This Week in Startups - Growth Hacks for Startups from Elliot Shmukler of InstaCart, LinkedIn and now Anomalo | E1926

Starting point is 00:00:00 Explain, Elliot, why it's such an important signal for startups to find a user who is willing to tell their friend about a product or service. Jason, very simple, there are no great products in the world that don't have phenomenal word of mouth, right? Word of mouth, people telling their friends about that product is always the number one growth channel that you can have. It's the foundation for everything. This week in startups is brought to you by NetSuite. NetSuite is the number one cloud financial system, bringing accounting, financial management, inventory, and HR into one platform, giving you one source of truth. Now through April 15th, NetSuite is offering a one-of-a-kind flexible financing program. Head to netsuite.com slash twist.

Starting point is 00:00:48 Squarespace. Turn your idea into a new website. Go to Squarespace.com slash twist for a free trial. When you're ready to launch, use offer code. Twist to save 10% off your first purchase of a website or domain. And HubSpot, join thousands of companies that are growing better with HubSpot for startups. Learn more and get extra benefits for being a Twist listener now at HubSpot.com slash startups. All right, everybody, welcome back to Twist this week this week. This week in startups, we do the show. About four times a week. I've been doing it

Starting point is 00:01:23 for about 13 years. We're almost to 2,000 episodes. Why do I do the show? Because I like to learn. I love entrepreneurs. I like to invest in companies. And I'd like to help build the future. That's why this week in startups exist in the world. That's why after 12, 13 years and almost 2,000 episodes, I keep doing it. I'd love to learn new strategies, techniques, and just to meet the people who are building the future. Today will be no different. Two of the great companies, two of the great products I use and love, Instacart and Wealthfront, they grew like weeds. You've heard of these companies, right? Wellfront had amazing tactics on how to grow. Incredible strategies.

Starting point is 00:02:00 And Instacart, just IPOed last year. Stock keeps growing. People love that product. I use that product every week. I use Wealthfront. But who is the growth officer? Who did growth at those companies? Today's guest was the chief growth officer at Instacart and the VP of Product and Growth at Wealthfront.

Starting point is 00:02:15 His name is Elliot Schmookler. I got it correct, right? Elliot, Elliot Schmooker. You got it. Exactly. Schmooke. And his company is anomalo. A.

Starting point is 00:02:28 We're going to talk about Anomelo in the second half of the show. For the first half, I want to do a primer on growth, modern day growth at startups for all the founders who are listening. So let's start with how does one, at a early stage startup, how should they think about growing their company? When should they start having a growth team? Or should everybody in the company have a growth mindset? Should it be a dedicated group?

Starting point is 00:02:56 How should this work? And welcome to the program. Yeah. Thanks for having me, Jason. It's great to be here. Let's talk about growth. Right. So how do you begin?

Starting point is 00:03:05 When should you start? Should you have a team? You know, one of the major things I want to make sure all your listeners understand is what I've traditionally done with growth is something you do after product market fit. Okay. So before product market fit, you should be doing whatever you can to get those initial users, right? There's no playbook, no techniques. Go out on the street and find people.

Starting point is 00:03:26 go to your network, right? Do whatever you can to get those initial users up because nothing else will really work. Paul Graham, I think has a great blog post about this. Do things that don't scale, which is kind of the opposite of the advice you get in later stages in your startup, right? You want to start scaling stuff? Well, before product market fit, just do things that don't scale. Got it.

Starting point is 00:03:48 So product market fit, it must occur before you start applying growth. why is this so important for founders to understand what happens if you add growth before product market fit? It just won't work, right? Or you're going to be wasting your time and your money. I'll give you some very simple examples. Let's say you're going to grow through paid marketing, right, or paid advertising. Well, before you've achieved product market fit, you don't know who to advertise to, right? You don't know who your users are or what messages they resonate with. There's really no foundation for you to build stuff. Or a lot of the things I did at, at Wellfront and Instacart, we're about taking an existing stream of users that are interested in

Starting point is 00:04:29 using these products and just magnifying and amplifying, right? Making sure more of those users actually use those products more heavily, right? Well, if you have no existing stream of users, there's nothing to magnify and amplify. So none of those strategies actually end up being working. So, you know, the first few years of a startup and I went through this at Anomelo as well are kind of dark times, right, where you're looking for product market fit. You're trying everything. And once something is working, once you have a set of users that are passionate about your product,

Starting point is 00:05:01 now kind of thinking systematically about growth can begin. So you get a small number of users to try a product. You can do that by emailing them. You can meet them at a grocery store. If you were doing Instacart, literally just talking to somebody as they're in the parking lot and saying, hey, I'm Jason, I'm the founder of Instacart. we want to let people order their groceries on their phone. Is that something you might be interested in?

Starting point is 00:05:28 Because I'm looking for 10 beta users. Like, it sounds silly, but that is actually the best way to get the first 10 users for Instacart. Is it not? That's exactly right. And my favorite is actually not from Instacard, but it's what Uber did in the early days. They would send people to sporting events, right? These huge events, right, thousands of people leaving the stadium that all need a ride. And they'd be like, do you need a ride?

Starting point is 00:05:52 right, here's an app you can try to get you a ride home. Wouldn't that be great? Yes, press a button, get a ride. Right. Things that don't have to pick up the phone. It does not scale. Completely manual, but hugely powerful. These were folks with exactly the need that Uber, you know, was trying to solve.

