Why do some start ups grow virally in the Web, and others remain stuck on the ground like a dead horse? This article explains the phenomenon of network attractiveness and how it can be triggered.
Isn't it surprising that companies like SalesForce, dropbox or Trello take off like a rocket, and others fail dreadfully? Is it all due to coincidence and bad luck?
In fact no! Off course coincidence and good timing play a role. But le us assume that your timing is good and that your product has a good enough value proposition and a big enough potential user group,
The good news is: there are major adjusting screws. The art is to properly handle them.
In my role as researcher at the Karlsruhe Institute of Technology, I analyzed 100s of companies to find, categorize and model the mechansims which made some ignite, and others not.
Infact, it is not classical inbound marketing which made the difference!
Many of us use dropbox or Trello, but I suppose you never were spammed by these companies, or were heavily confronted with Google Ads praising these companies.
So what is their recipe of success if not classical inbound marketing?
The Concept of Network Attractiveness
All of the successful companies we looked at had a dimension of attractiveness, which is depending on the amount of users within the network. A typical analogy is a telephone network. If there are only few users, it remains pretty unattractive. But from a certain critical mass onward, the attractiveness grows more and more. We call this dimension of attractiveness network attractiveness.
What are the parameters of network attractiveness?
Let us now consider network attractiveness as our adjustment screw to tweak our chances of success, To do so, we need to have a closer look at it, to look inside it.
Amount of users - First, it depends on the amount of actual users of a service.
Critical Mass - The second parameter is the required critical mass of users. You may imagine that in the case of a telephone network, it needs to be pretty high, but I will talk about cases later-on, where a critical mass of two users is already sufficient.
Sensitivity - A third parameter which is important is sensitivity of the targeted user group to the size of the user group. You may imagine that there are services, like for instance the supply of piano lessons over the Internet, where the quantity of users is per se not so important. But the telephone case is the other end of the range. You need many people to make it attractive.
Amplification - Lastly, the users' sensitivity may be amplified. Imagine a phone company who offers e.g., 10 hours of free communication for every new customer you bring in.
The Interplay goes viral
Looking at the interplay of parameters, network attractiveness can be defined as the exponential function:
`sf "Network Attractiveness Ai" = Exp ( sf "amplification" xx sf "sensitivity" xx sf "amount of users" / sf "critical mass of users")`
An initial formula was first introduced by MIT Sloan Professor for System Dynamics John Sterman and we applied it on the Cloud Services and added the amplification. In my research, I developed amplification factors for it and contextualized it to the world of web-based platforms and IoT platforms. The formula actually says that if you set the parameters right, growth can really turn exponential, until market saturation is achieved.
We played around with several examples and simulated their growth against competition. The graph below shows the so-called logistic function which buffers down the exponential behavior in function of saturation. But forget the mathematics behind. All you need to understand is the relevant factors and the potential growth behavior until saturation.
What we describe here is that magic thing behind the so-called viral growth effects. we rather call it network effect. Such a network effect takes place within a causal loop, where network attractiveness grows exponentially.
In the following we have a look at some examples of network attractiveness and then try to find out how to apply and tweak them within a Cloud-based service offering.
Examples of Network Effects through Network Attractiveness
Amazon is one of the great examples of striking network effects. Starting as an online book and CD shop, they rapidly reach a huge mass of customers. This mass of customers was big enough to exceed the critical size, needed to attract external suppliers. So in the moment where Amazon opened its shop to external providers, rapidly thousands signed up.
Amazon had network attractiveness to them. And now look what happened: the more external suppliers came on board, the more customers joined in, inciting more suppliers to come on board. We call that a cross sided network effect. It can go on until saturation is reached (either because the market is saturated by you or shared by you and a set of competitors).
But now the downside: lots of start ups came up with portals trying to copy the Amazon idea in specific areas and domains - and most of them failed.
What did they do wrong? In most cases they started with a platform without users and suppliers.
They neither had network attractiveness to suppliers nor to customers - as they simply lacked the required critical mass.
The important bit as what we call the base value proposition. This is an intrinsic value proposition which the platform has in the first place, and which attracts an initial critical mass of either providers or customers. In Amazon's case the customers were there as Amazon had its base value (books, CDs and other goods). Once the engine has started off, it is pretty inert and runs and runs.
Salesforce had a similar lucky starting postion. They started off with a pretty successful and well managed (and well priced) CRM-as-a-Service solution. Lots of customers were attracted and provided a critical mass for suppliers to provide additional SaaS solutions in Salesforce's market place. The traditional market dominators where challenged by this disruptive solution.
We could state many more examples like that: Apple with the App-store is perhaps the best known in the smartphone sector.
But: not everybody has that big luck of starting off with thousands of existing users. This leads to the big challenge what to do, when there is no critical mass to build on.
The Big Challenge: the Initial Base Value
If you head for network attractiveness - which requires a high critical mass of users, you need an initial base value proposition and run through the rocky road of building up the initial customer base.
However, being flexible is one of the big advantages in the Cloud business. Given you are a Bank as a Platform. You want to attract lots of interesting Fintechs to outperform existing banks. You might simply "buy" a critical mass by joining forces with a strong front-end who will benefit from your services. You might serve as white label bank to a big grocery store chain and give them the chance to have their own self-branded bank, offered to all their customers within their locations and/or from home.
Sometimes it also makes sense to invest into startups, which have managed to develop a sufficiently important user base, but lack the business orientation or capital to materialize on it. Looking at the Vodaphone approach of "share-swapping", this does not even require huge amounts of capital to purchase such a company.
