Use the network effects to jump-start your growth engine

Angele Lenglemetz at UXDX Community: EMEA Series. Video: https://youtu.be/3nk5YSQnTag

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

Introduction and three network products

[00:00:00] Angele: Thank you. So today I'm going to talk about how you can use network effects to jump-start your growth engine. Very quickly, a little bit about myself. I've been working in tech for the past six years. In a previous life I was a consultant working on capital markets, in particular in large banking institutions like Goldman Sachs and others. And it happened that after a while, despite working in tech, I felt like I didn't really understand what was happening in the background, which is when I decided to follow a coding boot camp called Le Wagon. For me that was really a life-changing experience, where I learned a lot and moved towards the product path.

[00:00:41] I've since worked in a few companies. I started at Streetbees, which is a market research company powered by AI. Then I joined Wise, a remittances company, where I was looking after their USD product outside of the US and grew it from pretty much zero to a billion monthly. And more recently I've joined Shares.io as a senior PM. Shares is a social investment platform where friends can invest and build strategies together. But anyway, enough about myself.

[00:01:13] The reason why I decided to talk about network effects today is because all three companies that I've just mentioned, despite being very different, are really impacted by and leverage network effects. So I will try to use examples from these companies today to show you how you can do it as well in your own company, and what the things are that apply in every single context or product.

[00:01:41] But first, what are they? Network effects are when an increased number of people or participants improves the value of a product. And it's not just Facebook, Instagram, LinkedIn or Snapchat. Actually, pretty much every single product has some sort of network effect.

[00:02:06] I'll walk you very quickly through my current company and previous ones, because I'll be using loads of examples from them, and I think that unless I explain roughly what they do, it will be really hard for you to get it. Shares.io, as I was saying, is an investment app where friends invest and build strategies together, and you can think of it as an investing platform and a social media. On the other side, Wise is a product that you can use to transfer money abroad easily and quickly, with very low-cost money transfers, and in terms of size they are quite big. They IPOed in 2022.

[00:02:42] And finally, Streetbees is a human intelligence platform which is disrupting market research, and the product is quite interesting. On one side you have a dashboard with insights available to customers like Don Pepsi [?], all the big FMCG companies. On the other side you have a mobile app that is used to collect always-on data. This data can be structured, like your usual survey, but also unstructured, like photos and videos. We had about 4.5 million users in 189 countries when I was there. And in the middle, they're using machine learning to analyze this data in real time and gather insights.

Direct, data and cross-sided network effects

[00:03:26] So really quickly: there are about 14 different types of network effects, but today I'll focus on these three, which I'll explain really quickly. The first one is direct, and this is when an increased number of users improves your product. You can think of Facebook or Snapchat or Shares: the more users the better. A very simple product, maybe the first one, was the telephone, where the more friends you have with a telephone, the more value you get out of it.

[00:04:01] The second one is data, and that's when an increased number of users or data improves the quality of your product. That's usually every product that has some sort of predictions or personalization involved. Flo, which is something you can use to track your period, is a good example, because it tries to predict what's going to happen. Google is personalizing your experience based on your previous searches and what other people have done. And Wise uses this data as well, in order to improve the predictions on how much the fees will be, or how long a transfer is going to take to reach the recipient.

[00:04:40] Finally, cross-sided network effects. That's when the value of a product increases for users as users of another type join the platform. That's usually all the marketplace products. You can think of Streetbees, because we have on one side those large companies and on the other side these users, or Etsy, with the sellers and the buyers. Obviously for the sellers, the more buyers on the other side the better, and the other way is true as well: for the buyers, the more sellers the better. So these are really cross-sided, and they have a little bit of a different dynamic.

[00:05:22] What's interesting as well here is that a lot of companies with very high growth actually combine different network effects. Facebook, for example, is using direct, data and cross-sided network effects, and this really gives them a huge advantage. The direct network effect comes from all your friends being on it. The data one from personalizing your feed and being able to provide you a very specific and, in theory, great experience. And cross-sided is in the marketplace sort of products and areas that they've built a bit more recently.

[00:05:59] I'd be curious, maybe not to have you raise your hand, but at least to have you reflect on whether you believe your product, or the company you're working at, is impacted by or leveraging some of these network effects.

Why it matters: the Wise flywheel

[00:06:16] But now, more importantly in a way: why do we care? Why does it matter? Well, products with a mix of these network effects have higher defensibility and explosive growth. I'll show you a little bit of the mechanism at hand in the background. Let's take Wise as an example. They have two types of network effects. The first one is direct, where paying another user on Wise is very cheap, fast and convenient. And they have data network effects, where the more data, the more accurate the predictions on fees and speed for the transfer, which is really important whenever you send money, especially if you're sending quite a large sum.

