This article is a 9 step guide how to develop a business model for your App. Note, that we are looking at Social Media Apps and Search Engines as an example. Most of the steps are relevant for all platform businesses.
What most people think that App development is about, is only Step 7 below. It means there is a lot more to do first. It also means that what most people think App development is about makes only a part of the whole endeavour!
- Step 1: Understand the overall (advertising) market size
- Step 2: Develop a Customer Value Proposition
- Step 3: Develop a monetisable Customer Value Proposition
- Step 4: Attract participants to your App
- Step 5: Develop the inner workings: network effects (NWE)
- Step 6: Platform design considerations
- Step 7: Digital properties
- Step 8: Data capture & use of it for monetisation
- Step 9: Sub-markets and monetisation
Step 1: Understand the overall market size
Let’s start with looking at overall markets. There are a few dozen. In step 9 we will see that we need to go far deeper within a selected market. But this a starting point.
The macro view shows how Google’s and Facebook’s revenues compare to the GDP generated by entire industries in the US. This typically is the revenue of all players in that industry (with some exceptions, notably retail). I have chosen industries in a similar ballpark to Google+Facebook. There are some bigger ones which are not shown.
But how do they make these extraordinary revenues when most others struggle to monetise their (wonderful) idea?
The business model flywheel
Before we go into the details (and as I say, the details matter most), let’s have a look at the high level. We have three interwoven flywheels, one for each participant type of the platform (users, businesses, advertisers). Here is the advertiser flywheel.

Step 2: Develop a Customer Value Proposition
The core of every business is its value proposition. Traditional businesses typically have to provide a value proposition to one type of end-customer. Platform business models differ in that they have to deliver distinct value propositions to more than one type of participant of the multi-sided platform.
These sides (=participant types) are:
- Users (as in “ordinary” users)
- Businesses
- Advertisers
Benefits for small businesses
The value propositions following next are suitable for all sizes of business, small and large and no longer to large businesses only as was the case before:
“The reason digital advertising works so well for small businesses is that it’s a great leveller. You don’t need to tip thousands or millions of dollars in building a brand or creating assets that might get used for a few weeks on TV or radio. The way the system is designed is that anyone can compete” Mike Rhodes, Founder and CEO, WebSavvy
Search and Social grew large not just because of small/medium size businesses. But they have brought many such players to the table. There have been the heavy-hitters, like Booking.com, who were said to have spent billions per year on Google Ads. But there are also many others.
Millions of large and small customers
This statistic from 2011 shows that the top 50 Google advertisers generate <$2b of $37b revenue, i.e. ~5%. There were some huge advertisers in the following years, but it’s also clear that small advertisers are crucial for Google.
Facebook says that they have 7m advertisers on their pages (as of June, 2020)
Google was estimated to have 4m advertisers in 2015.
Now that we know that our platforms (a) make a lot of money and (b) have millions of paying customers, let’s have a look at how they have done this.
Step 3: Develop a monetisable Customer Value Proposition: Value proposition for advertisers
Many start-ups struggle with this: finding a value proposition that someone is willing to pay for (beyond free participation which is necessary to have users in the first place). So, let’s have a look at how social networks and search engines have solved this.
Supply of new targetable advertising spaces with auctioning mechanism
Think of traditional advertising. As an example, there are limited TV advertising spots for which the big brands bid prices up. These often come from verticals with a large market size (e.g. consumer goods). Many verticals are priced out of TV ads because their market is too small. Customer lifetime value in relation to acquisition costs plays a role but for small verticals, this is not enough due to insufficient volume and lack of targeting.
Keyword search requires ads to be relevant to the user’s search term, thereby favouring vertical-specific bidders.
We can make similar observations for the geographical dimension. Local companies are often priced out of many forms of advertising. Small/medium companies can advertise on local radio or newspaper. But even then, a lot of it is displayed to people who will not be interested in the respective product/service making it ineffective.
Search and Social platforms add economic value to these companies by allowing them to reach the right consumers. Search/Social create new ad spaces on the screens of users who they can target well.
There are many ways of ad targeting. We will use the vertical and geographic dimensions as our examples.
On the next diagram, you see that the average cost per click (CPC) can vary by a factor of 10 depending on the vertical. This means that the lower cost verticals would be priced out if they had to compete for the same ad space.
Equally, if you look at the average revenue per user (ARPU) across different geographies (Facebook), you see a factor of 20. Now, ARPU does not equate to CPC but there is definitely a strong correlation.
From an advertiser’s perspective, significant observations are that
- an unprecedented amount of new ad space supply has been created
- combined with a targeting mechanism based on the detailed properties of the ad viewer and with a pricing mechanism that is
- based on open market auctioning