Starting point is 00:06:10 And you just needed to expose enough of them where you now had a core passionate group of users that would ultimately tell their friends and, you know, your growth engine could start. Now, you said something very specific. in there. They could tell their friends. People should at this point, pause, look up net promoter score and understand what net promoter score is and about virality and people telling their friends and why that is such an important signal. Explain, Elliot, why it's such an important signal for startups to find a user who is willing to tell their friend about a product or service. Jason, very simple. There are no great products in the world that don't have phenomenal word of mouth.

Starting point is 00:06:52 right word of mouth people telling their friends about that product is always the number one growth channel that you can have it's the foundation for everything you know when i was at lincoln right people kind of know lincoln about yeah through their it's viral growth right those invitations that you get that you accept and maybe you sign up for lincoln if you're not a user yet that was all important but guess what no matter how much we optimize those things about 20% of our growth was still word of mouth, right? You bump into someone at the office and you're like, oh, you should try this thing, this LinkedIn thing. It will help you. And they go and they sign up. So you must have word of mouth. And once you have word of mouth, you can kind of amplify that

Starting point is 00:07:37 word of mouth through other things. Right. You can have invitations blows that amplify. You can have other ways of capitalizing on it. A referral program would be the obvious low-hanging fruit, Instacart, Wealthfront, and Uber all had that. Let's pause for a second. you have users who are willing to tell their friends about the product, it means that they love the product and they don't want to sucker their friends into a bad product. So no amount of incentive is going to take a user and have them abuse their friends with a crummy product. If you don't like a restaurant and they say, we'll give you two for one entres, three for one entrees, 10 for one entrees, no reasonable person is going to send somebody they love or a friend

Starting point is 00:08:20 or even an acquaintance for a five-for-one entree if the food is terrible. So you must first have a meal that people love. So let's assume you have the hamburger that everybody's crazy about or the app, Wellfront, Instacart, Uber. How do referral programs work in today's environment? Why are they so powerful? And I'm sure you've worked on them because I know Wellfront, Instacart. They really did have really great referral programs, correct?

Starting point is 00:08:50 Yeah, absolutely. And very powerful growth drivers for both companies. So the absolute best thing is to build a product where referring people to the product is just a part of you using the product. If you think about LinkedIn, you know, LinkedIn is a professional networking product. Part of the product is connecting with people in your professional network, people that you meet, people that you see at a conference, what have you, right? So kind of referring people, inviting people to connect is a core part. of using your product. Slack is similar, right? Slack had amazing growth for many years because what you want to do in Slack is once you get on Slack, well, you want your team to be in Slack. So you invite them, right? Dropbox and Notion are very similar products in that sense as well. You have a document on Dropbox. You want to share it. You have a document and notion. You want to collaborate with other people on it. So there's more of an invitation mechanic that's just built into the product. That's the absolute best you can do. If you can choose a product to work on that has sort of referrals built into your day-to-day use,

Starting point is 00:09:57 that's going to be really amazing growth for you. All right, let's do some quick math. The less your business spends on operations on multiple different systems and on delivering your product or service, the more margin you have. And margin is everything today. We are past the peak, ZERP. We are now in brass attacks. And the more margin you have, the more money you keep, the more money you can invest.

Starting point is 00:10:18 And the more people will be interested, paradoxically, in investing in your company. But with higher expenses on materials, employees, distribution, and borrowing, everything is costing more. So to reduce costs and headaches, smart businesses are graduating to NetSuite by Oracle. NetSuite is the number one cloud financial system, bringing accounting, financial management, inventory and HR into one platform, giving you one source of truth. With NetSuite, you reduce your IT costs because NetSuite lives in the cloud. There's no hardware required, and you can access it from anywhere. That means you cut the cost of maintaining multiple systems and the task switching between them and the education and the updates.

Starting point is 00:10:53 All that nonsense is gone. And you improve your efficiency by bringing all your major business processes into one platform slashing manual tasks. And all of those time consuming and costly errors, over 37,000 companies have already made the move. So do the math to see how you'll profit with NetSuite. Now, through April 15th, NetSuite is offering a one-of-a-kind flexible financing program. Head to NetSuite.com slash twist. That's net suite.com slash twist. Now, that is particular to social networks.

Starting point is 00:11:23 Yes. But Instacart and Wellfront are not social networks. Yes, yes. Then the second best thing you can do is make the product something that people want to brag about. Oh, okay. This is not something I thought about. A product, people are willing to brag about. A product that confers some sort of status.

Starting point is 00:11:45 Got it. How does Instacart do that? And how did you do that at Instagram? So Instacard is an interesting one because, you know, the heavy users of Instacart are usually kind of families. They're heavily pressed for time. And so one of the things that confer status on them is being able to do more things, right, without, you know, in the same amount of time. Got it. Right.

Starting point is 00:12:09 And so that was what folks would talk about, right? When folks would discover. Yes. they would say, oh, it was no problem for me to put that party together. I just ordered all the things via Instacart. You know, look how amazing I am in discovering this new tool. Yes. With Wellfront, we had that with our sophisticated investment strategies, right?

Starting point is 00:12:33 Well, front offered things like tax loss harvesting that, you know, until Wellfront launched that you couldn't get it. You couldn't, you know, maybe you could get it at like Goldman Sachs with $10 million in your account. But, you know, the average investor couldn't sign up for that strategy. And so if you were with your friends and you're talking about investments, you can be like, well, are you guys doing taxless harvesting? That's a great flex. Yes.

Starting point is 00:12:58 And for people who don't know what tax loss harvesting is, it's a pretty sophisticated technique. You're looking at your portfolio. You have some gains over here. You have some losses here. So if you were looking at jatrading.com where I did the public trading. I'm going to start that up again this year. Let's start trading some more. I have like three of my 15 companies, huge losses, Warner Brothers, Stitch Fix.

