncover Hidden Value for Customers, with Alibaba.com's President

0:00When someone comes to my door, I know, because I'm using SimpliSafe. I get an alert straight to my phone and I can see, is this someone I should unlock the door for because I know them? Or is it someone I need to tell to get lost? But here's the thing I love most about SimpliSafe. It works when I am not looking at my phone too, because when I'm not keeping an eye out, SimpliSafe is. SimpliSafe is proactive with a double layer of defense that stops crimes before they start. First, AI-powered cameras spot potential threats outside. And then live agents step in, talking to the person through the camera and letting them know they're on video and police will be dispatched if they don't leave.

0:45They can even trigger a loud siren or spotlight. That's how you help stop a crime before it starts. And SimpliSafe makes it easy to use. There are no long-term contracts, no hidden fees. You can cancel at any time. It was named Best Home Security Systems by U.S. News and World Report for five years running. It offers a 60-day money-back guarantee, so you can try it and see the difference for yourself. This month only, take 50 % off any new system. This is one of the best prices you will ever see for SimpliSafe, so don't miss it. Just go to simplisafe.com slash problem solvers. Again, that is SimpliSafe, S-I-M-P-L-I-S-A-F-E, simplisafe.com slash Problem Solvers and lock in your discount.

1:33There's no safe like SimpliSafe. From Entrepreneur Media, this is Problem Solvers, a show that solves entrepreneurs' toughest problems. I'm Jason Pfeiffer, Editor-in-Chief of Entrepreneur Magazine, and each week I take a problem that you're probably dealing with and get tactical solutions from people who have been there and solved that. So let's get solving. Recently, I was talking to the president of Alibaba.com. My name is Kuo Zhang. I'm the president of Alibaba.com. We were talking about AI and these new AI tools that his company has rolled out. And then he said this thing that I just became, I mean, kind of obsessed with.

2:12He said this. In this world, there are two type of questions. One question, the answer is verifiable. The other is unverifiable. All right, so let's review. Kuo says there are two types of questions in the world. One question is verifiable. The other question is unverifiable. And what does that mean exactly? So the verifiable questions like mathematics, physics, or computer code. In other words, you ask a question and you find a clear verifiable answer. One plus one equals two. Go verify it. It is true. But then there are the unverifiable questions. And that's a question like, I want to do a product design, do a product prototype.

2:59Why this prototype is better than the rest? Haven't we all been stuck on these kinds of problems? In fact, aren't these kinds of problems basically the definition of the unpredictability of entrepreneurship? We're looking at multiple versions of something that we made and considering whatever, which product to launch to market or whatever, or which just which thing we want to take out into the world. And we know, we know that one of them is going to be better than the other. And we have no idea which one, which one will work, which is better, which will people like more? The answer seems unverifiable.

3:35Now, there's no such kind of, how to say, formula to give you a very quick result to prove that in ground truth that it's right. So there's no ground truth for that. Now, this is a nice observation by itself, right? Some things are verifiable, some things are not. Kind of hard to argue with that. But what if? What if you could take questions that used to be unverifiable and make them verifiable? How powerful would that be? And how would you even do that? Here is a related question that I've been obsessed with lately. Let's just put verifiable and unverifiable to the side for a second. We will circle back to it in just a moment.

4:17But okay, every so often, we discover a new source of information, rich, detailed information that was always there in front of us, but that we weren't previously able to see or to recognize. For example, DNA. The DNA that is inside of our bodies, a rich, detailed source of information that for most of human history, we had no idea existed. And once we did, it could tell us so much. A hair picked up at a crime scene in a pre-DNA world had very limited value. Now it can potentially tell you who the killer is. A tooth found in an ancient burial site could tell you just a little bit about the people who lived in those ancient times.

5:03Now it is a fountain of knowledge. In the past, we could only wonder about our ancestors or spend years or a small fortune trying to track down little pieces of paper that trace our family lineage, and now a simple swab of our cheek can tell us who we are and where we came from. DNA. It was always there. We had no idea. And once we discovered it, DNA turned verified. There are other examples like this too. tree rings, ice cores, brain electrical activity, cosmic background radiation. And I often wonder, what other examples like this will come in the future? What information surrounds us right now that we just don't know how to see?

5:46Is it something buried in our bodies, in the air, in the surfaces around us? I don't think this is a far-fetched question. I mean, for most of human history, we had no idea DNA even existed. It's not like we were looking for it and couldn't find it. We didn't even know it was there to look for. And it cannot be the only thing like this. There are other sources of information. More of them. More of them to come. Sources that take what we don't know and turn it into what we do know. Which now brings me back to my conversation with Quo and the idea of verifiable and unverifiable. Because Quo thinks of AI as that new source of information.

