Latest / Founder F**K ups / Building the wrong solution with Jingjing Zhong

Transcript

Building the wrong solution with Jingjing Zhong · Founder F**K ups

• Ben Bagg: time out guys. Lots of you are watching, ⁓ of you are learning from the founder fuckups that ⁓ all gone through, but please do me one little teeny favour ⁓ subscribe so we can keep growing the channel and get more great founders on. ⁓ Back the episode. Today's guest is Jing Jing, the co-founder and CEO of Superbench, an AI power platform revolutionizing how service businesses operate. Before founding Superbench, Jingjing was an investment banker and even a startup operator at Helpline where she saw firsthand how critical scheduling was for service businesses. She had a proven hypothesis, solve scheduling, increase conversion. she and her team spent four months building a powerful AI driven scheduling system, ⁓ the bottleneck they really thought they understood. ⁓ there was a fundamental flaw in their logic. ⁓ had built a brilliant solution to a problem customers weren't asking for yet. built the second step of the customer journey ⁓ the first had even been completed. ⁓ we're talking about the ⁓ hypothesis trap, ⁓ dangerous belief that a solution that worked in one context ⁓ work in another ⁓ how Jing Jing learned the hard way you cannot ⁓ a job that hasn't been sold yet. ⁓ Jing, welcome to the pod. ⁓ Jingjing: Just as you saying that found this mistake so stupid, you know? It's like, ⁓ can you schedule a job that hasn't been sold? Yeah, right? ⁓ It's common ⁓ Well, by the way, thanks for introducing me, but it's such a common sense. Why does she make that mistake? I know. Like it just like, yeah. ⁓ Ben Bagg: we're gonna dig in. We're gonna dig into why like, why did you feel that that was like such a short bet, right? What was your the kind of core assumption that brought you to hey, let's help them schedule to get conversion. Jingjing: Well, it's, ⁓ so before we decided, know, this is the problem we are trying to solve. My co-founder and I, we sat down for like three, four months to do ⁓ ⁓ the key thing is we are looking for things that we know better than other people. in terms of industry. And number two, based on our skill set, I know I'm, so I'm very passionate about sales. I can do seven sales, eight sales calls, and then still go to the gym and feel like jacked. Ben Bagg: Yeah. Jingjing: Whereas my co-founder lost building product that people love. So together, and then now there's a new technology AI, right? In a picture. Then what do we automate? What do we solve? That's number one. It's a big problem in terms of ⁓ people are impacted by this problem ⁓ it's expensive problem for business as they grow. They to, if they don't solve it, it's going to get very expensive. And it's a problem. constrained by one geographic. So a lot of countries have this similar problem. So going back, my experience running Helpling as a platform, because we don't run it as a marketplace model, it's almost like agency model. ⁓ ⁓ all the service staff, they're fully, consider as a full-time head count under Helpling's umbrella. So ⁓ are responsible fill them up, ⁓ right? ⁓ easy fill up one or two people. Ben Bagg: Okay. Jingjing: But when you try to fill up a thousand people and when you have hundreds of jobs coming in every day, that becomes a logistic nightmare. Ben Bagg: Yeah. much. Jingjing: Yes, you can only do that. I'll give you a I'm sure you can do it. ⁓ you'll figure out how to match a probably 10 minutes. ⁓ If give you the right data. Yeah, but ⁓ it very complicated when you have a lot more people. the biggest bottleneck, the biggest for Helpling was we have to match a job to a service provider. Ben Bagg: I hope so. Jingjing: fast and accurately without that a job is not sold, right? The job, like the sales process incomplete, the client can still leave you because they don't know when the job is actually confirmed. So this was the, by far the biggest driver in sales conversion. We know if you're able to match a job within five minutes, your sales conversion can improve by as much as 80%. I have proper data on this. Okay. Ben Bagg: Yeah. Hmm. Yeah. I literally, just to add, literally chose a service provider today. I reached out to three and I've chosen the first one that got back to me. I exactly what you're talking about. Jingjing: So that. not just getting back to you, the first one got back to you saying yes to your slot or tell you no by proposing a different slot. We found that it's OK to say no. I'm not available for the slot you prefer. ⁓ you have to propose new ones. ⁓ ⁓ the customers, they're OK with ⁓ new proposed slots. ⁓ that's how we can do schedule optimization. ⁓ staff, think about imagine in your calendar, right? Ben Bagg: Mmm. Yeah, like that. Jingjing: You have one air con team, full like a maximum utilization is eight jobs a day. Anytime when they travel, anytime when