Starting point is 00:13:18 I think Disney, maybe I'm even in Twilio. I'm down a little bit. So I had like maybe three or four names where I'm down. I could sell some Amazon, Facebook, Uber, where I had massive profits. And instead of paying taxes on those profits, I could then sell the losers take the loss and then those gains would be tax free. This is a very sophisticated strategy. The way I've explained is very simple, but across an entire portfolio.

Starting point is 00:13:41 you know, this is something very efficient that, you know, if you have like, let's say, income producing stocks, like dividend paying stocks that Andy put together at wealth front. And so that's a, that's a super flex for sure. Yeah. And, you know, Uber had this too. By the way, you know, I remember in the early days folks would arrive at the restaurant and in their Uber black car and that would be a kind of flex. There I'm like, how did you get here? They're like, oh, I just used Uber. It's super cool, right? Yeah. Or I have my Uber. waiting for me, would you like me to drop you off? Exactly.

Starting point is 00:14:14 When you think about it, integrating into the product, adding a stop, and now you can add up to five stops, that is a viral feature. The fact that I can have you come in and I could have drop you off after or before, and I can do it during the ride, which is kind of sophisticated when you think about it. I'm in the ride. I'm heading to my house. And I'm like, Elliot, where do you live? Oh, yeah, that's after me.

Starting point is 00:14:34 I'll just add it after me. You can use my driver. Pretty viral. Very viral. And Uber had this built in because one of the top use cases was going out to social things and coming back from social things so that you were guaranteed to run into people, right, that were arriving at that place at the same time or we're leaving that place at the same time. And so that was an optimal way for you to kind of spread the idea that you were using this new thing. Also group, you also have group ordering on Uber Eats, DoorDash, and Instacart now. So when my wife is ordering food, she will share it with me, hey, I have the basket, put in your email address and then a person, we've all experienced this now on Uber where you just, it made it so fluid.

Starting point is 00:15:17 You just send a URL to a group chat. Everybody put your Starbucks order in. Boom, you have. And you can even put a time. Like, I'm ordering in 15 minutes, get your order in or not. Yeah, exactly right. A huge, you know, I remember that feature from Instacart, huge for offices that were ordering snacks, right? You could send it around and be like, which snacks do you want?

Starting point is 00:15:35 just add them to the card, right? And suddenly you expose the entire office to the idea of Instacart, right? And the idea that they can get stuff delivered. What about Give to Get? Hey, get a free ride, give a free ride, get $5, give $5. Are people burnt out on that? Does it work? Is it like negative ROI for companies?

Starting point is 00:15:56 How do you think about that? Because Wealthfront had a feature that was a gift to get. Maybe you could explain and do those work today. It definitely still works. And it's a feature you must have. I launched it at both Wellfront and Instacart, right? So, well, from the future, you know,

Starting point is 00:16:11 the more people you invited to the platform, the lower your fee was for the Wellfront service. So you kind of got a direct discount for helping Wellfront grow. And they've actually enhanced it quite a bit since I left. Now you get a higher interest rate on your cash deposits at Wellfram by inviting folks. You get an extra half percentage point of interest, which is pretty significant for inviting folks to the platform. And then we had the same thing on Instacart, you know, the more people you referred to Instacart

Starting point is 00:16:40 that went on to place orders, you know, you got a coupon essentially. You got a discount off your next order or your series of orders. These things definitely work. They don't work as well as the first two, right? As kind of, you know, build in something viral into your product to begin with or to really create situations, create dynamics where I could brag. about this thing. Got it. So build virality in. Hey, we got a group shopping order. Hey, you're adding people to your network. You're adding people to your slack. That's the gold standard. That's number one.

Starting point is 00:17:15 Number two, hey, I can brag. I can flex. Hey, yeah, no, that's my Uber waiting for me outside. Do you need a ride? I'll jump on in. Number three, maybe referral programs give to get. Hey, I give you a ride. I get a free ride. Tell me about the abuse in that third item. Because it does seem people can abuse it. What was like the worst abuse you? you've ever seen? Do they have to have caps on it? How do you deal with abuse? And I have some stories of it, but I want to hear from your side about abuse. Yeah, yeah. Some of it is self-limiting, right? So if you set up your incentives correctly, some of the abuse can be self-limit. So, for example, at Wellfront, we would never give you the reward unless not only did your friends sign up,

Starting point is 00:17:57 but they deposited money into Well-front, right? So they took a pretty heavy weight action. And so it was hard for you to really abuse it in that sense. But yeah, there's some basic cases like you can, you know, have multiple accounts, right? And kind of psycho rewards through all of your accounts, hard to do in Wellfront where everyone needed a social security number and all those things. But for Instacart, we definitely saw those cases, right? And we had to put in caps.

Starting point is 00:18:25 We had to use, you know, machine learning signals to spot those patterns and terminate them. I had this happen to me at Robin Hood. I was one of the angel investors in Robin Hood. I have a lot of followers. So I tweeted, hey, join Robin Hood, use my affiliate link. And I did it also for Tesla. With Tesla, I wound up getting like, if you ever see me tweet my bulldogs in like Tesla cars, that's because I did so many referrals that I got two or three of those little mini cars,

Starting point is 00:18:53 the battery power cars for kids. Those things cost like six or seven hundred bucks. Yeah. So I probably got like $2,000 of dollars worth of those for free. But I had also gotten 100 people or 50 people to buy. model threes. And then with Robin Hood, at a certain point, I think I had 250 people had signed up for Robinhood because of me and they give you a free share every time that it said no more free shares for you, Jacob. And I said in Devladen, I was like, I don't care about the free shares. It's like

Starting point is 00:19:18 just a $3 share. I just sell it immediately. And you like because I don't want the clutter in my account. But, you know, like, you can leave my referral thing on and not pay me. I don't care. And I think they just actually reversed it. And then for Uber. I remember this with the free ride to get one. Somebody took their Uber link and they bought Google ads. Yeah. And it was a friend of mine. And they bought Google ads in the early days, you know, Uber in L.A.

