Login to access Everything Marketplaces
Don't have a member account? Join here.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
We're sharing a guest blog post from our community EIR Blake Hirt covering network effects, which also helps highlight the importance of having strong network effects when building a marketplace. This was previously shared as a post in the community here.
Hey EM, I’m continuing my series of posts on what I believe to be the anatomy of a successful marketplace.
In this post, I’m going to provide an overview of the second criterion: strong network effects. This post does not dive into all of the nuances of network effects, but the goal is to provide a better high-level understanding and what network effects look like for marketplaces. For further reading, you can reference some of the resources at the end.
Network effects is a term that’s thrown around a lot when it comes to marketplaces, but it is often misunderstood. I’ve found that many early stage marketplaces believe they have them, when in reality they do not. Additionally, network effects often get confused with other things like virality.
In this post, I’ll help define what network effects are, break down the main types, cover examples, and share tips you can use to build a marketplace that benefits from strong network effects.
The basic meaning of network effects is that the more participants you have in your marketplace, the more valuable it is to current and future participants. In other words, the product or service becomes even more compelling and useful as more people join the marketplace. This last part of the definition is key, since additional participants need to make the product more valuable to both new and existing users. If that’s not the case, network effects do not exist (or are very limited).
There are generally four types of network effects:
Network effects can be one of the most valuable and defensible aspects to your marketplace. Once your flywheel is at full force, you become the #1 player, and it becomes really difficult for another company to displace you.
From a metrics perspective, network effects improve your customer retention and LTV, and each subsequent cohort of users should have an even higher LTV because they’re experiencing a better product. DoorDash has shown this to be the case for their cohorts.
If you have a commodity business, network effects should improve speed or price (e.g. Uber) and for non-commodity businesses, network effects should improve selection (e.g. Airbnb). In both cases, retention rate should increase.
Network effects are often confused with virality. Network effects improve the value of the product, thereby improving LTV. On the other hand, virality means that users organically share the product with other users, which drives down CAC.
It’s still widely debated whether network effects reduce CAC. On one hand, having a more valuable product should make it easier to sell, resulting in a higher conversion rate. On the other hand, having network effects doesn’t necessarily improve your go-to-market or acquisition strategy. Personally, I think network effects should have a minor impact on CAC but more meaningful impact on LTV.
You can have network effects without virality and vice versa. Of course, it’s best to have both.
Below is a famous napkin sketch of Uber’s network effects. More demand results in more drivers, which results in more geographic coverage, which results in faster pickups (i.e. ETAs), which results in more demand. Additionally, lower driver down time (i.e. higher utilization) results in lower prices (due to drivers being able to do more trips per hour), which results in more demand.
It was a beautiful cycle that I saw upfront and personal for 4 years.
For marketplaces, cross-side network effects should occur for both sides of the marketplace. In other words, more supply should make the marketplace more valuable to demand, and more demand should make the marketplace more valuable for supply.
As mentioned above, in order to truly have network effects, more participants need to make the product more valuable to both new and existing users. Marketplaces that connect users with a therapist, financial advisor, personal trainer, or other monogamous relationship do not have network effects, as a new supplier does not add any value to a user once they’re already connected with someone they like. It’s arguable that these businesses are not marketplaces at all, but rather lead generation businesses. I’ll go into this in more detail in my deep dive on my fourth criterion: demand-side promiscuity.
The best situation for a cross-side marketplace is where there are four-way network effects, in which more supply makes the platform more valuable for both demand and supply, and more demand makes the platform more valuable for both supply and demand. Amazon is a good example here. Not only does more supply improve the user experience through better selection, but more users provide more reviews, which provides more peace of mind. Additionally, more demand benefits suppliers with more revenue, and more supply increases the number of eyeballs on the site, which provides other suppliers more potential buyers.
However, for many cross-side marketplaces, network effects can go in the opposite direction for those on the same side of the network. Most Uber drivers do not want more drivers on the road competing with them for rides (many of them have told me this to my face). And if I’m trying to request an Uber outside a bar at 2am, I don’t want dozens of other people requesting at the same time, as it’ll reduce reliability and drive up surge pricing.
Unfortunately, network effects are not something you can just magically insert into your marketplace. They need to be a natural part of the value that your users get from using your product.
However, there is one way you can build network effects quicker: constrain. By constraining the initial audience for your product, it allows for more density, increasing the odds that an additional participant will make the product more valuable for everyone.
OfferUp spent two years building up the Seattle market before expanding. They started with a very small geofence and slowly expanded over time.
You can also constrain by category, not just geographically. Ebay started with collectibles like Beanie Babies. This constraint ensured that each new supplier improved selection for buyers and that buyers were high intent on buying collectibles, increasing supplier sales.
Another way to help with liquidity is if your demand becomes supply and supply becomes demand. By having a single participant play both roles, it can improve the experience for everyone. Airbnb benefited from this dynamic.
There may be some surprise network effects that pop up. Danny Martinez noted that at Prolific, they discovered a bunch of user researchers were using the platform with their colleagues across research teams. The product was always built for single player mode, and this revealed that creating a multiplayer mode would make the platform more valuable and lay a stronger foundation for growth.
The main question you want to ask yourself as you grow is: Does adding this new user increase the value for all of my other users? If not, the solution is usually to constrain more so you can start to build more liquidity and network effects.
While network effects can theoretically be limitless, there generally are limitations.
First, if you’re a hyperlocal marketplace, the network effects are only valuable in a constrained geographic area. If I’m a homeowner in Chicago, then Thumbtack adding a handyman in Atlanta adds no value to me.
Alternatively, if you’re a global marketplace, your network effects can compound for a long time. Airbnb is a classic example here, and that’s why their competition has been much less fierce than that of Uber’s.
In fact, at Uber, Travis was transparent to the company that we had to raise as much money as all of our global competitors combined, plus be more efficient with how we deploy that capital, in order to win. Fighting all of these local battles was very costly. Uber ultimately had to raise $25B, while Airbnb only had to raise $6B.
Of course, even Airbnb has limitations. Adding a shack in the middle of Antarctica doesn’t add value to many users. But their theoretical limit is much higher than that of a hyperlocal marketplace.
Facebook also has limitations, as users only befriend people they know. However, the newsfeed has now expanded to show content that’s beyond what friends have posted, and the company has leaned more into Facebook Groups. This way, users can connect with an unlimited number of people that they’ve never even met.
Ultimately, the higher your theoretical limit, the better. You should have a good idea of where your marketplace stacks on the scale, even before launching.
Network effects are a core asset and moat for your marketplace, but only if designed in such a way. Ultimately, as more users join the platform, you shouldn’t only be celebrating internally, but all of your existing users should be celebrating as well.
A big thanks to Sameer Singh, Lazar Javonic, Danny Martinez, and Mike Williams for their input to uplevel this post and helping review. For additional reading, definitely check out the resources below.
If anyone else has any thoughts or questions, feel free to reply below or reach out directly.
You can connect with Blake to discuss this post in the Everything Marketplaces community here. A big thanks to Blake for also being an active EIR in the community, where he is often sharing his marketplace experience, insights, and helping early stage founders.