there's an empty gap, that's called losing money. In our mind, literally like making money, making money, losing money, losing money, making money, right? ⁓ you want to pack it up as much as possible. So all the jobs need to be very, very close to each other, better walking distance in the same condo. But it's really hard when you have so many jobs coming in at a different time, different preferred day. Ben Bagg: Yeah. Jingjing: Obviously everyone preferred to do their home maintenance on Saturday morning. ⁓ Then you have to reschedule a lot of jobs. So that was the biggest problem. we were like, Helpling has, this problem is not solved. Sales conversion does not improve. We know this. ⁓ two, Helpling had the most people sitting behind to do this job. So it's very expensive, important role to handle. ⁓ we know this after speaking to 200 companies. Ben Bagg: Yeah, yeah. Jingjing: All companies think this matching sales coordination job is the most important job that in the company without that function, the sales does not move. The field guys doesn't have job either. So like, okay, great problem to solve. It sounds very good to investors as well because we're not selling, we're not solving a sales problem. Ben Bagg: Yeah. Yeah. I mean, I'm excited. I'm excited right now. ⁓ Jingjing: because we know sales doesn't move because of this part, ⁓ ⁓ it's like, okay, it's a great problem to solve. And how can we solve it? Well, I know it. We had a team of people doing manual scheduling, regardless of how many jobs you have, they can match within five minutes. I have a system. ⁓ showed them a spreadsheet. was like, okay, step one, you check the travel distance. Step two, you better propose a slot. ⁓ right, you always propose a slot. You never say yes to a slot. Right? you try to move the client. Always try to schedule to a non-peak slot first, then saying yes to a peak slot. So we have logic behind. It's pure logic driven. ⁓ all we need is to use JNI to translate the algorithm's decision in the back end. That's it. Right? You just need to do the translation job. ⁓ no to the job, but propose one, two, three. Say yes to the job, and you're booked. Ben Bagg: Okay. So we spent four months basically you got that hypothesis from what you did at Helpling and you're like, wow, this is an incredible problem. Everyone's saying yes to your investors are like, holy shit, yeah, you're right. ⁓ is a big problem. You four months kind of ⁓ building and ⁓ just it, but ideating it. ⁓ At what did you speak to customers and ask them if they agreed? ⁓ Jingjing: We actually spoke to, so we stopped, before we start building, we talked to over 200 companies in similar industry. Yeah, and then. Ben Bagg: OK, that's one. Yeah, so everyone was kind of agreeing with it then. Jingjing: Yes. And we even built MVP with design partners. So we were building with clients since day one, but we underestimated that this is a data transfer problem as well as deeply rooted operating problem. Scheduling. Every single company has at least ⁓ ⁓ manager or sales coordinator that knows all the scheduling rules. And the every single company schedule teams in different ways is because they compensate the team differently. The salary is different the resources is also different. For example, okay, ⁓ con cleaning one, the driver is also the maintenance guy. Air con cleaning B, the driver is the driver. The maintenance guy is separate from the driver. Ben Bagg: care. Hmm. Jingjing: Okay? single company is different. Some companies they will send in two guys, some companies they send in one guy only. Some ⁓ they have vans, some companies they only have a car. Van means you can have a ⁓ ladder, then you can serve ⁓ unit. ⁓ Ben Bagg: Yeah. Yeah. Yeah. Yeah. You can go to landing properties and things like that. And so when you've, you've kind of, you've kind of walked me through what you built, right? So you, you that process on a, on the spreadsheet. You figured out how it would operate. Then you use genites make it like basically fully automated. You've got design partners. had 200 kind of customer interviews that are agreeing with you. At what point did you realize you were building the second step before the first? And what is the first? Jingjing: Yeah, the first step is so in the whole journey is qualifying the client. You scope the job. You provide a quote. You figure out where is the job to be done ⁓ time do you want a job to be done? Last step, who's going to do it? OK, qualifying quote. ⁓ is step. Then next one is scheduling is where, when and who. OK, ⁓ we're where, when and who first. Ben Bagg: Okay. Yeah. Okay. But the problem for most of your potential customers was actually that step one or more so. Jingjing: So we're selling it to businesses that sell their service to consumers. Their consumers, first question is always how much? Ben Bagg: Yeah. Yeah. Jingjing: If you cannot answer that