AirBnB started the other way round. They simply shortcut the way to have a sufficient base value by putting their site as front-end in front of an existing but "boring" online marketplace and homestay network for travelling. So they were attractive for consumers right from the beginning, giving them rapidly the critical mass to attract suppliers.
You might look at many other successful examples like Facebook. You always will find reasons and ways they generated a critical mass with an intrinsic base value proposition to start off.
But couldn't we reduce the critical mass to a minimum, let's say a critical mass of TWO ?
Yes we can. Let us now see how that works ...
Engineering Network Attractiveness with a Critical Mass of Two
Learning from Dropbox and Trello
The big challenge related to most business models is that they build on the assumption that the initial critical mass of users comes by magic. Unfortunately, recent history showed it does not.
So why not designing business models in the first place that do not require huge populations?
Dropbox and Trello are archetype examples of how to grow virally with a critical mass of two. The secret key to this are the so-called collaborative scenarios.
Let us look at Dropbox first. They started off as a file hosting service. This was a good base value for the users, as the included delta coding allowed for lean and error free upload and synchronisation across several client computers operated by the same user. But growth was pretty inert. There was nothing giving it any network dynamics or network attractiveness.
Collaborative Scenarios as Driving Force for Network Attractiveness
That changed when Dropbox introduced the collaborative scenarios. When the initial file hosting service became a file sharing service, the dynamic network effects got started. Suddenly, the users were driving Dropbox' growth: Each user was able to invite collaborators to co-work on files. Dropbox included an easy to handle button where the user could invite his colleagues. These were able to join free of charge limited on a certain storage size (freemium approach).
The inviting user was highly motivated to get collaborators on board. Different to a telecommunication network, he only needs one or two to provide value.
Amplifying the Sensitivity
But Dropbox added some motivation on top. For each new signed up Dropbox client, the inviting user got free storage.
The Viral Effect - Some Simple Maths
Just imagine each Dropbox user invites 3 new users, and the 3 new users again 3 more each. The viral loop is started - and Dropbox's growth figures reflect this viral growth in an impressive way. Having 1 had million registered users in April 2010, 2 million where reached in September and 3 million in November of the same year. It hit the 50 million users by October 2011, 100 million in November 2012, 200 million in November 2013, 400 million in June 2015, growing to 500 million in March 2016. Although we can see some attenuation (comparable with the simulated logistic curve above), the company keeps on growing.
Let us look at the formula for network attractiveness again. It has been defined as
`sf "Ai" = Exp ( sf "amplification" xx sf "sensitivity" xx sf "amount of users" / sf "critical mass of users")`
Dropbox managed to bring the critical mass down to the figure 2, as two users where enough to provide added value. With the amount of users growing steadily, the network attractiveness continued growing as well. But Dropbox also played on the sensitivity. The fact of offering added value to a user, when he invited others worked like an amplifiier to sensitivity. On top of gaining value through easy collaborative scenarios on files, there is a remuneration for inviting users.
Collaborative Scenarios worked with other Startups as well
Trello implemented pretty much the same concept. Collaborative scenarios motivated the users to invite friends. Having invited a certain amount of friends qualified the user for some free period of the professional version of Trello. To me this amplifier design is a bit weaker than in the Dropbox case, because you will need to pay after the testing period. The concept might be tuned, e.g. by increasing the value proposition gradually to those who have invited others. But still - the 4 year old startup was acquired for 425 Million USD beginning of 2017 and having a "user capital" of more than 14 million signed up people. What a success.
Why We Do Not Need Maths?
The good news is: There is no need to calculate values like critical mass or sensitivity. The market is too complex to create predictive models. Even the simulations we did in the past with our research team and those done by John Sterman at MIT were rather explanatory models, explaining a success or failure of a company based on historic data in a retrospective consideration.
So how shall we proceed?
Modeling Network Attractiveness Around Amplified Collaborative Scenarios
The advantage of Cloud-based applications is that we can base our reasoning on real-time data. This allows us to work in an iterative approach starting with a minimum viable product and optimizing it continuously. However we need to make sure that the initial design is viable. We need to provide a clear value proposition where we can be sure that there is a sufficiently large potential user group, that we are able to reach them and that these users are able or willing to join. You might test this through lean management based verification.
Cloud-based platforms and services may heavily benefit from Network Attractiveness and eventually grow virally. The biggest success stories in the Internet, like Amazon, Facebook, Dropbox or Trello benefited from that.
But you need to make sure that you set the basic parameters right. Make sure that your basic value proposition (base value) is properly shaped and visible. If you require a critical mass of users (either on the supplier or on the consumer side), make sure that your base value is able to build up that critical amount of users.
The easier approach is to design collaborative scenarios, where the users benefit even if there are only small amounts of users. Use the opportunity to increase the sensitivity through some intelligent amplification.
And stay tuned to the user behavior through continuous tracking of live data. Rapidly tweak and fine-tune your Cloud-based offering, based on live data analysis and suitable tactical enhancements and tests.
References and Further Reading
- Read more on Lean Startups driven by online analytics and reasoning.
- Read more on FinTechs
- Read more on general strategies on sales automation
- Academic research and publications on the network effects and network attractiveness around Cloud Platforms by Ulrich Scholten
- Image "Dropbox" courtesy of Ian Lamont, Flickr (License: CC by 2.0)
- Image "Amazon on Mars" courtesy of Actualitté, Flickr (License: CC by-SA 2.0)