Starting point is 00:19:43 get a ride. Uber in Santa Monica get a ride. They just picked like, you know, certain like long tail keywords. And they had, I think, like 300 free rides. And then Uber found it out because they tracked back the URLs and they just turned them all off. Yeah. And my friend was pissed and I'm like, it's against the terms of service to do that. But in a way, it doesn't really make.

Starting point is 00:20:02 matter that arbitrage because, like, does it matter? Like, I mean, it's annoying, I guess, for the growth person because it perverts the statistics, but it doesn't matter because the person did take a ride. Yeah, it doesn't really matter, right? If you structure your program correctly, it's their own. And we had the folks doing, you know, Google ads for Wellfront, right? Because there was a monetary incentive. And we actually figured out that most of the lost money on the Google ads, because it's actually hard to be ROI positive when you don't control anything in terms of where the person lands and, and how they're going to sign up.

Starting point is 00:20:34 And so, ultimately, it doesn't matter. One thing that drives me crazy is when I want to buy something on a website, but there's too much friction when I go to checkout, right? I want to make the impulse purchase. I'll be totally honest. I see something I like. I got these new reader glasses. You know what?

Starting point is 00:20:48 It was so easy because there was no friction. That's what I want. I want to be able to have a smooth and easy payment experience, but it's going to be safe and secure, right? I don't want people hacking my credit card. And I like to have many different ways to pay. I want to use my digital wallet, obviously. And then, you know, other people might be,

Starting point is 00:21:02 and the buy now pay later period of their life. No judgment's there. If you want to be on a payment plan, that might be a good idea for you. If you're a merchant, you got to be able to offer alternative payment methods, APMs. This is critical. And you know who's the best of this? That's right. Our friends at Squarespace, Payments makes it easy for customers to make purchases, which equates to more revenue for your business. Super easy to set up, and it's going to keep your revenue growth going up and to the right. So check out Squarespace.com slash twist for a free trial. Squarespace.com slash TWIST. You get a free trial when you're you're ready to launch, you just go to Squarespace.com slash Twist, and you get 10% off your first

Starting point is 00:21:36 website or domain purchase. Squarespace has the best design. They constantly release great features like Squarespace payments. They study their customers, you and I, and they ask you, hey, how can we make this better? And when they make it better, they give it to you and it's for the same price. That's what I love about Spurspace. Every day I wake up with new features in that same affordable price. Squarespace.com slash twist for a free trial. Okay, so we have three so far. Build virality into the product. Got it. Number two. the flex, be proud, tell people, hey, I'm saving time, I'm saving money, I've got a cool car, I got my own driver, I can drop you off. Number three, give to get programs. What else is in your

Starting point is 00:22:14 playbook for growth? What other items are there? Yeah. So that's kind of the critical viral stuff, right? Then once you have some virality going, one of the critical growth things you should be doing is, you know, what the technical term is conversion rate optimization. Okay. Any sequence of steps that a user has to go through to take an action that you care about. It could be signing up, could be deposit of money. In Wellfren's case,

Starting point is 00:22:39 could be placing a grocery order on Instacart, ordering a ride from any kind of sequence of steps that they go through. You should be collecting all the data you can about how folks are flowing through that sequence of steps and generating tons of ideas in terms of optimizing that process.

Starting point is 00:22:59 Okay. But the idea that the idea that the, simpler you can make that sequence of steps, the fewer steps, the easier the steps, right? The more, whatever the incentive is, is reiterated on those steps, the more people will actually complete that process. Because that's what really hurts a lot of growth efforts. You know, you might have a great viral hook that gets 100 people to come to LinkedIn because they were invited. But maybe only 10 of them sign up. So kind of 90, you know, are gone. They were there, But it was just too hard or not compelling enough for that way to go through the process.

Starting point is 00:23:36 So the optimization of funnels is critically important. Walk us through what, who does that? Who is the optimizer? Because you have designers, U.S. designers, they like to make things pretty in some cases. They want to win design awards. You get developers, they want to push code. They want to use the latest tools. Okay.

Starting point is 00:23:57 And they have like a product manager. they are kind of like a maestro, some CEO is yelling at them. You've got a CEO who's probably got their important ideas. And then you have growth people. Where does this land in an organization? Let's assume we've got a company with 10,000 users and 20 employees and $3 million in funding. Now, we're starting to build a team. How do you think the team should be composed?

Starting point is 00:24:24 Because I know I talk to startups, hey, who's responsible for the funnel? And they're like, yeah, I guess, well, the developer built it and the CEO said what they wanted. And there's a UX designer. They did some of it. There's a product manager. This is a growth team. Like, that's five different people. A lot of cooks in the kitchen.

Starting point is 00:24:42 What's Elliot's version of optimizing this kitchen? Absolutely. So typically the closest role, the closest functional role to doing the right things is the product manager. Think about product managers. Ultimately, they're, you know, the good ones, of course, right, are outcome oriented. They want to have some sort of goal. They want to sign up as many users as possible. They want to get as many pictures shared as possible.