[00:06:58] And what happens in the background is that this better experience that they can offer from having more data and users is very likely to increase retention down the line. It means that the acquisition is very, very optimized. I think that when I left, about 70% of users were coming from word of mouth, so very much organic. Which leads to, obviously, better monetization, because the experience is great. Users stay for longer and have a higher lifetime value, and the acquisition costs are quite low. So they have a higher average revenue per user, which means you can invest in your product, leading to higher retention, creating this virtuous circle.

Ten tips for building network effects

[00:07:46] It's all nice. But then the bigger question now is: what can you do, and are there some things that you can do in every single product in order to start building those network effects and those virtuous loops? I've gathered my top 10 tips and tricks. I'll go through each of them and give you a concrete example of how we did it in one of the companies that I've previously mentioned, so that it gives you an idea of how, even in a small company, even in an early stage one, or also in a large one, you can actually build them.

[00:08:23] The first one is identify your most valuable users. The second one is define your minimal density. The third one is network density is more important than total size. Number four is bootstrap your content. Number five is collect your data smartly and automate gradually. Number six, have a strong initial value proposition. Number seven, do one thing well. Number eight, read The Cold Start Problem by Andrew Chen. Really great book. Number nine is use your first audience to build the next one. And number 10, don't lose faith, because it only takes one successful event.

Most valuable users and minimal density

[00:09:04] Number one: identify your most valuable users. In every single network product, and especially the ones that are cross-sided, there is a side of the network which is a little bit more valuable, and we usually call it the hard side of the network. It's very important, before you even start doing anything, to be clear on what this side is. Usually for a marketplace it tends to be the seller. In a dating app it's going to be very attractive people.

[00:09:43] And in an example like Streetbees, it was actually users who were a bit older. The reason why is because in every survey, whenever you collect data and run this market research, you want a sample of users that is representative of your population, so demographically representative. You need them to be of every age, gender and wealth. What was challenging for us is that our value proposition for users was: do this survey and take some photos to earn like 30p. And if you're an older, rich person, that just isn't very attractive for you. So we really had to think a lot about our pricing and how to adapt the product in order to be able to retain these users, and ensure that our CAC for them was under control. And that meant that we actually gave them, very regularly, specific tasks designated for them, even though we didn't necessarily have a customer behind them. And that was something which really made a difference for us.

[00:10:51] Then, identify your minimal density. Every product has a minimal density, and the density is how many users you need for them to start getting value out of your product. The reason why that's super key is because it's actually going to determine your launch strategy, and also how much defensibility your product has down the line. If you think of a telephone, you only need two users. They can call each other. Great, they get value out of it. If you think of Slack, it's going to be 10 users, basically the size of a team. If you think of Visa or Mastercard, it's actually probably roughly the size of a city: 100k users and, I don't know, a thousand shops. So you can imagine that the investment to launch Visa or Mastercard is very different. And knowing this in advance is very important, because it's going to really help you define your launch strategy and how you want to get started.

Density over size, and bootstrapping content

[00:11:53] Then, very important: your network density is more important than your total size. Once you know who your valuable users are and how many of them you need, well, you need to get it going. And something important to remember is that usually your first network, and how many users you need, is actually smaller than you might think. The size of your atomic network, and an atomic network is the smallest network you can get on your product where they get value out of it, can be very small, but it needs to be extremely interconnected. There's a reason why Tinder, Facebook, Snapchat all started in universities, and that reason is probably because college students tend to be very social and interconnected, and they have a bit of time. So they are perfect to launch a viral product and get those network effects going from the start.

[00:12:52] An interesting learning for us at Shares was that when they launched in March, they had quite a big budget and acquired about 120,000 users in just a month. It was quite insane, number one in the App Store. Unfortunately, those users were not deeply interconnected, and we didn't manage to retain them, because they didn't get much value out of it. They were just like, "Well, I don't have any friends on your product, so I don't get it." On the other side, Tinder organized, apparently, a very cool party at UCL, on a university campus, and got 500 users who were very much interconnected, and that was enough to get them started. So remember: density over size. That's what matters.

[00:13:38] Then number four is bootstrapping your content. If your product relies very much on content getting created, think of a Reddit or these sorts of communities, it's likely that, unfortunately, it will be hard at first to get enough users who create valuable content. And my recommendation is: fake it until you make it. I know that when we launched a community product at Shares, what we did is we asked pretty much every single employee to start creating content and posting, and to be very active on those community pages. And we then actually contacted some investors who wanted to become content creators, giving them an incentive to post regularly. It's a strategy that's been used in many products, and the whole idea is, until you have enough organic content creators, just try to insert them, or fake it. Because network products are, in a sense, based on their hype, and no one wants to go to an empty party. No one wants to come to an empty product. So make sure that you really put all the chances on your side by creating this content.

