The Freemium Business Model and Viral Product Management

By Scott Sehlhorst September 16, 2009


  • Fixed costs. These costs are the same for the company, no matter how many users there are; additional users add no incremental costs.

  • Variable costs. These costs are the same per marginal user-—incremental users add incremental costs.

  • Total costs is the sum of fixed and variable costs.

  • There are also two important ways to look at profitability: overall and per-product sale:

  • Total profit is the sum of all product sales minus the total costs to make and sell the product, including overhead.

  • Contribution margin is the difference between product revenue and the variable costs to make and sell the product.

When the total revenue from product sales exceeds the total costs to make and sell that product, the product is profitable. From a decision-making standpoint, the contribution margin needs to be positive. The number of products that need to be sold for the company to be profitable is the fixed costs divided by the contribution margin.
Here’s an example using a Software as a Service (SaaS) pricing model:

  • Your business incurs $10,000 per month in fixed costs.
  • Your product has a variable cost of $0.10 per month per user.
  • 1% of your subscribers pay for their subscription; 99% subscribe to the free version.
  • You price your product at $20 per month, per user (per unit subscribed).

That looks like a very profitable product-—some people will pay $20 for something that costs you a dime. But looks are deceiving. You have to cover the costs of the free subscribers, and you have to cover the fixed costs of making and selling your product.

Ever scratch your head and wonder why you can use your favorite application for free? How can a business actually make money (and stay in business) when they offer their product for free? When does it make sense for a company to offer a free version of a product that competes with their own for-a-fee version of the same product? The freemium model is one where the company offers two (or more) versions of a product. The basic version is free to use. You have to pay for the premium version.

Once you do create both free and for-a-fee versions of a product (here the word “product” also means services and combinations of products and services), how do you approach positioning and pricing of the products so that your company stays in business? Do you encourage viral growth, and what is it about a product that makes it inherently viral?

Economics of a freemium business model

One way to look at the freemium business model is to consider the choices each user makes. By definition, a freemium model is where one user is faced with a choice: Do I use it for free, or do I use it for-a-fee?
We will discuss how to encourage users to move from the free version to the for-a-fee version later. For now, we’ll just look at the impact of that choice.

Billing Peter to pay for Paul (freemium)

Every free user gets some benefits at no cost. Every for-a-fee user pays to get the full benefits of the product. The customers who pay for your product also cover the costs you incur when providing it for free to other users.

As a company, look at your aggregate user base to analyze the economics. What makes this analysis meaningful is asking the question, “What percentage of your users will pay when a free version is available?” Basecamp, from 37signals recently celebrated its fifth anniversary and serves as a good example. Note that 37signals expressly does not share this “conversion rate” information (the percentage of free users who are converted into for-a-fee customers), so we have to speculate.

In a 2006 interview with Ryan Carson for ThinkVitamin, Jason Fried, co-founder and president of 37signals, indicated that their conversion rate was “more than 0.87%.” So we’ll call that 1%.

In a 2009 interview with Brad Spirrison for, Fried indicated that 90% of revenue for 37signals comes from subscriptions to web applications. Spirrison points out that 37signals earns $40,000 monthly from its job board, so we’ll estimate $360,000 per month from subscriptions.

We can sanity-check our 1% estimate. Fees for 37signals for-a-fee products range from $24 to $149 per month. If the average customer pays $36 per month, then there would be 10,000 paying customers-—1% of a million. We could tweak the numbers (the average might be lower; there may be more than a million users, etc.); but this data is consistent with a 1% conversion rate.

Blogger Jed Christiansen did an analysis about a year ago where he estimated about $5,000,000 per year in 37signals revenues-—numbers that are consistent with the Spirrison interview. Jed built his estimates from the usage stats that 37signals reported, along with some assumptions for converting from usage to number-of-users. His estimates would put conversion somewhere around 0.5% to 1%. He provides a spreadsheet of the model too, if you want to tinker with it.

This feels reasonable-—100 free users for every paying user. Even if that number is wrong, the rest of this article holds true, but it sometimes helps to have a number to think about.

The left hand doesn’t know what the right hand is paying (not freemium)

There are other ways to “pay for” providing a free product, but freemium only applies to the situation where two versions of the same product are offered to the same users-—one for free and one for a fee. The following describes a perfectly valid way to do this-—just don’t mistakenly label it as freemium.

This left-hand/right-hand situation is where a user gets a product for free and a different user gets a different product for a fee. Technically, this is not a freemium model; the same user does not choose between the two options.

The following are examples of this model:

  • A company offers a product for free to (primary) users and charges advertisers (secondary users) to display ads to the primary users. This is an ad-supported business model.
  • A conference offers the opportunity to speak/present (for free) to lecturers and charges attendees to listen to the lectures.
  • A government offers waivers on corporate or property taxes to a company to build a new facility and levees payroll taxes against the employees for the privilege of working there.
  • A shopping mall hosts free events (such as holiday pageants) for the general public and charges the retailers for rental space in the mall. These events make the mall a more attractive location for retailers, leading to higher rent or occupancy or both.