Starting point is 00:25:03 They want to have as many rides, whatever. They're typically very outcome-oriented. And they understand the worlds of design and they understand the worlds of engineering. And often they can get deep into the data, right? Or work effectively with data analysts, data scientists to get deep into data to combine through those three perspectives, design, engineering, data. But with this sort of outcome orientation, what can, can we do now, whether in engineering or in design or using data to achieve our goal? You know,

Starting point is 00:25:35 they don't care typically if they're going to win awards for the UI design. They care more about that UI design being effective in achieving that goal. What's the most effective sign up today? I run into all kinds of signups. I notice on my phone, there's a lot of apps that are just like, give me your phone number, here's your secret code. They don't even ask you to put a password in. They don't ask you to name an account. It's just like, you're phone number's enough. That's our single source truth. Then I have other things I log into and they're like, hey, Google, hey, use the Apple sign up. What sign up is the most effective at getting people through the process quicker? And then how do you deal with this like multi-signup madness that's occurring

Starting point is 00:26:14 right now? Because I will sign up for something using Apple. But then I try to sign up my phone number. It's like, oh, do you want to merge these accounts? This just seems to be one of the massive friction points. If you were advising a startup, what would you tell them to do? Yeah, yeah. So it tends to be very product-specific. At Wellfront, we actually found that if we made the account creation easier, folks wouldn't come back as much after they created their account than if we made it just a little bit harder. So we worked on making it just hard enough where after creating the account, folks felt like they had, you know, invested in the game.

Starting point is 00:26:54 Skin in the game. Okay, so explain to me what? additional information you asked for. Well, everyone's very heavy, right? You know, if you're going to open a funder account,

Starting point is 00:27:04 you got to have your address, your real name, your social security number. Okay, yeah. It's very, very heavy. But we found,

Starting point is 00:27:11 even in the early steps, if we didn't have you create a password, right? If we used one of these, you know, you're going to put your phone number and we'll send you a code, folks just wouldn't come back enough.

Starting point is 00:27:22 Wow. Even thinking of the password and the, the big growth- Intuitive. Yes, but again, it's a different kind of product, right? In a different kind of app, I could believe that using the code or how it is the best thing to do. When ultimately, I want folks to give us their social security number and put money in, you know, a few steps later, making it seem more solid and more secure and more trustworthy by asking for that password was actually important. You know, it's super interesting. I force people when they sign up for my event. events, like in-person events, liquidity, which we're doing in June, like VCs and GPs.

Starting point is 00:28:02 I forced them to have a, they apply for a ticket with their credit card, and then we will either accept them or not, but they have to have a LinkedIn. And one out of 50 people, one out of 100 people is, I don't have LinkedIn. I don't use LinkedIn. And we're like, okay, well, you need to have a LinkedIn to come to this event. So create a LinkedIn professional account. We want to make sure we know who you are or send us some information, you know, email us from your corporate account kind of thing. And what we found was this was like the ultimate way to weed out people we didn't want to have in events. And we do this also for the All In Summit, which is a $7,500 ticket. But having the apply process, looking at it, we didn't want the events to be infected with

Starting point is 00:28:41 too many service providers because, you know, you have like 50 real estate brokers show up for a hundred person or 200 person event. Now, everybody's trying to sell you a home or finding out how much your home's worth. And it just ruins the event. No, no, no, no shade to a real estate. I'm sure there's some great ones out there haven't met them yet. But, you know, the people who tried to circumvent it were service providers who were trying to hack the system. And so a little bit of friction really good. You know who else did friction was Thumbtack. When you went to Thumbtack, we were the first investors in Thumbtack, Jonathan's company and Marco. And in Thumbtack, when you said, I want to paint this room here, he would say, okay, how many rooms are you painting?

Starting point is 00:29:20 He said, fine. So, okay, are there any cabinets? Say, yeah, it's cabinets. Do you want to paint cabinets, yes. Okay, are you painting the inside or the outside of the house? Are you providing the paint or are we providing the paint? Okay, is there molding? Okay. And by asking these questions, do you want to strip the wood? And you could say no, or I don't know. But by creating a longer dossier, the longer you filled out, the more vested you were, then they had more information to give to the service provider, which increased their chances of wanting the customer, and then there was just more invested in it, which is exactly what happens when you go to buy a used car. they keep you on the lot a long time.

Starting point is 00:29:56 So when you get to two hours on the lot, leaving means you lost the two hours. Nobody wants to lose the two hours. But you can use this, Elliot, to your advantage. If you say, you know what, I've been here two hours, I haven't gotten this done. I can buy a Tesla and I could have been done right now. My time is valuable to me.

Starting point is 00:30:12 I'm leaving. You have my phone number. If you want to sell me this car, this is my price, but I got to go right now. I've got a meeting and you just walk off the lot. They will chase you off the lot. No, no, no, no, no, no. You say, sorry, I can't do it. You have my phone number. I can come back tomorrow for 15 minutes.

