By 📝Alexander Shartsis in 📝Seed to Sequoia:
A founder asked me this question and it’s a great one. So here’s an answer. If you have other questions, please just reply! I get a lot of responses to these emails and knowing that they are useful makes them much, much more fun to write. The only thing that’s better is questions about what to write about next 😉.
What are network effects and why are they important?
Network effects occur when your product or service becomes increasingly valuable as more people use it. Network effects can also create distribution. For B2B products you can absolutely have network effects within a single customer and in the broader marketplace.
Network effects are behind some of the most successful businesses of all time: Microsoft, Meta (aka Facebook), YouTube, Google, Bloomberg, Walmart, etc. Distribution network effects can lead to the kind of hyper growth VCs dream of. Docusign, for example, grew because someone would receive a request for a signature, and then sign up. Evite had a similar growth pattern for invitations.
The thing with distribution network effects is that they work for anyone in the space. HelloSign launched and benefitted from the same tailwind as Docusign. Paperless Post grew the same way Evite did–and tools like Partiful or Lu.ma are growing the same way today. SurveyMonkey grew to $20+ million in revenue with a handful of people and no marketing or VC, but later entrants (TypeForm, for one) rode the same distribution network effect to displace it.
Reducing customer acquisition costs is a great thing. But the network effects you really want are like Facebook or LinkedIn: the more people join, the richer the network becomes for each user, or the better the product becomes. Retention network effects.
Retention network effects aren’t limited to social media—they show up in less obvious places. Adobe and Microsoft's dominance stems partly from industry-wide adoption, creating a standard that newcomers naturally gravitate toward. Microsoft also encouraged development on its OS, so that more and more companies needed Windows to run other programs. Even though Linux was free, because it wouldn’t run your favorite Windows programs–it didn’t displace Microsoft Windows.
In SaaS there are many less obvious examples. Carta benefitted from both kinds of network effect, distribution and retention. Investors would discover Carta because startups would invite them to track their shares on it. Early on it took some convincing–but not much. Then new startups would be directed to Carta by their investors (or lawyers) who were using it, creating both distribution and retention network effects. Eventually competitors emerged, and were able to gain share using the same distribution network effects–but this is a “winner takes most” market, and Carta has continued to innovate and maintain market share.
Single player and multiplayer experiences
Single player refers to applications or products that are useful on their own. Photoshop, for example, has a great single player experience: you can create digital designs or edit photographs. Quickbooks, similarly, let you keep your books digitally and saved you (or your accountant) a lot of time on double entry. Single player apps can build communities or user bases that are loyal and reluctant to try alternatives. Just try to get your accountant to try a new accounting software and see what happens.
Multiplayer applications are only valuable if you have many people using them at once. Slack, for example, is useless if you’re the only person at your company using it. Fax machines would be useless if other people didn’t also own fax machines (remember those?).
Becoming multiplayer
From these examples it almost goes without saying that some products are inherently single player or inherently multiplayer. You can’t make a single player version of a communication platform like Slack. But what’s interesting is that you can take a single player application and turn it into a multiplayer must-have, beyond simple user familiarity or preference.
Substack, Beehiiv (where this is published), Tumblr were all originally great single player experiences. You wanted to publish a newsletter or blog, they enabled that. Beehiiv came out of MorningBrew, intended to solve a very clear problem for a person who wanted to build a subscription based newsletter business. Yet they show real network effects. Reading this newsletter, you’re exposed to Beehiiv and perhaps consider using it yourself. Beehiiv and Substack have referral networks, making them stickier and creating value for their users. Can you migrate? Definitely. But do people do it often? No.
Other examples might be less obvious. Bloomberg benefits from a lot of network effects in that a lot of investment professionals use it and rely upon it. If you work at a hedge fund, you have a Bloomberg login, and when you switch jobs, you’ll have one at the new fund. There are network effects to having a global staff of reporters, and data partnerships. But what’s surprising about Bloomberg is that investment professionals use it to communicate with each other, as a trusted closed network (like email–without the spam). Bloomberg took a great single player experience–a research and data platform useful to an individual investor–and turned it into a must-have professional communications network.