question, don't. They would never give you their post-gond. Ben Bagg: true, ⁓ true. ⁓ ⁓ Jingjing: That's where it's not like we realized that we thought if we built a backend quicker, we can enable the people to say yes to us a lot faster. But what we miss, what we miss out is that we expect the human to start using our tool immediately as part of the conversation. So we wanted human to handle the first part and we give them the second part so they don't have to go back, go to the backend to ask so many questions to, to go through 50 people's calendar. Ben Bagg: Was this that you felt that you missed in your initial research? Was there perhaps a bit of a disconnect between kind of your understanding or maybe you were like tunnel visioned on we're scheduling, that's what we do, this is our area and you got a bit too focused? Walk me that. Jingjing: were just too excited with the problem we're trying to solve. It was in our spreadsheet, but we just kind of ignored it. Ben Bagg: Yeah. It was just totally ignored. And so what was happening when you were speaking to customers was did that correlate into like your own lack of conversion? weren't able to close your own customers or did you start to see churn because they weren't using it? What happened? Jingjing: We see clients not really using our tool, even if we built it and they can immediately ⁓ who is available at which location. ⁓ we were them to collect all the client's information. So allowing on a scope, quote, once you figure where, when. Right? You input this information, it will tell you who can go or which team can go. They're not really using it. It's also so partially it's because number one, our tool might be the UI is not very good. Number two is that they were so used to their own way of checking who is available. They don't need our tool to solve that problem. For them, it's a behavioral change. Ben Bagg: Yeah. Yeah. Jingjing: Yeah, and also the reason we chose to build a scheduling part is because if we have their schedule, we have the source of truth. And we saw that as our molt. Yeah. Ben Bagg: Thanks Okay. And I guess also from the behavioral changes, the reason why, should we say blue collar service businesses probably ⁓ least digital. We it this way. This is we've done it. We've got Karen in the office. She's been doing it for 20 years. Like leave Karen alone. You're not taking her job, ⁓ you know, of mentality. And that's you were probably feeling like you were competing with a little bit on the behavior side. I imagine I might be putting words into your mouth. Jingjing: Yeah. No, like you were right on that one. getting them, it's like suddenly telling you to go from Google Calendar to some other random calendar. That will help you to give everyone a rating and give it like this and that. It's like, you just want your good old Google Calendar. Ben Bagg: Ha ha! Why are you giving me more work? Jingjing: Yeah, exactly. And plus, guess what? You need to move all your Google Calendar to our calendar first. Then we can schedule everything in a much easier way. If you don't move it over, I will double book you. Ben Bagg: Yeah. So onboarding ended up being probably a right pain in the ass because you've got to do that transfer and it's either a super manual process for your customer or like you guys are going to pick it up somehow, right? So that must've been a challenge too. Jingjing: I'm manually key in one by one. Our CTO did the first round of transferring, and I have a spreadsheet to track all the differences. And then in the evening, after they fall asleep, that's where we do the manual adjustments. Ben Bagg: think we've all been there doing like the non-negotiable things. Not non-negotiable, non-scalable things, right? We've all heard of the story of like ⁓ in the beginning going around to people's houses and taking photos, trying like, ⁓ know, making more attractive than the shitty photos people would have had on their phones back then. ⁓ But that must have been really challenging. Obviously we four months, which ⁓ isn't the of the world, but you know, there's a cost to that, there's a morale. Jingjing: Yeah. Yeah. Ben Bagg: there's runway, there's all these things. you keep the team sort of motivated through that after basically, not quite throwing away four months of work, but kind of focusing in the wrong area? Like how did you lift get people ⁓ cracking doors again? Jingjing: It was tough because it was just three of us. Thank goodness there's three of us, not two. ⁓ Three of us came together. We had to vote. You had to vote for one direction. Either we thought, yeah, we're like. Ben Bagg: Yeah. Very democratic. Jingjing: Either you stop building, we continue this path. We continue crank it through. We have to define how long and what's the that we're expecting ⁓ ⁓ do we get from here to there? ⁓ we start building the sales side, which is a chat bot first and hopefully eventually