Value proposition to advertisers
The fascinating truth is that you can basically rattle down a laundry list of microeconomic (others call them strategic) concepts that Google, Facebook, Twitter, Snapchat and Pinterest (and other successful social media and search engines) provide as additional value propositions to their advertisers.
- Barriers to entry to digital advertising are low.
- Barriers to exit are also insignificant
- The scalability of advertising
- Variable cost only (for measurable results)
- Timely returns & measurable metrics (even possibility of negative cash conversion cycle)
- Advertising tools (search, transaction cost saving)
Anyone with a microeonomic or strategic background will be awe-inspired by these compelling benefits for advertisers (and it’s not even the whole list). I am sorry to throw this at you like this, but rest assured I have a lot more explanation on each of these within the premium resources.
This is on top of the multitude of value propositions for businesses. At the end of the day, advertisers are a subset ob businesses (Facebook says that there are 160m+ businesses on their platform – hence the paying 7m advertisers make about 5% of the total business base).
Step 4: Attract participants to your App / Multi-sided platform (MSP)
Another significant difference are the platforms’ direct relationships to the customers. They interface directly with their ad customers without middlemen / gatekeepers (there are ad service firms but the platform’s relationship / accounts are always with the end customer).
Where in traditional industries would you see this? Often margins are eroded by powerful middlemen who interface with the customers.
Amazing similarities
It is amazing how similar search engines and social media are in their value propositions for advertisers. But it’s equally riveting that the same holds true for the other sides of the platforms. You will notice how the value proposition to users also follows important economic concepts, in this case from the area of search and transaction costs.
We can actually find a level that is not superficial where all these platforms are comparable (which is part of the detailed resources). The great value of doing this is that we can branch down from there into our own ideas. Many of you share with me their ideas. This is why I know that doing this is of enormous value.

With this, we have concluded the prelude.
Step 5: Develop the inner workings: network effects (NWE)
Network effects are often referred to as the most important competitive advantages of platform businesses.
Network effects (NWE) are the effects that incremental participants and participation have on the value of the network to other participants.
In the context of multi-sided platforms (=platform businesses), we distinguish between two different types of network effects:
- Direct network effects, also called same-side network effects, are the effects of participants on one side of the network on other participants on the same side of the network. An example on Social is the creation of relatable content among users (esp. close connections)
- Indirect network effects are effects of one side of the MSP on the other side of the MSP and are also called cross-side network effects. Content created by businesses is an example for cross-side network effect on Social and crucial on Search
Positive and negative network effects: Network effects can be positive or negative. Network effects on Social among close connections tend to be mostly positive. But they can also turn negative (bullying, harassment, etc).
Enhancing positive network effects and reducing negative ones is the most important activity of a platform business
Then there are wider negative network effects (and externalities) where e.g. the platform is misused to spread disinformation (more later). Negative network effects need to be managed by the platform. Facebook had 30,000 staff (2018, pdf) to manage the multitude of negative network effects of which they say:
“This work will never be finished, but I now believe we’ve built some of the most advanced systems in the world for dealing with these issues.”
So, if the value of the network grows with the number of participants, where can we see this?

Connections underpin NWE. But they are not NWE as such. It is often pointed out that network effects are based on Metcalf’s law (which says that the number of connections increases at the square with the number of connections, i.e. N^2 with N=number of participants).
I would not say that this is wrong but like to call out that you should not use it as a formula but rather as a guide. It is a correlation that was used in another context and needs to be overlaid with many other factors, firstly, with who is actually connected to whom.
Platforms, by no means, try to bluntly maximise the number of connections. That in itself will lead to confusion and overload, burying relevant content among irrelevant content. Platforms try to do quite the opposite, by aiming to (algorithmically) provide well-matched connections (which is why you see me use this phrase so often).
Take it, not all connections are equal.
Closer connections drive more engagement. Snap Inc CEO Evan Spiegel says: “Your top friend in a given week contributes 25% of Snap send volume. By the time you get to 18 friends, each incremental friend contributes less than 1%” Now, consider that half of Facebook users have more than 200 connections (average of 338).
The previous statement of CEO Evan Spiegel continues to say: “This means that in order to grow our business we need to make sure that we help all Snapchatters communicates with their best friends. Finding best friends is a different problem than finding more friends, so we need to think about new ways to help people find the friends they care most about. We can’t establish network effects if our users can’t use Snap to communicate – so we need to work hard to make sure that all Snapchatters have best friends they can communicate with.”