Starting point is 00:30:27 You might get the price you want, but that's why they keep you on the lot. Yeah. Hey, everybody, I am obsessed with AI right now. You know that every Tuesday we do a bunch of AI demos and we're trying to figure out on this very program, how do we take all this AI potential and make it a reality? So when we're running our startups, our startups are more efficient, and we get more done with less, and we delight our customers. I just read a report by our friends at HubSpot, and they're talking about all the different ways to use AI to improve specifically sales marketing and customer support. Now, if you're running an organization, you know that you live and die by yourselves, and that marketing drives yourselves, and that customer support is how you keep your

Starting point is 00:31:04 customers. These are three of the most important pillars in your startup. And the team at HubSpot did something really interesting. They surveyed a thousand early stage founders. They asked them, how are you using AI to do sales, to do marketing, to do customer support? First, using AI to segment your customer list. What a great idea. Second, personalizing everything with AI from your website content to SMS to email. Third, how do you encourage your team to get creative with AI? And fourth, how do you balance between investing in AI trained experts, right, and developing your current team? Now, you're going to read this report and you're going to get all of this information, these weapons for massive growth. WMGs are just waiting for you. If you're in your

Starting point is 00:31:42 podcast player, check out the link in the episode description. And if you want to save 70% on HubSpot for startups program, head over to HubSpot.com slash startups. That's right. HubSpot love startups. They were a startup themselves. Visit HubSpot.com slash startups to see what discounts you qualify for. And start using their powerful solutions and pricing that will break the bank, HubSpot.com slash startups. The key is you have to find the right level of friction for your individual product, right, that makes you successful. Usually lowering friction, making stuff easier and easier is the right thing to do.

Starting point is 00:32:12 But there is a kind of balance that you need to strike. Your other question, Jason, was about merging accounts and merging accounts. multiple signups. And this is a huge deal. It's really only a huge deal when you get to scale, right? I think most startups are not going to have a huge problem with duplicate accounts, right, or having to merge users. So I wouldn't worry about those folks out of the gate.

Starting point is 00:32:37 But I can tell you, LinkedIn was creating, you know, a million duplicate accounts a week. Oh, was there? Yeah. And so we actually spent quite a bit of time. I had an entire team in my organization that figured out all. the possible ways that you could have a duplicate account on how we could detect it and would merge them in the right way so that we stopped creating duplicates. And so at scale, you definitely have to worry about that kind of stuff.

Starting point is 00:33:05 Tell me the most brilliant for each of the companies you worked at. Tell me the most brilliant little growth hack and insight you had at each one of them. So take a minute. And then tell us the most frustrating moment at each of those. and what, you know, like you couldn't solve or you had a hard time solving, and whether you resolved it or not, but that was just particularly challenging so people can get a sort of overview of that and then we'll get into your new startup. Yeah, absolutely.

Starting point is 00:33:29 So I'll do them with kind of reverse chronological order. Let's start with Instacart. So the most unexpected thing about growing Instacart is kind of how word of mouth worked at Instacart and how geographic it was. You know, when we launched, you know, say we launched in Dallas. right? And Instacart is very similar to Uber, right? You need folks on the ground, right? You need some delivery drivers and those kinds of things available. So typically launching a new geography is quite a bit of work. So when we launched in Dallas, typically we'd launch downtown Dallas, right, the city center. What was an amazing insight is that as soon as we launched the city center, all the zip codes that surrounded the city center were suddenly a lot more interested in getting Instacart. Fascinating. Because there was a kind of geographic word. trip out. And so eventually, you know, part of my work there and, you know, one of the most rewarding things we did was was changing our geographic expansion strategy. We used to kind of watch Dallas and be like,

Starting point is 00:34:30 all right, Dallas is working. Let's go to the next city. Go to Houston. Yeah, exactly. No, you want to go to the ring outside of Dallas. You want to cover everything around Dallas. So smart. far as the eye can see, right? As far as, you know, you can find grocery stores, right? You want to cover it completely. And so we changed our strategy and we, you know, launching the city center and then just rapidly expand in the circle. So smart.

Starting point is 00:34:59 That's so smart. Huge. Because it's easier. The person's already been primed. Their friend used it. So when you look at marketing, there's like some marketing funnel that sales people use, marketers use about like they have to be aware of what you do, then they have to understand what you do, then they have to try what you do, and then they have to like become like regular

Starting point is 00:35:20 customers. It's like people have different words for that. I don't know what the name of that funnel is. But in this case, the first step or the second step are kind of taking care of. They've heard the name, Instacart, and they know somebody who's used it, right? So it's like, yeah, that conversion funnel is so primed. That's right. And before we were breaking the word of mouth, right? You could have folks that were just outside the coverage area that are hearing from all of their neighbors and their friends about this Instagram, but they can't sign up because we're not theirs. We were actually stopping the word of mouth and its tracks, right? And it was great. At Wellfront, one of the surprising things was account minimums. Maybe this is obvious to folks,

Starting point is 00:36:06 but it was kind of very surprising to me, but that was one of our biggest growth drivers, right? We'd sign up these folks and we'd know they have, you know, a million dollars to invest. But they put in 10,000 into welfare. And we're like, well, you know, if you like it enough to put in 10,000, why aren't you not you putting in? Yeah. Why aren't you putting in more of your assets? You know, what is stopping it?

Starting point is 00:36:31 And just folks didn't think that way. They were trying this thing out, you know, they weren't fully sure. They didn't know if all of their assets should be. in there or some of their assets. You know, there's difficulty in moving money around, right? And so just us offering products would hire minimums was a huge growth driver. We can say, oh, it's great that you put in $10,000. But if you put in $100,000, you get this extra level of taxless harvest.

Starting point is 00:37:01 Yes. And you put it. Largent loans, whatever. Sure, whatever it is. If you put it $500,000, you get even more. concierge time, whatever. Yeah. And so those kinds of guidelines,

Starting point is 00:37:15 especially in a product where, you know, we weren't getting on the phone with these folks, right? We weren't talking to them or telling them what to do is all online. You know, providing these kind of guidelines, here's how much you should have in Wellfront if you can afford it. Right. It was a very, very powerful way. Interesting.