Collect data smartly, a clear value proposition, do one thing well

[00:14:54] Number five is collecting data smartly and automating it gradually. That one was a learning from Streetbees, where our whole product was based on machine learning. And what most people underestimate is how much data you need in order to get started with ML. Honestly, unless you're sitting on a pile of cash, you will need to generate revenue before that. And the general thing is that it's fine. It's just a matter of being smart about how you collect data, and also how you automate it gradually.

[00:15:28] In terms of collecting data, we built a few strategies to collect as much information as possible. One of the things was, we told users that if they were doing free surveys, so unpaid ones, they would be more likely to get the paid ones. That was very important, because it allowed us to get more content, but also to keep them engaged by having content in the app all the time. And finally it created a little bit of hype around how to get those paid surveys.

[00:15:59] And another learning, maybe even more important, was around the fact that you really want to set yourself up for success. At first, when we started, we didn't have the ability to run machine learning. We didn't have enough data. So a lot of the analysis was manual. Imagine you collect a thousand pictures of bottles, people showing you what they are drinking, and you have someone go through them: this is orange juice, this is milk, this is coffee. This data is actually extremely valuable for training a machine learning model, and that works with annotation. Unfortunately, because at that stage we didn't really know where we would end up, a lot of it was done in random spreadsheets, and when we actually wanted to start doing machine learning, it was really complicated to bring all this data into a database. So even if you start manually, be smart about it, and think about how you will automate this down the line, to build the right data infra from the start.

[00:17:04] Then number six is offering a clear value proposition to your audience. Wise is a very good example of this: fast, cheap, convenient. I can explain this to my friend in less than 10 seconds. And it's important because usually you want network products to be recommended. You want friends to tell each other, "This is great, you should use it." And personally, I don't recommend products if it takes me more than 15 seconds to explain the value of them, and I think most people are the same. So keep it simple, very easy to explain. It's really important down the line, to make sure that users recommend you and your product is likely to go viral.

[00:17:45] Then, doing one thing well. I think it can be very tempting to build a lot of things, but when you look at a lot of network products, they are very simple. Uber lets you order a ride in one tap. And interestingly, when Uber was first built, the engineering behind it was quite simple, and they had actually outsourced it to a company in Mexico. Meaning that when they had to scale the product up, because it was actually working, and started hiring in San Francisco, they had to give a Spanish to English dictionary to most of their engineers so they would be able to translate the codebase. Similarly, Facebook was started by college students, and no offense to college students, they tend not to be very experienced. So focus on the simple thing.

[00:18:39] At Streetbees I had this example where we had two options. One was improving the sign-up funnel, optimizing it basically, which honestly isn't super sexy. The other was adding new shiny features, which was obviously a lot more exciting. But when looking at the impact, it was quite clear that improving the sign-up funnel was way more impactful and important. So that's what I did. And indeed, by removing a few steps, making it clearer how long it was and how many steps there were, and doing a few UI tweaks, we managed to improve conversion by 24%, and the average time to complete went from over a minute to 34 seconds. Which was really, really important, as we had to acquire lots of users really quickly for them to do surveys. So don't necessarily add a ton of things to your product, but really focus on quality and doing one thing well.

The Cold Start Problem, using your first audience, and not losing faith

[00:19:41] Number eight is reading The Cold Start Problem. Honestly, it's such a good book, and I think loads of the things that I'm talking about here are covered in this book, but in way more detail and with a ton of super fascinating examples.

[00:19:55] Then number nine is using your first audience to build the next one. That's something we started doing at Wise: you set up a transfer for your friend, and at the end we would ask you for the email of the receiver. And the way we used it was, we would send an email to that person and let them know, "Oh, this person has set up a transfer for you, and the money is on the way, so don't worry." And it would do two things. For a user, it stopped you from having this super awkward conversation of, "Hey, did you send me money?", which no one likes to have. And for us it was actually a great way to show that this great experience was provided by Wise. So next time you send a transfer, why not use us? That's the kind of trick that you can use to get your first users to convince the next ones to join you.