that will link to the scheduling side. And three of us voted and we agreed that we will start building the chat bot side. So now ⁓ I started this AI sales. ⁓ We call AI salesman. ⁓ And immediately everyone wants it. It's so much easier for people to understand, you're selling an AI chatbot that helped me to sell. Yeah, I want that. When you tell people that I want to sell you an AI automation that helped me to do scheduling. Scheduling what? Scheduling me or scheduling my staff? how? Like my rules so complicated. I don't trust you can do that. If you help me to sell. Yeah, yeah. Selling is easy. I think you can help me to sell. Then that thing start to take off and we... Ben Bagg: Yeah. Jingjing: managed to like batch onboarding client cause the demand was quite high. And also that four months, was GPT three to GPT four transition. So we had to build with GPT three, we would have burned a lot more tokens and we it's going to get a lot more hallucination, ⁓ but GPT has gone better. So we kind of lucky in the way that we did not build that first. Ben Bagg: Yeah. Okay, so I ended up, I mean, a lot of startup world is a bit of luck sometimes. So you end up getting luck in terms of the burn and the hallucinations and the cost. You ended up refining from the more customers you spoke to the real problem, which then created a pull. Like people were like grabbing your hand, grabbing the hand off you to get this. Cause they were like, yeah, I get it. want to sales, I want to AI sales to close my jobs. That must have, soon as that pivot was done, you guys must have been super excited and the customers ⁓ of kept flowing in. And were able to then connect that ⁓ straight to the scheduling so that you were selling both together or did you sell them separately? ⁓ Jingjing: We had to sell them separately because we realized GenAI data structure does not match traditional engineering. So traditional, let's say if you build a CRM and calendar system, what we call schema, they have very rigid schema. They have like tables, whatnot in the backend. work with unstructured data. So people are like, ⁓ you can have structure output. No, it doesn't work that way. It hallucinates. So ⁓ it's not possible to connect them right away because we build the backend first. ⁓ platform, the scheduling platform, and we built JNI after, right? And then after that, we realized we cannot connect it. So we just finished building V2 of the platform. Now they can connect. Thank you. ⁓ Ben Bagg: Okay. Now they can connect. Well, congrats. Quick one, because I feel like I skipped a question. We were talking about how you did the vote. Was there like a specific ⁓ moment ⁓ a conversation with a customer ⁓ a conversation with an investor or a team member that triggered that vote? Was like ⁓ that moment, we've got to do something different? ⁓ Jingjing: We had two failed calendar migration. Ben Bagg: So that's the onboarding when you're migrating to someone's Canada. when you say failure, how big of a failure was it for the customer? they lose their power? Jingjing: It's just you miss, you miss one or two recurring jobs. And then the outcome of that is the cleaner would not know they have a job that day. They didn't show up and the customer call you and pissed off saying that where's my cleaner. Ben Bagg: Yeah, yeah. And then your customer calls you goes, the fuck? Where's money? Okay. ⁓ So, so two, was like two trigger points. ⁓ Shit. We to something about this. ⁓ Jingjing: Yeah, and we realized the problem is way deeper than moving them to a different calendar. It's a whole data migration problem. Their specific built to solve this problem. It's a completely different feature now. Ben Bagg: Yeah, okay. Superbench has dramatically shifted from the beginning to that sales piece. How's the... Jingjing: Well, I have to correct you. Our whole intention is to increase our client sales conversion. Right. But it's just that we wanted to solve the scheduling piece first because we think that's the key to stop the sales to complete sale. ⁓ But we move to the other side of equation. Instead of solving the scheduling first, we solve the sales part first by answering facts, collecting all the right information. Ben Bagg: show. Yeah, yeah. Yeah. Jingjing: yeah, that was the second part of the product, which is supposed to be the first. ⁓ Ben Bagg: Yeah. now it's the first step of the product, even though it's the second, it's our first step. How has that ⁓ growth for you since that pivot, since that shift? ⁓ you seen like a really big uptick? Where are you selling at? What markets are you positioning in? Jingjing: Yeah, of course. So we're selling it in Singapore and Australia mainly. And honestly, ⁓ It's just been going quite well. Also, the Singapore government has been telling everyone to use AI. So ⁓ we're so public, we're built ⁓ in public, right? I'm on Instagram and ⁓ everywhere. So whenever people need