Network Effects

What are network effects?

Network effects arise when the benefit of a product or platform to each individual user rises the more other users use it. What matters is the higher benefit for the existing user, not a lower cost base or greater brand awareness of the provider. The simplest picture is the telephone: a single connection is worthless, and with every further subscriber every connection becomes more useful. Economically, this is a demand-side effect, because each additional user changes the value of the product for the others; economics also speaks of network externalities.

Network effects are typical of platforms, marketplaces, communication and collaboration software and payment networks. Founders highlight them in the equity story of a funding round, because they can justify a competitive advantage that grows with size. Whether the claim holds is clarified in due diligence on the basis of usage and cohort data. German competition law also uses the term. Under Section 18 (3a) of the German Act against Restraints of Competition (Gesetz gegen Wettbewerbsbeschränkungen, GWB), direct and indirect network effects must be taken into account in multi-sided markets and networks when the market position of a company is assessed.

Types of network effects with examples

• Direct network effects: The benefit rises with the number of users in the same group. A communication service is more valuable the more people to talk to use it, and software for exchange between companies the more business partners are connected.
• Indirect network effects: Two or more groups benefit from each other reciprocally. More providers make a marketplace more attractive to buyers, and more buyers attract further providers.
• Local network effects: The effect operates only within a segment, such as a city, an industry or a language area. An additional trade business in Hamburg is of no use to a client in Munich, and every new region starts without a network advantage.
• Data-based effects: More use improves the quality of the product for everyone through collected data, for example in recognition or recommendation systems. Whether a genuine network effect exists must be examined case by case, because the additional benefit of further data often diminishes.

Network effects can also work negatively. Too many providers reduce the visibility of the individual offer and intensify competition among them; too many participants lead to congestion and falling quality. Platforms counter this through selection of providers, rating systems and participation rules.

Critical mass and the chicken-and-egg problem

Network effects take hold only from a critical mass. Before that, the product is worth little to the individual user, because the other side or the other users are missing. On two-sided platforms, the so-called chicken-and-egg problem arises: providers come only if buyers are there, and vice versa. Widespread solutions are:

• Limiting the market: The platform starts in a region or industry where sufficient density can be reached quickly.
• Subsidising one side: One side receives the service at a discount or free of charge, or the platform initially provides part of the supply itself.
• Benefit without a network: The product already offers the individual user a benefit on its own, for example as administration software, and the network arises as an additional layer.

For marketplaces, liquidity is the most important metric. This does not mean the ability to pay but the probability that a request or an offer leads to a transaction within a reasonable time. If it rises with the number of participants on the other side, this is a first piece of evidence of an indirect network effect.

How investors assess network effects

Network effects are among the claims that investors examine especially critically, because they can carry a considerable part of the valuation. The test is whether the product becomes better for an existing customer when another joins; if not, there is growth but no network effect. The effect is evidenced through data:

• Cohorts: Users who joined at higher network density use the product more and churn less often than earlier cohorts or users in thinner regions.
• Completion rate: The share of successful matches rises with the number of participants on the respective other side.
• Acquisition costs: As the network grows, the share of customers who arrive without paid advertising rises, and the cost per acquired customer falls.
• Pricing: The platform can hold or raise its commission rate without losing participants.

Against a strong effect speak the parallel use of several platforms (multi-homing) and low switching costs. A further warning sign is when providers and buyers, after the first contact, conclude deals directly with each other, bypassing the platform. The GWB expressly names parallel use and switching effort as criteria of market position. Mere size is not evidence. A high market share, many users or a well-known brand can equally rest on sales spending, price discounts or early market entry.

In valuation, evidenced network effects work through the assumptions of the financial model: lower churn, falling acquisition costs and stable or rising commission rates. Through these assumptions they justify higher revenue multiples (revenue multiple), not through a flat premium on enterprise value.

Network effects in a company purchase

In acquisitions of platform companies, merger control gains importance, that is, the review of corporate combinations by the Federal Cartel Office (Bundeskartellamt). Besides the revenue-based thresholds, the GWB has a threshold that is tied to the value of the consideration, put simply to the purchase price (Section 35 (1a) GWB). It targets acquisitions of companies that are highly valued, for example because of their network or their data, but still generate little revenue in Germany. A precondition is, among other things, that the acquired company is active to a significant extent in Germany. The thresholds in the current version apply. An acquisition that must be notified may not be completed until clearance. Clearance is therefore agreed as a closing condition, and the parties' law firms clarify the notification obligation before signing.

