Practical Advice

Network effects: ClearScore & Salary Finance co-founder Dan Cobley on building the ultimate startup defence strategy

Practical Advice

Network effects: ClearScore & Salary Finance co-founder Dan Cobley on building the ultimate startup defence strategy

Words Dan Cobley

July 26th 2022 / 8 min read

This article is excerpted from Dan Cobley’s talk on Network Effects at our 2022 Founder Day.

Defensibility—how can you build the ultimate competitive advantage for your startup? Broadly, there are four ways to build defensibility.

Firstly, scale: how vast and far-reaching are your operations? Amazon has the upper hand on nearly every e-commerce provider on account of its unparalleled scale and the prices they’re able to offer as a result. Then there’s brand: how recognisable is your brand, and is it synonymous with quality? My kids no longer talk about buying something: they talk about ‘Klarna-ing’ or ‘Monzo-ing’ it. Thirdly, there’s embedding: is your product or service embedded in such a way that makes it hard to replace? The way that GoCardless, for instance, is embedded in checkouts. 

But most effective of all is network effects. Network effects are found in 20% of startups, but are accountable for 70% of the total value created in tech since 1994. If you’re looking to build a moat around your startup, network effects are really your most valuable tool. 

Key takeaways

  • Understand that network effects, crucially, are different from viral effects—and are a far more effective tool for building sustainable value in a business 

  • Learn about the different methods for building network effects, and how successful startups like Uber, Facebook, and OpenTable have deployed them

  • What the ‘cold start problem’ is for building network effects, as well as how to overcome some of the other challenges you’ll face as you scale a network

Interested in receiving more insights like this direct to your inbox? Subscribe to our newsletter, The Startup Bulletin.
Subscribe here

What are network effects?

My preferred definition of network effects is: 

A network effect occurs when a product or service becomes more valuable to its users when more people use it

\It’s important to distinguish this from viral effects, which is where a product or service encourages users to recruit other users in order to grow the product. Take Wordle, the game that took the world by storm in early 2022. Many of us, myself included, started playing as a result of friends’ recommendations. This is somewhat of an in-built growth hack, deployed by many startups as a way of acquiring customers (often incentivized by referral benefits—e.g. HelloFresh or other DTC brands). 

Crucially, the experience of the product doesn’t improve with the more people that join it (aside from some bragging rights). The result? There’s little defensibility to the product, evident in the inevitable slump in Wordle’s daily users.

Products with network effects have both this in-built growth hack and a core defensibility, from the fact that the experience improves with each new addition to the network. Those who used Skype back in the day will understand that, with each new user, the experience improves as there are more people you can speak to on Skype.

Different ways of creating network effects

There are several different ways for you to build network effects into your business: 

📞 Physical infrastructure. One of the early examples of this was the telephone. This is a simple, physical, very direct network, in which users are all homogeneous (carry the same value). For the very first person with a telephone, it had zero value; for the second user it had slightly more, but still limited value. Only after there were lots of users did the network have real value, growing with each additional user. 

🌐 Data networks. This is one of the most popular network effects that businesses can build into what they’re doing. The idea is that more users > more data > a better experience > more users. Navigation platform Waze taps into this: the more drivers who use Waze increases the data volume they have, thus creating a better/more accurate experience for drivers. 

📱Platforms. This is an extension of infrastructure—specifically the platforms that are built on top of this infrastructure. PlayStation, for instance, is a two-sided platform for both players and developers. Here the network effects look like: more users > more games developers > more games > more users. 

🤝 Marketplaces. Here, network effects correlate to your breadth of choice. OpenTable is a two-sided marketplace for consumers (literally) and restaurants: the growth of one feeds the growth of the other.

I used to work at Google, who own YouTube, a more complex three-sided marketplace for users, content creators, and advertisers. Each of these groups were a central consideration for us when trying to drive new content categories: we’d analyse which of these three groups wasn’t playing and would try to incentivise growth. For instance, when we tried to boost more “How To” videos, we saw that the advertisers weren’t really biting. So we went to the creators, guaranteed their revenue in order to increase the quality and quantity of their content, which would in turn pull in advertising revenue. 

Challenges associated with creating network effects

Building a business around a potential network effect isn’t a guaranteed formula for success. There are a number of challenges that founders need to be conscious of overcoming:

The cold start problem

This is the first, and arguably biggest, challenge you’ll face—a classic chicken-egg conundrum. If the value of the business is in its network, then how do you get those first initial customers on board before the network exists? 

