A Playbook for Building a B2B Network Business
For both LiveRamp and Datavant, a key factor in our success was internally viewing the businesses as a “network” from day 1, and methodically following a network playbook.
From VISA to telecom companies to cable companies, many of the largest companies ever built are networks. These businesses have literal “network effects,” because their value to customers increases as their network expands. VISA is incredibly convenient for consumers because it is widely adopted by merchants, and it is critical for merchants because so many consumers have a VISA card. Network density becomes the killer feature for network businesses.
The playbook for building a network business is subtly different from businesses with other strategies. “Escape velocity” and “minimum viable network” become core concepts in company building, as getting to critical mass as quickly as possible becomes existential. Most B2B companies are not network businesses, so many of the playbooks and pattern matching by investors and experienced executives don’t perfectly apply to network businesses.
Below are some of the approaches I’ve found helpful in building out network businesses.
Tactic 1: Have Clear Network KPIs — nodes & edges
A simple step of building a network business is to rally the company around network metrics. Without explicit network KPIs, revenue and other metrics will domniate as the focus of a company, which are generally imperfectly correlated with network expansion. A lack of metrics also leads to lots of internal arguments over whether strategy A or B is right for the network, without a clear optimization function for resolving debates.
At Datavant, in the early days, we had two metrics that we tracked relentlessly (as closely as we tracked revenue): nodes of the network (what we called “installed sites”), and edges in the network (what we called “links”):
- A node was a company that was using Datavant technology; once they used us once, it was very easy for others in the network to do business with that company in the future.
- Edges were unique relationships between different parties in the network; if company A sourced data from company B and company C, that would be 2 edges. This was a rough metric of network throughput.
Establishing these metrics and prioritizing them relentlessly made many other aspects of our strategy clear, as you’ll see below.
Tactic 2: Look for subnetworks to get density
Generally, what we think of as a “network” actually consists of lots of subnetworks, each of which have their own internal network effects and where it is easier to get to critical mass network density.
For example, Facebook was built by being a social network at Harvard, then a social network at Ivy League schools, then a social network at all colleges, then a social network for everyone. This meant that at the early colleges, Facebook had critical mass density, well before it had critical mass everywhere. Had they tried to launch everywhere initially, even if they had the same initial number of users, they would not have had critical mass density anywhere — and would have failed to get traction.
For Datavant, we found that the ecosystem around commercial claims data was a subtly different ecosystem than the ecosystem around oncology data, which was subtly different than the ecosystem around public sector, and subtly different than clinical trials; all of these networks needed to be connected in the long-term, but we could get to “critical mass density” more quickly by focusing on specific areas. So we methodically focused on different submarkets where critical mass density was feasible.
Tactic 3: Go narrow on features.
Common advice is to go after a product with a large total addressable market; I think that this is perhaps the worst advice that VCs frequently give to entrepreneurs. In reality, while it is good to build sequentially into a larger and larger TAM, you want to go after the smallest viable TAM to get there: narrow products are much easier to crack a network effect. You want to go narrow on two dimensions:
- Size of network. In the Facebook example, it’s easier to build “the social network for Harvard students” than the “social network for everybody” precisely because it’s a smaller market.
- Size of feature set. You ideally want to have the minimum viable number of features in the early days. In a true network business, each individual feature has its own chicken-and-egg network effect (eg. “phone calls” and “texting” both require critical mass adoption, even though they’re on the same network), so you want to avoid fighting for critical mass adoption on multiple fronts simultaneously. Additionally, because ubiquity is so important, you’ll want large incumbents to rely on you as a channel partner (more on that below), and so reducing your feature set decreases your level of competitive threat — letting you more quickly get network scale.
Tactic 4: Cultivate network-referred sales
In Datavant’s early days, one of the most amazing features of the business was that ~90% of new business was customer-referred — meaning our best marketing channel (by far) was existing customers. This is true in a lot of network businesses: your power customers get value from your network expanding, and so — if you align closely with them — they refer lots of additional customers your way.
To make this work, obviously it was important that our product added enough value for us to be referred frequently, but it was also critical for us to make the sales reps at the partner want to bring us into conversations. This involved us obsessing with being perceived as quick & easy to work with, willingness to not take too much of the pie, and having sales reps who could deftly handle three-way conversations.
Tactic 5: Look for supernodes
When most companies think about their wishlist of perfect customers, they think about who can drive the most revenue potential.
In a network business, the perfect customer instead maximizes the network throughput (which is imperfectly correlated with revenue).
For Datavant, a counter-intuitive part of our initial strategy was focusing our sales on the ecosystem of analytics companies servicing pharma, instead of servicing pharma companies directly. Almost every VC we spoke with had the view “all the money is in pharma; you should focus on them;” all of our competitors went in this direction. Instead, our view was that analytics companies are power users of data and drive more network activity than pharma, and have faster sales cycles — so it’s the faster path to network ubiquity.