Worked example: measuring network effects on a marketplace

A marketplace brokers maintenance orders between industrial companies and service providers in two regions. In the North region, 120 service providers are active, in the South region 360. In the last quarter, the North received 2,000 requests, from which 900 awarded orders resulted; the South received 2,400 requests and 1,680 orders.

• Completion rate North: 900 awarded orders ÷ 2,000 requests = 45.0%
• Completion rate South: 1,680 awarded orders ÷ 2,400 requests = 70.0%
• Orders per service provider: North 900 ÷ 120 = 7.5; South 1,680 ÷ 360 = 4.7
• Share of clients still placing orders after twelve months: North 38%, South 61%

The higher density of service providers in the South improves the clients' completion rate by 25 percentage points and goes together with retention that is 23 percentage points higher. This points to an indirect network effect, provided the regions do not differ in other characteristics such as industry structure or price level. The effect, however, is local: the 360 service providers in the South do not improve the situation of clients in the North. At the same time, each service provider in the South receives fewer orders than in the North; if this number falls further, service providers lose interest, and the effect reverses on the provider side. Investors therefore value such a platform region by region, rather than extrapolating the figures from the South to the overall market.

Distinction: network effects, economies of scale, viral loop, lock-in

• Economies of scale: reduce the cost per unit as volume grows and work on the supply side. Network effects increase the benefit to the customer and work on the demand side.
• Viral loop: a customer acquisition mechanism in which users invite further users. Virality lowers acquisition costs, a network effect increases benefit; the two often occur together but must be evidenced separately.
• Lock-in effect: binds customers through switching costs such as data migration, training or contract terms and can exist without any network effect.
• Scalability: describes whether costs grow less than proportionally to revenue. Network effects can reinforce it because they lower acquisition costs, but they are not a prerequisite for it.

Common mistakes

• Network effect without evidence. A claim without cohort or liquidity data is not accepted in due diligence and weakens the credibility of the other statements.
• Confusing growth with a network effect. Rising user numbers alone do not show that the benefit to the individual user is rising.
• Calculating local effects globally. If the liquidity of a dense region is transferred to the overall market, market share and the pace of expansion are overestimated.
• Overlooking circumvention of the platform. If participants conclude follow-up orders directly with each other, the benefit of the network arises without the platform earning from it.

Frequently asked questions

How can a network effect be proven?

Through cohort and liquidity data. Users who joined at higher network density must measurably use the product more and churn less often, and the completion rate must rise with the number of participants. The usual tool is cohort analysis.

Is a network effect the same as a lasting competitive advantage?

No. A lasting competitive advantage (moat) can rest on network effects, but also on switching costs, intellectual property rights, cost advantages or a strong brand. Network effects are considered particularly durable if they are not limited locally and parallel use is low.

Do software companies for business customers need network effects?

No. Many valuable providers for business customers have no significant network effects but rest on high switching costs, deep integration into processes and good customer retention. A model presented openly without a network effect is more robust in investor communication than a claimed effect that the data does not support.

Does the value of a network grow with the square of the number of users?

No, only the number of possible connections grows roughly that way. With n participants there are n × (n minus 1) ÷ 2 possible pairs, so 4,950 with 100 participants and 19,900 with 200 participants. In reality, most users interact with only a small share of the others, so the benefit grows much more slowly. This rule of thumb, known as Metcalfe's law, is therefore unsuitable for valuations.

The Roemer Capital view

Roemer Capital adopts a network effect in the pitch deck and equity story only when the client's data supports it: cohorts, completion rates and the development of acquisition costs. If it cannot be evidenced, Roemer Capital deletes the term and describes the model's actual strengths, such as switching costs or integration into customers' processes.

If the effect is only local, Roemer Capital builds the financial model by regions or segments, each with its own ramp-up phase up to critical mass. This presentation is less impressive than an extrapolation to the overall market, but it withstands investor scrutiny and makes the use of proceeds of the funding round comprehensible.