Improving Segment’s Growth Model for Two-Sided Marketplaces
We hear the term network effect used very often when we talk about some of the fastest growing start-ups in the world.
Looking around at companies like Airbnb, Uber, Etsy and even crowdfunding platforms like Kickstarter, the network effect is undoubtedly one of the primary driving forces behind their success.
But what is the network effect really? How can one define it?
I recently read an interesting article published by the data integration SaaS company Segment, that elaborates on the subject quite well.
Inspired by HBR’s Strategies for Two-Sided Markets, blogger Andy Jiang, explains how Segment, the company he works for, models growth in a two-sided marketplace.
In this article, I will recap some of the most important takeaways from the original article and later expand on the model. For a more thorough understanding, you can refer to the original piece found here.
What are the forces that drive marketplace growth?
There are 6 forces that contribute to the network effect in a marketplace made up with buyers and sellers.
- Buyer-to-Seller Cross Side: Buyers refer sellers to the marketplace
E.g. “I want to rent a place. You should list your place on Airbnb”
2. Buyer Same Side: Buyers refer other buyers to the marketplace
E.g. “Why would you rent a place on Craigslist? Craigslist is sketchy, you should check out Airbnb”
3. Direct to Buyers: The marketplace directly promotes itself to buyers
E.g. *Airbnb Ad — Ditch the hotel! Book a place to stay on our platform!*
4. Seller-to-Buyer Cross Side: Sellers refer buyers to the marketplace.
E.g. “I’m renting out my place on Airbnb. Sign up and make a reservation!”
5. Seller Same Side: Sellers refer other sellers to the marketplace
E.g. “I listed my place on Airbnb to rent. You should too!”
6. Direct to Sellers: The marketplace directly promotes itself to buyers
E.g. *Airbnb Ad — Rent out your apartment to make extra money! *
Looking At How Segment Models Growth
As a SaaS marketplace company made up of buyers (customers) and sellers (integration partners), Segment models each of the 6 growth channels as its own component. This allows Segment to examine each of the forces in isolation, assign quantitative metrics to track them, and organize marketing resources around them to improve measurable performance.
For example, Segment’s product team is responsible for managing customer same side growth. With a goal of enthusing customers to refer more, the product team measures their success on referral rate, and devises marketing tactics to improve the ratio of new customer gained per existing customer.
Segment also uses this model to make data-informed decisions of where to best allocate marketing dollars. To do this, Segment runs optimization simulations, keeping growth rates of 5 of the 6 forces constant and changing one. After running every possibility, Segment identifies which growth channel to invest in in order to maximize total growth.
In the example that Segment uses, the company projects that an investment in partner-to-customer cross side marketing will increase new customers per partners per month from 4 to 8. After running simulations on each of the growth levers, Segment can determine whether this is the most effective marketing investment. If so, then Segment will invest more in the partner team to increase customer sign-ups attributed to Partners.
Improving on Segment’s Model
I think Segment did a great job of fleshing out a two-sided marketplace model. What stood out to me was how they well they aligned the quantitative metrics to informing business decisions.
However, I do feel as if some improvements can be made.
First, the original model has no churn mechanism where customers are lost. Currently every new customer added is assumed to stay forever. Not only is this not how marketplaces work in the real world, but this kind of compound growth over a long period of time is going to heavily inflate growth numbers. If the purpose of this model is to efficiently allocate marketing efforts and maximize growth, this kind of oversight may heavily skew decision making.
Second, the way Segment defines total growth doesn’t seem ideal. A basic gut check tells us customers and partners should definitely be valued differently. If a partner brings in 3.8 customers a month, and an existing customer brings in 0.02 customers per month should the model really be valuing the partners and customers same the way?
Lastly, if we think about the purpose of the model as a tool to help us deploy marketing dollars as effectively as possible, shouldn’t we think about valuing total growth as some kind of dollar value? This will allow us to compare apples to apples.
Adding Churn
I added a Churn mechanism to emphasize the importance of retention rate in a two-sided marketplace. It is important to be able to add new customers to the marketplace. But, if your product is not sticky you will lose customers faster than you can add them!
This churn rate I added is the weighted average churn rate based on average churn rates in customer growth channels. It is hardly a perfect calculation, but for simplicity’s sake it will have to do.
Let’s see what would happen if monthly churn rate doubles.
Yikes! Churn is scary!
If we follow Segment’s previous example of aligning the most important levers to internal project teams. The Segment team should integrate churn into this equation because churn/retention may be the most important growth metric.
Optimizing for Total Customer Value
If we evaluate total growth the same way Segment did, then it is a no brainer that 10,000 customers are better than 1,000, and that Segment should allocate marketing resources to capturing that 10,000.
However, let’s say Segment adds 10,000 customers with 10% monthly churn rate, and $4 monthly recurring revenue. This is equal to $400,000 in total customer value.
Now the 1,000 customers have 5% monthly churn rate, and have a monthly recurring revenue of $25. This is equal to $500,000 in total customer value.
Now it becomes clear whether more customers isn’t always better!
Marketplace growth isn’t always about maximizing the total amount of customers. A lot of the times it can be better to add fewer customers but ones that are more loyal and spend more.
In order to calculate customer lifetime value in Segment’s model, I’ve added some additional variables:
Growth Simulation
Following Segment’s previous example, what we’re going to do is keep 5 forces constant and change one. We’re going to apply the same aggressive growth rate to partners- to-customers cross side growth.
However, this time we’re going to look at the $ value of our customers (customer lifetime value). To get this, we multiply customers from the respective growth channels to their respective average customer lifetime value.
Then we can look at projected growth case vs. base case across the entire projection period.
After running the simulation all 6 times, we can then determine which growth channel maximizes total customer lifetime value, and invest marketing resources there to maximize ROI.
Playing with the Numbers
Hope you enjoyed this article! If you have any questions feel free to reach out.
I’ve added the updated model Google Sheet here. Try playing with the numbers and see what happens to the growth case if you change:
- Recurring monthly revenue
- Retention rate