Step 6: Platform design considerations
There are many design factors for Search/Social platforms. These will determine the value proposition and affect network effects, search/transaction costs and other factors of the platform. It is these factors that will instrumentally determine the trajectory of the platform in the long term.
1. Connections types
One-way following does typically not require consent from the one being followed. There are different rules on who can be followed.
- On Twitter, everyone can be followed by default. On Facebook, this depends on privacy settings.
- Two-way following requires mutual consent. This is called “friends” on Facebook, on Snapchat everybody is a friend.
- On Snapchat, one-way and two-way connections are shown in different digital properties and Facebook is also slowly separating things.
- Further, different platforms follow different strategies in suggesting further connections. This is most prominently featured on Twitter.
- On Google, any indexed page can be accessed (website owners can tell Google not to crawl & index their pages).
There are 12 other social media / search engine platform design considerations (many of them transferable to other platforms), that I am showing. They can help your creativity. ….
Step 7: Digital properties
In the next step, we are looking at the *thing* well, the thing that the users see. I call them digital properties.
Digital properties are those powerful hubs (like the News Feed) where some of the most important elements of platform business models come together! Digital properties are key elements of an app or website, such as the News Feed, Stories, the Like button, the Knowledge Panel, etc.
Digital properties:
- Are instrumental in delivering the value proposition at low search / transaction costs
- Bring together defined sides of the multi-sided platform and create network effects
- Capture data
- Create ad spaces (i.e. are instrumental in monetisation)

We are going into more depth than I can cover here. But for starters, here are some overall principles of digital properties.
- Most platforms’ apps consist of several digital properties.
- They are instrumental in delivering the value proposition and do so by connecting different (typically, selected) sides of the MSP.
- The user interface of the DP should be designed to reduce search / transaction costs of key activities (searching for, consuming / creating content) and minimise risks of acting on content.
- Switching to other digital properties should be low friction. The most valuable DPs should be reachable most easily.
- Each digital property creates its own network effects across sides of the platform
- Digital properties capture valuable data in order to provide the value proposition
- This data can be used for (micro-)segmentation
- It will shape the segments targetable by advertisers
- Digital properties should support native ad formats
- Monetisation value can be benchmarked against other DPs and relative to the platform
- Opportunity cost can be measured by whether it created incremental or replacement engagement
We are looking at some examples in terms of digital properties and then providing a framework that you can use. We are also tailoring search and transaction costs which I had to skip here altogether.
Step 8: Data capture & use of it for monetisation
We have said that accurate ad targeting is one of the most important value propositions for advertisers. Essential for this is the capturing of user data. And it should not be an accidental by-product of the design of a platform. Rather we should design our platform and digital properties with data capture in mind.
There will be digital properties used for engagement, provision of the value proposition with little need for data capture. But there can also be digital properties that can provide a great value proposition when capturing data. In this case, the data can also be used to deliver well-targeted ads. The reality, of course, looks a bit different. Data is captured “because we can”.
Three important reasons (among others) to capture data are:
- Provision of a better value proposition / customer experience to users: this is the best reason
- For continuous improvement and innovation of the platform: this requires only anonymised aggregated data. It should not be linkable to a user, thus not be used for advertising (yes, this is a tough stand!)
- For better targeting of advertisements: this should only be a secondary reason if reason (1) is satisfied. Specifically, data collection for the purpose of providing “personalised” ads should not be a reason by itself
We are going to look at some methods on how platforms collect data.
Given I am taking a tough stand on data capturing, I will spend some time to show you three great examples for capturing location data for a good reason. I hope it will inspire you to come up with similarly good ideas. Equally, it will future-proof your innovation given underlying platforms give users more control over the data that is being collected.
Then we will be learning about how it relates to penetration, segmentation and ultimately targeting of users with (the right) ads.
What data do platforms capture?
Let’s start with a really sad example. A torch app (“Brightest Flashlight LED – Super Bright Torch” with 10 million installs) asking for permissions like:
- Precise user location
- Access to user’s contacts
- Send SMS messages(!)
- Permission to directly call phone numbers(!)
- Permission to reroute outgoing calls(!)
- Access to camera
- Record audio via microphone
- And more …
Really?
Collect data for value in return – deep dive: Location targeting
I strongly encourage you to capture data for the predominant purpose of providing a great value proposition.
As an example, let’s use real-time location data capturing.
Let’s compare three great platforms: Google Maps, Google Waze and Snapchat’s SnapMap. Starting with a summary, note the bolded differences.