Starting point is 00:37:31 You know, at LinkedIn, lots of crazy stuff, right? But, you know, one of the most interesting insights I'll tell you, and we used it a number of different ways was, how often LinkedIn was about kind of power relationships, right? How often, well, yeah. So the number one reason that folks would accept an invite to LinkedIn is because their boss invited them or someone who's like who was at a different, you know, higher social level. Well, that's, you know, self-perceived, right? So famous investor, right?

Starting point is 00:38:06 I invited them to LinkedIn. And so that was an amazing insight. It's not obvious at all that that would be the dynamic, but it makes sense in terms of human nature. So then you have to come up with a reason of why the boss wants you to be on there. And I just did this. I, when we had a job listing, in the job listing, you can add people to have the hiring ring on LinkedIn.

Starting point is 00:38:30 So I took like four or five people from our investment company. I said, hey, here's the job posting for this new researcher. please get on that researcher and put the hiring ring on and post it to your feed. So the act of doing that, if those people weren't, didn't have LinkedIn pages, I'd be inviting them. But I am just getting them to engage because I want to activate their networks to promote this job. So that is a great insight. Tell me about your new company and what you're working on now.

Starting point is 00:38:59 Yeah. So the company is an envelope and it was driven out of struggles that my co-founder, Jeremy and I have had many companies, right, I had these struggles back when I was at LinkedIn, which is, you know, every company today and certainly LinkedIn well for on Instagram are huge users of data for everything, right? You're tracking how your users are using your product. You're running experiments and deciding which things you want to ship or not. You're kind of analyzing patterns of usage. You're analyzing your marketing effectiveness. You're analyzing all of these things using that you're making recommendations using machine learning models using data. And so what happens

Starting point is 00:39:37 in a lot of these organizations is, well, sometimes the data is wrong or it doesn't arrive or big chunks of it are missing, right? Or, you know, a column of data that you expected to be in US dollars is now in foreign currency. Right. So all of your dashboards and all of your models that are expecting US dollars are now misbehaving, showing you the wrong numbers. And so these become huge fires in these organizations. And now are becoming fires in every company because everyone is trying to use data to do stuff. And so Anomelo is kind of the tool we wish we had back in the Instacar, Wellfron, and LinkedIn days to catch these data issues.

Starting point is 00:40:18 Right. So our customers point Anomelo at their data warehouse where all of their data lives. And we automatically use AI and machine learning technologies to spot issues in their data and alert them about these issues, you know, before that causes things to break, before that causes the dashboards to be wrong, then bad numbers to be reported, bad decisions to be made, or, you know, machine learning models that are using that data to kind of go off the rail. Got it.

Starting point is 00:40:47 And so who uses the product and maybe some examples of them having success? Absolutely. So our end users tend to be kind of data scientists, data analysts, folks that are deep in the data inside these companies, and many, many examples, right? But usually they'll spot, for example, oftentimes companies are pulling in third-party data, right? Instacar used to pull in grocery feeds from, you know,

Starting point is 00:41:10 every inventory feeds from every grocery store that we worked with. Sometimes those feeds would be wrong, right? And Anomelo can spot those issues immediately. Or they're pulling in reports from, you know, marketing tools, right? Because they're running huge Google campaigns or huge Facebook campaigns, right? sometimes that data will be wrong or data will be missing and they'll spot those things instantly. Sometimes their product is broken, right? And so the data is going to indicate that there's some breakage or some unusual behavior.

Starting point is 00:41:40 That's happening. And again, anomaly will spot those deviations pretty quickly. Tell me a little bit about, you know, to the extent you can, I see you've got like some pretty serious customers here, pretty serious investors like Snowflake invested, data bricks invested, or their partners, why are those people getting involved in your company? And aren't they competitors to a certain extent? Yeah, so Databricks invested, Snowflick is a partner. They're getting involved because, you know, companies like Snowflick and Databricks, right,

Starting point is 00:42:10 which are operating these huge data warehouses, right, that were enterprise or storing their data. They hear these issues from customers all the time. You know, every time they talk to their customers and they're talking about, you know, using data, bringing in more data, overaging data. one of the top issues that customers talk about is data quality. They're saying, well, I want to use this data, but it has all these issues and how can I be certain that it won't have this issue in the future? And that's where Enomelo comes in, right? Anomelow is kind of your safety blanket, your insurance, right, that sits on top of your data and we will let you know if there's an issue.

Starting point is 00:42:49 Now, you had a hard time maybe explaining the value of this product to customers. and maybe engagement was low at the start like any other startup. So how do you get people to try a product like this where it's kind of like maybe they don't even know they need it? How do you have you gotten people to engage? And obviously, this is a couple of years ago. I know you got some really big clients using it, but you did struggle to get that product market fit like all of us do.

Starting point is 00:43:16 So maybe explain the journey of product market fit here. Because this is a vitamin or a pain killer, you know? Yeah, it's hard to say, right? It's hard to say. The struggle we had was actually not folks identifying that they had this problem, right? Every time you talk to data people and you ask them, do you have issues with data quality? They'll say yes, right? It was very rare, even in the early days, that we would talk to folks and we said, like,

Starting point is 00:43:44 do you have any data quality issues? They'd be like, no, our data is perfect, right? You know, and some folks did say that. They do exist, but it was pretty rare. Most people say, oh, no, we have all these issues and, you know, we reported the wrong thing to the board the other day and we had this model go off the rails. So they knew that they had the problem. The difficulty we had with product market fit is convincing that them that our solution was what they needed, right? Particularly because our solution is heavily automated.