The cold start problem
Businesses that are multiplayer by nature without a single player experience are much harder to start: this is known as the “cold start” problem. Some, like Slack, *seem *easy but that’s because they found a niche that was whitespace. In Slack’s case, this was startups. Slack replaced IRC (internet relay chat, or a real time chat app you could host yourself) that was popular on engineering teams at the time. Two companies I worked at in the ~2010 era used it extensively. One even had a chat bot built in (if you misspelled a certain teammate’s commonly misspelled name, it would harass you).
Slack turned out to be difficult to sell to larger companies. They simply weren’t interested in a new way to communicate, and without getting much of the team’s buy in–especially leadership–Slack was dead on arrival. My wife’s 1,000+ person company wouldn’t even consider it even though they were based in San Francisco, surrounded by hundreds of companies using Slack every day. Eventually Microsoft entered the market with Teams, and won over a lot of enterprise customers (it helped that it was free!). Who I guess just wanted a crappier product from a better-known vendor.
For a startup founder, unless you can find that pocket where your multiplayer-only app resonates, it’s far better to find a great single player experience, grow that, and then add the network effects later. At Drawbridge, we launched as an ad network (which, ironically, we were able to do without any network). Eventually we built the network effects into the business, but at first we were just a great way for advertisers to acquire mobile app users.
Building network effects into your business
We can’t all be Microsoft, with a network of nearly every company on the planet as a customer–making it easy to launch new products and reach critical mass quickly. But we can be thoughtful about how we structure our product and ecosystem to maximize network effects.
Network effects in SaaS and B2B products often involve creating ecosystems or integrations that strengthen with adoption. Slack has added cross-company communication, which not only helps it reach more customers but also helps make it stickier. Slack also integrates a lot of applications, and encourages more apps to be built within it. Platforms like HubSpot or Salesforce become increasingly valuable with each additional integration, app, or user, expanding their ecosystem and deepening lock-in.
To foster network effects in B2B SaaS:
- Encourage integrations: Make your software essential by integrating deeply with popular tools already used by customers. This can also help distribution.
- Leverage collaboration: Integrating other organizations makes anything sticker. Who wants to ask their VCs to change from Carta?
- Aggregate valuable data: Offer analytics or benchmarking insights that become richer and more accurate as more customers contribute data.
- Create community: Vibrant networks of users and service providers can help you both grow and retain customers, making it harder for customers to switch.
- **Build in-company network effects: **Get more users in your customers’ company to use it, making it stickier.
You can also reach massive scale, creating network effects. Bloomberg may be a digital example of this. Amazon is a great physical example: it has so many customers, and so much share of wallet amongst those customers, it has built a delivery network that can get a product to a customer within hours. This makes the product demonstrably better (faster delivery) and it makes Amazon harder or even impossible to displace by a competitor. No single player experience can compete with the Amazon delivery network.
Same company network effects
If your product or software is only used by one or two people at a company–beware! This is very bad. Even if this is super valuable software, if that person or people leave–the whole account is at risk. A common example is pricey data providers. A data provider might have a 1 seat subscription. Bob in data science logs in every once in a while and downloads some data. When bob leaves, and the CFO finds a $100k line item in the data science team’s budget–what do you think happens next?
Contrast that with JIRA, which every engineer and product manager uses every day. When Bob in data science leaves, his tickets get reassigned and life goes on. Nobody at Atlassian (JIRA’s parent company) loses sleep. They might even pick up a customer if Bob’s new job doesn’t use JIRA yet–Bob might suggest it to management, and now JIRA has retained a customer and added a customer. That’s good for NRR!
It is tempting to focus on power users and dive deep on their needs, but if you can, it is equally important to create broad adoption and network affects within your customers. My personal experience with Notion is instructive: we had a teammate who had used it elsewhere and really liked it. She built out a few cool integrations and automations in Notion. She’s gone now. We downgraded. The learning curve felt steep, nobody else really figured it out, we stopped needing one or two of those integrations, and it was easier to just rebuild things elsewhere or live without. Had Notion been more successful at getting others in my company to embrace it, things might have gone differently.
Building network effects within your customers can be as simple as adding an “invite” step to your onboarding flow. You do need good reason to invite people, and if it seems like a blatant way to get another monthly per-user fee, it’s unlikely to be successful. This is also why I don’t love per-user pricing in most instances: it disincentivizes same-company network effects. Just have Bob go do the thing, rather than pay for another seat. Or if you need to do it, get Bob’s login. That leads to churn.