There are a couple of ways to address this. First of all, you should try to enable a ‘single player mode’—some utility to those users who are coming in before there are any network connections. Adobe, as a way of trying to entice users to their full Adobe suite, has Adobe Acrobat as a free single player mode, allowing users to read Adobe (PDF) documents before paying for their software to collaboratively create and edit them. At Clear Score, we needed to onboard vast swathes of customers before some large credit card providers would partner with us: so we offered free credit reports in a very consumable and interesting way, along with tips on how to improve your credit score. 

Hyper-local launches are another way of addressing this. Rather than exploding into a large market at once, it can be more effective to address a much smaller market. Uber doesn’t launch into France all at the same time: they launch into central Paris or Lyon, and get a density going in that space. 

You may also have to be prepared to subsidise one side of the market. How many people will use Uber when there are no drivers; and similarly, how many drivers will register if there are no fares? Therefore in each market, Uber recruits a bunch of drivers and pays them a minimum fee for idling: that way, they have a bank of drivers ready to cater to the first customers. 

Finally, you should think about services you can offer on top of the network. Before launching their marketplace, OpenTable went into every restaurant with an iPad booking system to replace the primitive pen and paper method: once a number of restaurants were using this, they added them to the marketplace to entice customers. 

There are also a number of challenges associated with scaling your network. To build true defensibility you should consider how your business can overcome the following:

Network congestion

Networks are great; big networks are even greater. But can a network get too big? In some cases, yes. Take Strava. Once upon a time, it was myself and my ten best friends sharing our cycling activity with each other. But now, I have hundreds of contacts on there, and I barely look at the feed. Sometimes, too much activity can reduce the value and excitement of the product. 


What stops an Uber driver from registering for Lyft, Bolt, etc? What stops an eBay vendor from using Facebook Marketplace, Depop, and so on? There are no downsides to using both, and therefore I’ll likely use whichever pops up first. To avoid this, platforms have to really offer something unique or impactful to their network as a way of keeping them engaged, or make it impractical to run two systems.  


There’s also a risk of a platform becoming redundant. You might witness this on a more basic level with babysitters or plumbers: as soon as you’ve made contact with the individual, there’s no advantage to using the platform anymore (especially if they take a cut). So consider this: what can you do to avoid being cut out?

Asymptotic effects

Essentially, this is a point you reach where the value-added by an additional network participant adds no incremental value to others.. If there is one Uber driver in my town, adding another arguably doubles the value and halves my waiting time. But when there are 1000 drivers, and I never have to wait more than 2 minutes for an Uber, adding another driver doesn’t add any perceivable value. This is a specific problem for networks where one side is fungible (every driver performs the same function), and not for a marketplace like OpenTable where every option is differentiated. 

At Clear Score, we’ve faced this challenge: once you’ve onboarded most high street banks, any addition has minimal added value. So we’ve deployed two strategies:

  • Actively sought out underserved markets— with regards to our market, this means working with providers who are reaching those with poor credit scores, or those who wouldn’t consider applying for a credit card

  • Increasing our data leverage— this is behind our latest product, Drive Score, where we’re using black box data created from the DriveScore app to convince insurers you are a good driver (and get better rates) without an actual blackbox

Network effects come in many flavours, and when properly established, they can add huge utility to your customers and defensibility to your business. It is worth consciously exploring how you can maximise their impact and it will likely pay back many times over in terms of the value you create.

If you’re interested in reading more about network effects in startups, then check out the NFX ‘Network Effects Bible’

About Dan Cobley

Dan Cobley is Managing Partner for Fintech at Blenheim Chalcot, a leading UK venture builder. He is the co-founder of ClearScore and SalaryFinance. Before that, he spent eight years working at Google as MD of UK and Ireland and VP Marketing EMEA.

Enjoyed reading this article?

We’ve abridged it in a short Twitter thread, which you share with your friends & followers here

Words by
Share article

Advice from world-class operators and founders

factory news

Investing in AtmoCooling—atmospheric control for global deserts

Learn more about our latest Blue Action Accelerator investment, and how they’re building dynamic technology to address the heat and aridity of the hottest deserts

founder stories

TrustedHousesitters’ Mathew Prior on growing a category-creating pet sitting marketplace

Learn about the story behind TrustedHousesitters, the marketplace for pet owners and sitters, recently acquired at $100m valuation

factory news

Investing in CyferAll—data-centric cybersecurity for the post-quantum era

Learn more about our latest Fastweb investment, and how they’re building an integrated solution against the most sophisticated classical and quantum level threats to data