Starting point is 00:44:13 It's all machine learning, AI-based, right? You point us at your data and you kind of have to trust us to find the issues, right? Whereas folks were used to using more manual approaches, we're used to kind of establishing rules for their data and kind of being very confident. Well, I put this rule in, so I'm very confident that my data is going to fit this rule, right? How can I trust that anomaly is going to find the same thing?

Starting point is 00:44:38 So that was our difficulty. And the great insight we had was we just built both, right? So we now let you put rules in, right? Because folks are comfortable with that. That's like a starting point. that they're very comfortable in. But then over time, they see that that's unnecessary because our machine learning is finding issues above and beyond.

Starting point is 00:45:01 What are the typical issues? Like, somebody has, like, their zip code, their address is wrong or it's not optimized. I notice when, like, I have an address now, and I put my address in from my, like, you know, little saved poolbar or browser. A lot of times, like, merchants are like, by the way, is this your address?

Starting point is 00:45:20 And then they have an address that's, like, you know, road is spelled out instead of RD or Boulevard, instead of BLVD is Boulevard. And then instead of a zip code, it's a zip code with the dash and the more precise one. Is that what we're talking about here in terms of like the issues people have? Yeah, those are some of the issues. I mean, so those kinds of issues are more common in terms of human entered data, right? And that's some of the data that we examine. But we also examine a lot of machine generated data, right?

Starting point is 00:45:48 when folks are tracking clicks on your website or tracking marketing campaigns, right, or tracking transactions, sales, right, on your e-commerce site. You know, that's kind of, there's some human enterprise pieces of that, but sort of the machine that's running that transaction is outputting that record of that transaction. And so cleaning up what Google, Facebook, meta, TikTok, everybody's doing in their ad networks and normalizing it, this is a major issue for folks. Yes, absolutely. So marketing data is a mess, right?

Starting point is 00:46:18 and it's changing all the time. But even your own internal data, even your own internal transactions or user behavior or tracking are tracking all the time, are changing all the time, you know, often in unexpected ways. At Instacart,

Starting point is 00:46:29 we once lost all of our Android tracking for six months. And no one noticed, right? Because we didn't have a product like an Amelon. We kind of noticed when we needed the data, right? So missing data, right? Or kind of data changing its shape or its distribution in a profound way, right, changing kind of the definition of what's in the data are some of the largest issues that we see.

Starting point is 00:46:56 How's AI going to impact this? I know you're doing machine learning, but, you know, these AI tools are really starting to do interesting things. Yes. They're moving very quickly. Large language models looking at data, you know, I forgot what they called it, but chat GPT4 had a name code interpreter maybe. Yeah.

Starting point is 00:47:16 or like, yeah, upload a PDF, upload a C-As me, and then it just starts telling you about the data. Yes. It's, is it a parlor trick? Is it valuable or, you know, is it not yet valuable? And maybe the juice ain't worth the squeeze with the sort of language model looking at data. Yeah, yeah. So language models themselves, you know, aren't really able to make a lot of sense out of

Starting point is 00:47:40 structured or tabular data, right? A list of transactions. So, you know, what, how code interpreter works. even today, I think they've renamed it. They renamed it to like data analysis or something. Code interpreter basically uses the LOM to write code and, you know, say, Python to analyze that data. The LLM itself can't do it, but it can write a program. And so in some ways, it's a part of which I could still useful, right?

Starting point is 00:48:10 That's what data scientists do with data. They're right programs to analyze, especially when it's at scale. So it is actually quite useful, but it doesn't bring the capability of understanding that kind of structured data to large language models. You know, where large language models are interesting and where they will ultimately apply to products like Catamolo and many other products is in summarizing, right,

Starting point is 00:48:33 and contextualizing a lot of the insights that we spot using other techniques, you know, creating, if you will, kind of news feed for your dataset, where the large language model can consider, assume a lot of insights and signals from the data set and say, this is what's going on. Here's a human readable summary of what's happening in your data. How can people learn more if they want to try the product? They want to engage and learn more.

Starting point is 00:48:57 And by the way, you've been incredible guests. Thank you for being so generous for all this knowledge. And when is the perfect time for a client to use your products and services? Is it when they have 10 people? Is it when they have 100, when they have a million transactions, when they have 1,000? When should they start thinking about this? We're really looking for, you know, large teams, right? So most of our customers might have certainly tens, if not hundreds of people working with data in their organization, right?

Starting point is 00:49:24 These issues start coming up or just incredible data complexity, right? We have customers where, you know, the volume of data is so massive, right? They need every tool that they can manage to get to monitor that data. But folks can learn more about Anamolo on our website. So it's A-N-O-M-A-L-O dot com. Nalo.com. Eli, you have been amazing. I think the audience is going to really get a lot out of this.

Starting point is 00:49:48 So I really truly appreciate you coming on the pod. For those of you listening, you can subscribe to this week in startups and see us. Maybe it's a feature. Maybe it's not. I don't know. It depends on your jam. Go to YouTube and type this week in startups and then subscribe to the channel. Put the alerts on.

Starting point is 00:50:02 There's a little bell there. Speaking of growth hacks. And I'm going to start doing some Q&As. So if you want to get in on the Q&As, you get the alert. You subscribe to the YouTube channel. Go to this week in startups on the YouTube and subscribe. and just write a comment. And that's always appreciated.

Starting point is 00:50:15 We'll see you all next time. Bye, bye.

This Week in Startups - Growth Hacks for Startups from Elliot Shmukler of InstaCart, LinkedIn and now Anomalo | E1926

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.