Beware fake network effects
Trying to manufacture network effects without genuine user value rarely succeeds. Consider failed social networks. The catch with a network dependent business is if you fail to attract the network you don’t have a business. Ello, marketed as a Facebook alternative, spiked briefly but lacked authentic engagement. Google+ was used by every Google employee–but nobody else. Because it failed to capture genuine user interest, it was shut down. The closest one to succeeding, Path, which was started by an early Facebook employee to make a more personal network. The thing with that–specifically, throttling invites so your network was just actual friends, not everyone–was that it killed the network effects and after raising $68 million, went poof.
Other products might look like they have network effects at first but, in fact, don’t. EV charging software is a great example of this. In theory, EV charging companies run networks. By having a larger charging network, the software should be more valuable–both to drivers and others in the industry but also to the hardware manufacturers. This is, however, not the case.
The reason is that the hardware and software are interchangeable. They are built on an open standard (called OCPP), and all the software and hardware do pretty much the exact same thing. Just because a driver has an app and, in theory, would want to use that app to charge everywhere, the charger purchase is large enough that price sensitivity prevails and nobody really cares which app they have (they’re all free to download, anyway). A contributing factor is the person buying the charger is usually not the person charging EV at it, or choosing the software that goes on it.
Of course, none of this stopped a lot of EV charging founders from raising on a network effects promise. But it’s important to adjust your strategy as you go, and clearly network effects in this space have not played out as expected. If they had, one of the largest EV charging networks in the US–Chargepoint–wouldn’t be nearly insolvent and valued at $286 million, down from a peak of $8 billion in 2021.
If you’re adding network effects to your product, make sure that they’re genuine and real. They might not be as valuable to your users as your single player experience, but they should be valuable. At TripIt, we launched a company plan where whole companies could share plans amongst their team. It was maybe 1/50th as useful as the core TripIt product–but, it encouraged some additional adoption and created a new revenue stream. It also opened up the ability to partner with travel agents, which became a growth channel for us. Your network effect feature doesn’t have to be perfect or even very good to be useful to you strategically, as long as it’s real.
Pricing power
Network affects are directly correlated to pricing power, even if it’s nonobvious and even if pricing doesn’t go up. Netflix, for example, benefits from network effects and has a history of raising prices over time. Bloomberg also, famously, raises prices every single year. But even though companies like Slack don’t raise prices often–they also never have to lower them.
Software is a zero marginal cost product, and should therefore trend towards $0 in price. After all, if your incremental user costs you $0 but you make $1,000, someone should be willing to come in and charge $100. Or maybe even $1, if they can acquire the user inexpensively enough. Customer acquisition cost (CAC) has become the variable cost in building a software business and can create some defensibility–but only some. Network effects can create much more.
My favorite example of network effects gone wrong today is YouForm, which I’ve tried and use. TypeForm is ungodly expensive for what it is (just one guy’s opinion). They did that, probably, because they built great network effects and have lots of great integrations. But YouForm not only copied their software’s key innovation–sleek, minimalist design–but also copied many of their other features. And, guess what? It’s also a survey software so it also has distribution network effects. But at a small fraction of the cost of TypeForm. If I were TypeForm, I’d be changing my pricing defensively. This always ends badly.
Today the biggest threat to software companies is AI generated software. Last year Klarna, the lending company based in Sweden, let slip that it had left Salesforce and around 1,000 other SaaS vendors for a combination of in-house built software and a handful of SaaS vendors. You can bet the vendors it kept had some powerful network effects. If you only have a single player software, be worried. If you’re getting a $1mm annual license, and a developer could build your software in six months using Cursor or Replit, well, that’s not good either. I think those most at risk are PE-owned SaaS companies that have shifted to value-extraction mode, but it might get any of us.
Conclusion
Distribution network effects can help your company grow quickly. Retention network effects help you retain users and extract value over time; they provide more of a moat than distribution network effects, because usually the distribution tailwind that helped you is endemic to your category and will help your future competitors.
It’s best to choose a space where network effects are possible, but start with a single player application to avoid the cold start problem inherent in a network dependent business. Then add network effects to your business later, as you mature, in order to retain customers, create pricing power, fend off competition, and keep growing.