6:29It's something that I hadn't really thought about before because, you know, AI doesn't actually create new information. That's kind of the whole point. AI is only as good as the information that it's fed, which means that it's just arranging existing information that humans already created. But that by itself is also a new source of information. Like to see patterns that you weren't able to see before, to predict things that were previously unpredictable, that is new information. And how do you productize that? Well, here's a great example. Alibaba.com recently rolled out a new product called Axio.

7:08That's A-C-C-I-O. And it works like this. You give it an idea for a product, and then it basically just does all the work of bringing it into existence. Quo gave me an example. He said, let's say that you go to Axio and tell it, I want to create clothing for children with Attention Deficit Hyperactivity Disorder. It will analysis about this request, do market research, and understand what kind of a specific key element there are, and give you the product design, and give you a breakdown of the product design, and then list the kind of suppliers who can help you do that. What you need to do next is verify this business plan, whether it's valid or you want to do some change, and then click send inquiries to relevant suppliers.

7:59That actually simplifies the whole product design and market research to a few clicks. In other words, you have an idea for a product, just a really basic idea, and any idea for a product used to come with a bundle of unverifiable questions. What should this product look like? Who would want it? Would anyone buy it? Who's going to do the best job of making it? And then Axio basically answers all those questions for you one by one by drawing upon the mountain of data that Alibaba already has about which product features and designs work best in the market and who manufactures them the best. You just like you can tell the engine, just say the words what I'm going to do and it can do the to rest.

8:49Which is pretty cool, right? Today on Problem Solvers, we're going to do something a little different than usual. Quo walked me through the process, like the thought process of developing Axio, and I'm going to play that for you, but this isn't just an interview episode. You will hear me and Quo, but you will also hear a lot of me unpacking this idea, this idea of the unverifiable question becoming verifiable and helping to apply it to your own work because this is a big idea and a really useful one too. And it can help you think differently about the role that AI can play in your business. It's all coming up after the break.

9:32All right, we're back. So today I'm talking to Kuo Zhang, the president of Alibaba.com, about this idea of verifiable and unverifiable problems and what we can learn from his new product, Axio. But before I get back into the Alibaba story, I want to add a little more color to this big idea. Because as I thought about it, I realized Kuo was not the first person to tell me about a framework like this. About five years ago, I had a very different kind of conversation with a woman named Elizabeth Wardle. She teaches writing at the University of Miami in Ohio and just came out with a new book called Writing Rediscovered, Nine Concepts to Transform Your Relationship with Writing.

10:15And we were talking about education and how people were taught to write. And I was telling her about how despite growing up to be a professional writer, I don't feel like school prepared me to be a good writer. Like the way that I and we all were taught to write makes no sense, actually is kind of infuriating. And so here's what I told her about that. And then her reply. I remember having a realization later about how weird it was that I was taught things that have no practical future applications. So for example, that there was a certain structure to a essay that I remember being taught, right?

11:02Like this is an essay and it has these paragraphs and this is the paragraph that does this and this. And then eventually I, I mean, I don't know what happened. I guess I went to college or something and I looked around and I was like, wait a second, that doesn't exist outside of school. You taught me something that has absolutely no practical application in the real world. Why did I learn that? So what's going on there? Can you diagnose that? Yes. So you're talking about the five paragraph theme, which we absolutely hate. But it's easy to assess, right? So I like to talk about it in this way. There's well-structured problems and there's ill-structured problems.

11:45Well-structured problems have one right answer. Two plus two equals four. Ill-structured problems do not have a right answer. Every writing problem is ill-structured. There's a bunch of ways you can do it. Some of them will work. Some of them won't. They're more or less effective. But school, because of the system I was talking about, really likes well-structured problems because they're so much easier to assess. So what you've just named there, the five-paragraph theme, is a great example of trying to take the ill-structured problems that always go with writing tasks and make it into a well-structured problem because I can assess it easily.

12:20Do you have five paragraphs? Do you have an intro? Do you have one sentence that names your three points? And do you summarize your three points at the end? That's a well-structured problem. But nothing in the world ever gives you a well-structured problem that you can usually solve with writing. Yes! And there it is, right? Quo's verifiable and unverifiable problems are Elizabeth Wardle's structured and unstructured problems. Same basic idea, expressed and applied differently. And in Elizabeth's case, she's arguing for keeping ill-structured problems ill-structured, or really for teaching people how to navigate ill-structured problems by embracing that they are ill-structured, because there's no one right answer with writing.