We especially doubled down on the handful of customers that became “supernodes;” at our peak, our 5 biggest customers were driving ~25% of our network activity. Even if they weren’t giant from a revenue perspective, this drove our network to take off.
Tactic 6: Become channel agnostic
If the primary goal is to get as much network throughput as quickly as possible, a useful accelerant is channel partnerships. The main downside of channel partnerships (assuming you can successfully set them up — they are tricky to operationalize) is that you lose the relationship with the end customer, and ultimately capture less value per customer. In a network business, these are generally very reasonable tradeoffs for accelerating your footprint more quickly.
At LiveRamp, when we were $30 mm ARR, ~80% of our revenue came from 5 channel partners; we only started going direct when we were much larger. While this was an unusual path, channel relationships shaved several years off of our path to becoming perceived as the universal standard in the industry and getting to escape velocity.
Tactic 7: Short-term success = Long-term success
In every business, the CEO has a complex job of balancing short-term goals (hit the number this quarter) with long-term goals (eg. build the product successfully).
In a network business, I think short-term goals and long-term goals are extremely correlated: your long-term success is determined more than anything else by hitting network escape velocity. Being successful in the short-term is the most important determinant of long-term success because it unlocks more network density, which in turn makes your product more compelling. Short-term momentum also creates a sense of inevitability in the market and swagger within your team, which becomes self-fulfilling.
So I generally am “short-termist” in goals, looking for accelerants for how we maximize our success this quarter/month/week — not because I don’t care about the long-term, but precisely because I believe short-term success is the biggest driver of long-term success, and escape velocity is everything.
Tactic 8: Make it a no-brainer to work with you.
As you establish network ubiquity, customers are making bets on whether they want to be in your network — and you want the answer to be an obvious yes. This means you need to be relentless about being low-friction to work with, and never lose deals on price or extraneous contract terms. Remember, nodes are valuable — even if they don’t pay you anything. That doesn’t mean you should make your product free by default (that rarely works as a strategy to increase adoption), but it does mean you should have a mechanism to discount aggressively when needed, and make the standard on the sales team to never lose a deal to competition on price.
Similarly, you want to be seen as easy to work with — quick to turnaround contracts, great support, and a general culture of excellence + velocity. Make the choice to partner with you as much of a no-brainer as possible.
Tactic 9: Look for step functions on network expansion
The vast majority of growth for network businesses should come from a relentless grind for more and more nodes and edges one at a time. Beyond pushing this grind forward, part of the CEO’s job is also to find the step functions.
Some step functions are more complex partnerships — setting up the right channel partnerships, or finding the incumbents who can help “king-make” your network. In the early days of VISA, they spent several years obsessing over their relationships with banks as their most critical partners.
Network businesses also lend themselves to more early-stage M&A than other business models. The most straightforward M&A hypotheses are usually finding complimentary subnetworks, and are along the lines of “we are the leading railroad of the east; you are the leading railroad of the west; we can combine and become the leading railroad in the country.” When that’s actually true and expands the network dramatically, 1+1 can equal 5 in combining the networks.
Tactic 10: Make ubiquity part of the brand
The killer feature of most network businesses is the fact that they are ubiquitous; phones are great because everyone has a phone, and VISA cards are great because they’re accepted everywhere. All the other features are secondary.
As you build your company brand, it’s helpful to make ubiquity an explicit part of the brand. Just like VISA is “everywhere you want to be,” you want to be perceived as ubiquitous. This can be done with how you describe your company, the sense of momentum that you create for your company, the conferences you host, the advisors you engage, etc. — but you want to ensure that there is a sense of inevitability around you.
Before you have established a massive network, your customers are making decisions to bet on you; if they believe that you will be ubiquitous, that’s an easier bet to make. So building the sense of inevitability becomes self-fulfilling.
Tactic 11: Look for ways to eliminate network effects in the early days
Once you’ve established a minimum viable network, network effects are an amazing feature of a business model by creating a strong moat and driving customer-referred sales.
Until you reach that point, network effects can work against you. Why would anybody join a network that doesn’t have critical mass?
One hack around this is to look for “single player mode” approaches. In this analogy, multi-player video games are games that have network effects; they require enough other players to want to play simultaneously. An effective way for games to get to that point is to have a great “single-player mode” — where people play against the computer directly, so the product is valuable whether or not there’s a network already playing.
In other words, if it’s possible to eliminate the network effects to start, that can be an effective way to bypass the chicken-and-egg problem until you have a critical mass of customers. At Datavant, the use cases we serviced initially were around connecting internal data sets (which did not require a network), before we had the critical mass to also connect external data sets.
***
These tactics all exist as a means to reach network escape velocity. For a network business, network density and throughput is everything, and escape velocity is the moment where the density goes from holding you back until the network effect begins to work in your favor.