AI, they come to me, they PM me. So because of that, we're getting quite a bit of growth and we able to batch on board clients. Each batch we have seven clients. ⁓ We did three of them ⁓ until the end last year. ⁓ Ben Bagg: Hmm. Jingjing: So it was very good growth for us. And then now we're starting on boarding again to both sales and scheduling. So for clients who have been with us, we are going to really ⁓ the vision that we wanted to do. Yeah, finally, after two years, right? We're going to do this for them. ⁓ And on of that, we realized sales is not just the customer journey I share that. Ben Bagg: Yeah, finally. Jingjing: talking the sales piece and the scheduling piece, there's so much more in the middle. Like most clients, they just stop somewhere in the funnel and they just go to you. Yeah, but we can follow up, right? How do you follow up? What's the right messaging? That's also another piece and you need to get them curious, reactivate them and ⁓ eventually they will tell you the schedule. So there's so much more that we figure out along the way. So nowadays our clients, I mean, ⁓ Ben Bagg: Christ. Yeah. Bye. Jingjing: I'm going to post a new Rio I think next week. Like she literally went from 60 K to 80 K in a month. Yeah. Ben Bagg: Boom, And I like that you're building in public social selling. was literally doing a webinar about social selling yesterday. I'm not sure if you know Serena Lam. Yeah, so Serena was talking obviously she does fuzzy AI. But yeah, the social selling is really powerful. And do you find that to be a really great channel for these more blue collar service businesses then I guess because of the Instagram ⁓ than LinkedIn? Because ⁓ yeah, go on. ⁓ Jingjing: Yeah. Yeah, of course. Instagram and TikTok and YouTube. Ben Bagg: I need to get on those channels. I don't have any social media. free from the world. ⁓ Jingjing: You got a scar from somewhere. I posted about Serena like last week, because Serena, me, and Sherry, three of us, call ourselves, we're the beauty corn, beautiful unicorn. Ben Bagg: excited. I love it. I love it. I should laugh that hard. I should laugh that hard. Jingjing: yeah you see yeah like like we were eating dragon fruit in a park ⁓ then i saw serena's tongue turn purple i was like that aligns with unicorn's color i was like guys ⁓ tongue out i need to take a selfie and i took a selfie and that's the photo i posted Ben Bagg: Okay, I need to go check it out, I haven't seen this. I need to see it. Before we digress, we've got through your founder fuck up, right? We kind of built on a hypothesis that was perhaps incorrect, well, was incorrect and kind of pivoted the business. Luckily, quickly enough that we didn't kind of run out of cash and... So you're in that position. You've now found an opportunity where people are kind of like pulling themselves to that. Hey, I need this service. Hey, so you're kind of getting that mainstream is the wrong word, but you you're getting some adopt good adoption. What would you advice could you give to another founder? Because there's, would say half my audience are aspiring founders, right? And so what would advice would you give them if they think that they've got this proven hypothesis? and they're ready, raring to go, just like you were, just how passionate you were when you were describing this proven high-value system. What advice would you give them? Jingjing: you know, I saw your question. was like, retrospectively, would I take the same path? I think I probably would. Could I do it faster? Potentially, ⁓ we really need to go through two rounds of failed onboarding to really learn that this is not the right order to start. And ⁓ we got lucky, lucky that a newer model was released. Ben Bagg: Sorry. Jingjing: And so the first part, sales agent becomes a lot better and we see this proper product that we can build from there. If we started out with GPT two or three, we actually did that, which is not a product. We just saw that thing hallucinates. asks you, like ⁓ guy was like, I ⁓ to book ⁓ cleaner for my mom's ⁓ then AI literally asks ⁓ is your mom a guy or a girl or something like that? Because they didn't know ⁓ Ben Bagg: Yeah. Mm. Yeah ⁓ Jingjing: they need to know whether they have gender preference for the cleaner. stupid things like this that happened before, right? So yeah, I guess the moment you realize you're in a dead end, you need to have that courage to pivot and also bring the team along to pivot. It's good have three people on the team, not two, even Ben Bagg: Yeah. I have one quick question ⁓ alluded to this twice now. ⁓ need to have at least three people. ⁓ one of the team members vote the other way? Okay, if it was two, would have been headlocked ⁓ is you're saying. I just to check. Jing Jing, thank you so much for coming on the pod ⁓ and sharing your with us. If people want to get in touch, how? I'm gonna attack her anyway. Thank you so much