13:05You cannot tell someone, this is the only way to write an essay, so I'll teach you this one structure and then you can always write to it. That doesn't work. Writing has infinite variety, as it should. But in business, there is actually great money to be made by creating certainty and structure. Business is a game of risk. Anything that you can do to lower the risk and therefore lower the cost of taking that risk can directly translate into dollars earned. Which means that if it's possible, you actually do want to err on the side of verifiable questions. You want to make everything as verifiable as possible.

13:45Which brings me back to Kuo. When he was telling me about Axio, which, as a reminder, is this AI tool that'll do your consumer research for you and design the optimal versions of a product and find the best manufacturers, I had to ask him, what was the process of developing this? Like, how did Alibaba.com start to think about where it could add new value to its users and how to find questions for them that were once unverifiable and then make them verifiable? That is functionally the service that this is providing, right? It's like saying, hey, you entrepreneur, you have a unverifiable question and we are going to try to add data to it and make it at least a little more verifiable.

14:29I mean, it's not going to be perfect. It can't tell you exactly what's going to work in the market, but it can give you more information and it can help you make a more informed decision. And that shifts from unverifiable to verifiable. But how do you even start to think about that? How do you realize that this is the kind of product that you should make for your consumer? And he said this. So what we can add value in there are in three kind of factors. The first one is Alibaba.com or Axio see a lot of products, 280 million product in daily basis. and we handled like more than$60 billion transactions every year.

15:14So what type of products look like a kind of a bestseller or winning product? So it has some sense there. The second part is about the industry domain know-how or expertise. So like doing a good prototype design, at least what are the factors you need to weigh in? So I need to do the market the research, know who is doing similar stuff and what the pros and cons, and then start from there, we can do some improvement. If it's not kind of pure, nothing exists, you create a completely new category. Sometimes it's just an improvement of an existing product. So if you weigh in all these kind of factors, more or less, actually, you are getting out some result in a good way.

16:02And the The third one is iteration. So in the end, the product will go to Alibaba.com. Somebody is going to produce it or you're going to sell on Alibaba.com. So we'll see the sales. So it will iterate the model and the kind of industry about its domain expertise part to improve the model itself. So in every day, actually, this model just keep improving since we see enough And we have in the street domain know-how to see these factors that you need to weigh in. And also, in the end, it's kind of closed loop that we see the result and we can keep improve and iterate. All right, let's debrief here.

16:44What did Quo and Alibaba.com really see here? Alibaba's users are people who create products and need to find manufacturers and other sourcing for those products. And these people often face the same unverifiable problems like, Will this new product resonate with my consumers? What kind of design will break through? Which suppliers will deliver the best quality? Historically, the only way to verify any of that was to just launch and wait, which is costly and slow. You can blow your life savings on the wrong hypothesis. Quo's point is that if you break down the categories of information that you have access to as a business and then match those categories to the questions that your consumer is asking, then you can create a higher degree of certainty that didn't exist before.

17:32You can, for example, model human behavior at scale. With trend prediction and demand forecasting, you can compress time by simulating thousands of scenarios in seconds. You can aggregate dispersed knowledge by drawing patterns from data that no human could actually parse. It's just too much of it. Now, Kuo didn't say this is perfect, because of course it isn't. the world contains too many unknowable factors. But AI can move questions up the verifiability spectrum, which is a phrase I just made up. The verifiability spectrum, what once required months of market exposure, very at the bottom of the verifiable spectrum, can now be tested in minutes with data simulations or generative prototypes.

18:17And that is really interesting. That is a really interesting way of thinking about AI is just moving things up the verifiability spectrum, either for you, like the decisions that you make internally, what things used to be unverifiable that we can now do a better job of, if not verifying, at least moving up the verifiability spectrum. Or how can you offer that service to your consumer by saying, hey, I know that you have unverifiable problems that keep you up at night. here we can help them be a little more verifiable. So how does a founder use this to their own work? Here's what I think. It starts by going through a process like this.

18:58Step one is listing your companies or your customers unverifiable problems. Examples are like, will customers like this new feature? What's the best pricing model? How should we expand internationally? Then number two, step Step two, break that down into components. Unverifiable questions are messy because they contain multiple unknowns. So break them into sub-questions like Quo did that might be testable. So if you take will customers like this, you can then look at what language do they respond to in our copy tests, as well as what price points drive click-throughs. And then step three, ask what part can AI now help make verifiable?