[00:20:52] Another example was at Streetbees. This is Beno, and Beno was one of our power users in Kenya. At some point, Pepsi came to us and was like, "Well, we would like to have a million photos of Pepsi, in as many shops as possible, everywhere in Kenya, in 3 days." And honestly, we were like, "Okay, we don't have that many users in Kenya, definitely not even a thousand. So what should we do?" And what we did is we contacted Beno, and he created this WhatsApp group, started inviting lots of people, sharing the survey with them, telling them to sign up on Streetbees. And it actually went pretty viral, and within a few days we had so much data that we actually had to close the survey. And that's a strategy that we then reused in multiple markets, using those people, who became our ambassadors basically, where we created a real partnership. Some of them then came for an internship during the summer, and things like this. And that was instrumental in getting us where we needed to be, to fulfill these surveys that we were getting quite often from large customers.

[00:22:07] And finally: don't lose faith. It only takes one successful event. As I was saying before, Tinder had one great party, and that's actually how it all started. On the right side, this is Obimi [?], an app that went completely viral in Ukraine, that allows you to connect with your friends and people you're very close to and share your mood. I was talking to one of their early employees recently, and he was explaining that for eight months the product was pretty much ready, but no one was using it. And it was actually one very successful TikTok video that went viral which made the whole thing start, and it's become one of the top five apps in Ukraine. So as long as your product is solving a problem, give it some time, and also give yourself chances and opportunities for it to be successful, because there's a bit of luck, and you have to trigger this luck. One thing at some point will take, and it's hard to know which one, but just don't lose faith, basically.

[00:23:10] So the conclusion here is, first, network effects are not just for social networks. They are really everywhere, and they are a very powerful way to increase the defensibility and growth of your product. What I would ask you to do is first identify yours in your company. Can you think of any of those networks or loops that you have? And start leveraging them. I hope one of these will apply to yours. There are little bits here and there, and I think it depends on the stage of your product. Some of you might be working on a really large one, or a very early stage one. I've tried to put in a mix of small and large companies. But hopefully you find one that you can apply and start using, and in the worst case, just read The Cold Start Problem, because it's very good. Thanks a lot for listening.

Q&A

[00:24:12] Host: Excellent, thank you very much. I really enjoyed that, because you made it very tangible with all the different examples. I thought it was very, very well done, so congrats on that. One thing that just jumped to my mind was Udemy. I don't know if you've heard the story of Udemy, the online courses. They had that problem. They had no courses and they had no people. So they just made up their own courses to try to get it started, and that was enough to start the ball rolling.

[00:24:45] Angele: Yeah.

[00:24:46] Host: So please do share your questions if you have any. I have one from Sean here, but please do share some more, and I'll put them to Angele. The one from Sean: you might have touched on this actually as you went towards the end, but you mentioned, with Wise, the value flywheel, that it was word of mouth that drove it. Was that word of mouth just that email that you described, or were there other ways that you were trying to drive word of mouth?

[00:25:18] Angele: I think in general, the thing with Wise is that they had a lot of these little strategies here and there. Another example would be, when deciding where to open the next market in terms of international expansion, they would have this sort of waiting list, and this was a really great way to evaluate how much demand there was for the product, but also to then email all the people who were on the waiting list when we would launch. So it's a mix of many small strategies like this. The one that I shared in this presentation is one of many. I think in general the increased retention comes from providing a great experience to your users. That's probably the first thing: solve a problem that exists, build a good solution, and add on these little tricks. But the idea is those little things can just bring you a bit further than you would otherwise get.

[00:26:15] Host: Excellent, I love that. And oh, here's a question: what was the most interesting problem to solve at Shares, and what was the complexity around it? That's from Karen.

[00:26:30] Angele: I think the most challenging part is probably connecting people in a meaningful way. We're building an investment product. At the end of the day, in a way, it's not simple, but it's not that hard. It's already been done, and there's a bit of a playbook around it: you need to enable people to pay in, and then to buy stocks, and then to sell them, and to withdraw their money. It's quite straightforward. On the other side, I think connecting users in a meaningful way, especially in the context of finance, is quite hard, because there's a kind of taboo, and some people might not really want to mix friends and money, actually. So the whole value proposition might not work for them. And I think that's one where we're slowly learning: do users actually want to connect and talk about this with their friends? And if yes, how do we bring value to this, and how do we make this conversation meaningful? That's one of those challenges.

[00:27:34] Host: Yeah, that's a great example, because I didn't think of it, but yeah, the whole mixing friends and money might not actually be something that people are too keen on. Any other questions? Just looking at the different channels. The zero to one problem, I guess, is the biggest. I haven't read the book, The Cold Start Problem, so I must give that a read, because you've given it a glowing review. Yeah, there's my Christmas present sorted. We're almost out of time, and I don't see another question coming in. So thank you very much for sharing. I really enjoyed that. I hope everybody out there did as well. So thank you very much.