This should give you a good entry-level understanding. We are comparing these three great platforms to each other to learn about how to capture data in a good way. Here is how Waze does it!

Step 9: Sub-markets and monetisation
The fact that over 75 search engines and over 200 social networks have tried to succeed shows how difficult it is to monetise successfully to survive and thrive!
Take video advertising. It is one of the fastest growing segments (though obviously not the only one).
Video ad spending is also growing faster than the average ad industry. Among this, video ad spending is clocking in double-digit growth which is forecasted to prevail until 2021 (before Coronavirus).
As part of the shift to mobile, it happens that video ads also move further to mobile phones. The platforms we looked at all offer video ad formats (in some cases it is called rich media when it’s not embedded in a player).

And there are all sorts of different approaches on how to embed video ads.
Here some thoughts on which digital properties allow for the most native way to embed video ads:
- Pinterest has many autoplay (and autoloop) ad videos. Given they are embedded among pins (mainly images), they should not be considered native. And they certainly impact the customer experience negatively (esp auto looping ones). But they (likely) attract a higher price point, thus revenues. I anticipate – like most platforms – Pinterest will prioritise customer experience higher as they grow. A typical observation among platforms
- It is very important to test the impacts of video ads on customer experience
- Take Snapchat: On their Discover digital property, videos don’t autoplay
- And as mentioned, Stories (first invented by Snapchat and successfully copied by Facebook) are a great digital property for video ads. I would think this is the most-native digital property for embedding video ads (far more native than YouTube mid-roll ads!)
- Then there are video ads which are embedded within live streams or other video content. In this case, there are revenue-sharing agreements with the content creators (booked under traffic/content acquisition costs)
One could think that embedding video ads within video content is the most native form of doing this. But that is not necessarily the case.
There are different forms of video ad placement with pros and cons.
- Pre-roll: video ads prior to the premium video, often non-skippable for at least a few seconds, typically around 6secs. They are often seen as the most recallable video ad placement. Much less interruptive than mid-roll
- Mid-roll: video ads during premium content (one downside is that these are typically seen as the most interruptive ones)
- Post-roll: video ads at the end of the premium content (less interruptive, easy to skip and typically lower recall and purchase intention)
Ideally, platforms would only use pre-roll ads but that would considerably limit the available ad spaces. Prices would increase but overall revenues would certainly suffer. Hence, we are likely to still “enjoy” mid-roll.
Some studies try to benchmark the various options on the competing dimensions. Here are two examples that touch on this (one from YuMe, one from AppNexus, before renamed to Xandr).
I would recommend taking these with considerable caution. Any platform needs to check these and other factors (such as the impact on engagement, measured in session duration, number of visits/day, etc) to make sure that they don’t adversely impact important engagement factors.
Hence these tests need to include (which the above studies don’t include):
- Customer experience
- Value for advertisers (marketing ROI)
- Platform engagement metrics
Video advertising is one of the sub markets. We are looking at 15+ charts to look at segments of the overall advertising market in which we can play a role in.
This closes the circle!
We have started with the overall market. We looked at the value proposition which is what companies offer in return for getting paid. I hope you have noticed how staggering the value proposition of social and search platforms is.
We then took a deep dive into the key underpinning principles of multi-sided platforms including network effects.
The long list of platform design considerations will determine its potential.
Then we emerged from the underground to look at the visible elements, these powerhubs called digital properties. They are the linchpin from a user’s perspective.
Data capturing is crucial for platform business models to work and I hope we agree that we want to do this in a good way. You have seen one of several examples that I cover in the advanced resources.
We have closed the circle by looking at one of the fastest growing advertising market segments, video ads.
May this help you with your endeavours whatever they are!