19:41Because everything that I just said above is also something you can learn through consumer research and testing, and you should, you should never, ever stop doing that. But how can AI help? How can it speed it up? How can it make it cheaper? Well, you could use AI for synthetic market research. You can simulate demand signals by analyzing consumer chatter or running microtests. You can use AI for design iteration. You can generate hundreds of product variants and run preference tests. Or you could use AI for prediction models, forecast supply chain risk or customer churn with data you already have.

20:18And then finally, step four, iterate faster than ever. AI doesn't eliminate uncertainty, but it shrinks what we might call the unverifiable zone. I like this, right? We've got the verifiability spectrum and now we've got the unverifiable zone. I'm just making up phrases this episode, but it shrinks what we might call the unverifiable zone, right? you get what that means immediately, that you don't have to be stuck in this very large space with unverifiable questions. You can spend more of your resources on high quality bets instead of shots in the dark. There are all sorts of ways that founders can use this.

20:55Like product development, instead of guessing which features customers want, founders can have AI simulate user journeys, test prototypes in virtual environments, and flag friction points before launch. You could use it in marketing and messaging. Instead of unverifiable brand voice experiments, AI can AB test thousands of variations of ad creative across synthetic audiences and real-time data. So you don't have to wait months to see what sticks. And hiring and team design. Culture fit is notoriously unverifiable up front. AI-powered assessments could surface patterns like what language you use and decision-making styles that could drive a more thriving environment in your company, which turns an ill-structured decision into something more structured.

21:43Anyway, in fitting with the theme today, I shared all of this, all of this stuff that I've been thinking about with ChatGPT, and then I asked it for an example of a company that is trying to turn unverifiable questions into verifiable ones. And it gave me a really interesting answer. It said Netflix. The entertainment industry has always wanted to know, what will people want to watch? But that was a largely unverifiable problem. It is why Hollywood just pumps out so many movie sequels all the time. Because if the first Spider-Man movie does well, then it's likely that the second Spider-Man movie will do well too, and that's the closest the creative industry can ever come to a truly verifiable insight.

22:24And then Netflix took a different path. First of all, it built out a strong AI-driven recommendation engine that helps people find things to watch based on what they've watched before. This means that people keep their Netflix subscriptions longer because they can constantly find the service to be valuable. There's always something new there that they can say, ah, this is a reason to hang on to Netflix so I can watch this next thing. But Netflix is also using its viewer data to create the entertainment in the first place. I don't know if everyone knows that. They learn things about what people like and what actors work best in different markets, and then they greenlight projects in part based on that.

23:00It's not perfect, of course. There are too many factors to control for, like changing consumer interests and the difference between a script and an execution of filming and so on. But, you know, the results don't lie. Netflix changed entertainment. The way I see it, the most powerful lesson here is not just tactical. And it isn't actually just about AI. It's just about mindset. Too often, founders accept unverifiable problems as unchangeable. They assume the only way to know is to try. But Quo's framing suggests a more proactive approach. Always ask what part of the unverifiable can be made verifiable, or at least moved up on the verifiability spectrum, because we now live at the dawn of a new source of information, like when DNA was first discovered.

23:50This can then become a habit of thought. Don't accept we can't know. Break down the question. Use AI to model, test, and iterate, and then accept that you won't eliminate uncertainty, but you can shrink it dramatically. In time, in theory at least, this mindset can reshape how you approach your work. You develop more rigorous systems. Resources are allocated more intelligently. Failures happen faster and cheaper. Successes scale more quickly. And if you can make your customers' or clients' lives just a little more predictable, well, that is something that they will pay for. Anyway, thank you to Kuo of Alibaba.com, who I met at the recent CoCreate conference that they held in September in Vegas, for setting me down this path.

24:39This was super interesting. I've been thinking about verifiable and unverifiable ever since. And that's our episode. Now let's keep the conversation going. I write a newsletter called One Thing Better, where each week I offer one new way to be successful and satisfied and build a career or company you love. You can find it at one thing better dot email. That's a web address. Just plug it into a browser. One thing better dot email. And if you get one of those emails and reply, it goes straight to my inbox and I will get back to you. Problem Solvers comes out every Monday morning. So make sure you're subscribed so you don't miss an episode.

25:23The show is produced by Emily Holmes and Money News Network. My name is Jason Pfeiffer. See you next week.