In each of these scenarios, the users who get the free product are not choosing it relative to the paid product. Different users are targeted for each.

Freemium product costs and prices

Isolating the freemium business model from other revenue-generating opportunities, shows that finding a profitable model can be tough-—you have to correctly control costs and set prices. Assuming our data from the preceding example is representative (and I don’t know that it is), if 1% of customers are paying customers, then each paying customer has to cover the costs of 100 free customers to have the possibility of being profitable.

The diagrams below show how long it would take for your product to be profitable with both linear and exponential growth curves. A linear curve is what you might expect from “traditional” marketing investments-—where $X spent yields Y users. An exponential growth curve is what you might expect from a viral marketing approach-—where each user tells X users, who then tell X additional users… and so on.

In the linear model, you have to get to 100,000 subscribers (1,000 paying customers) just to break even. This takes much longer than you would expect when selling dimes for $20! Even a 25% per month growth rate can’t help you early on.

The exponential growth does start to compound, but it also delays the break-even point. This delay happens because you have to cover the costs to serve 100 free-account subscribers with the revenue from each paying customer. The contribution margin is the key here, and three things have to be true, or you shouldn’t have a freemium business model:

  • You have to have a contribution margin that is positive when taking into account the ratio of free users to for-a-fee users.
  • You have to have a sufficiently large customer base (number of paying customers) to cover your fixed costs.
  • You have to lower your costs (if your contribution margin is not positive) and grow your customer base (if it is not large enough) fast enough to become profitable-—before you run out of funding.

Growing your customer base—word of mouth

There are a two ways to grow your customer base: traditional marketing and word-of-mouth marketing. If you’re relying on word-of-mouth marketing, there are two dynamics that drive word-of-mouth [thanks to Jonathan Berkowitz of for this insight!]-—altruistic and selfish:

  • Altruistic. This product helps me; it will help you, too. You should use it.
  • Selfish. It helps me if you start using this product. You should use it.

To leverage word-of-mouth, you create a product and a context where people want to tell others about your product. Ideally, your customers (free or otherwise) will like your product so much that they want others to use it, not just know about it.

There are two considerations in generating word-of-mouth results: marketing and product management:

  • Word-of-mouth marketing. People in marketing, PR, and corporate communications talk a lot about viral marketing. Viral marketing is when you create a message that is implicitly viral, causing exposure for your product. A great example of a viral message is the Mentos/Diet Coke videos. People shared the video because of the video’s entertainment value, not because of the intrinsic qualities of the Mentos and Diet Coke products. Viral marketing is different than viral product management.
  • Word-of-mouth product management. Viral product management is the action we take to make a product self-propagate or self-promote —as opposed to the explicit viral marketing we do to generate awareness. You can think of viral product management as implicit marketing: a feature or capability that might have a direct impact on your product’s word-of-mouth. At a high level, you simply need to create a product your customers want others to start using. But here’s the catch: They have to want it enough to encourage others to start using it.

iPhone market data

Pragmatic Institute likes to remind us that our opinions, although interesting, are irrelevant. So here is some data. Greg Yardley of Pinch Media shared a very interesting YouTube analysis of ad-supported iPhone applications versus paid applications. What is most relevant to viral product management is the graph on slide 26 of his analysis.

What the graph shows is that the top 10% of ad-supported applications break away from the other 90% in terms of usage. The reason (in that presentation) for looking at the data was for measuring the CPM-based revenue (cost per thousand impressions) for applications in comparison with the charge-for-the-application model. Ultimately, the free version makes sense, but only for the top 5% of applications, according to Pinch Media. For the rest of the field, a for-a-fee application will likely generate more revenue.

It’s clear from the data that the top 10% of the applications (the discrete red line at the top of the graph) are different from the other applications. Something causes users to want to use those applications significantly more than others.
What Pinch Media’s data does not show is how viral the applications are. In other words, do the top 10% of applications (in repeat usage per user) also have the fastest growth (in user count); and, if so, is it by word of mouth? My basic assumption is that the applications that are most pleasurable to use are the ones that can achieve viral growth. These are the applications you want to encourage others to use.

Pleasurable products

As a user-centered product manager, you are likely spending some of your time on customer delight features and capabilities. These are the capabilities that wow customers and cause them to tell others about your product because it is a cool and useful product. You’re also spending time on the more is better features and on usability concerns that make a product a pleasure to use.

You keep making it better, because there is clear ROI to increasing your user base. Improving usability makes it more likely that people will tell others about your software. Eventually, your product will be good enough that people will start telling others, and your growth will start to climb.

One of the dangers of this incremental product growth strategy is that you catch a case of featuritis. Featuritis is the malady where adding too many features to your product makes it less pleasurable to use.

Very rarely will you hear people clamoring to remove features from your product. You’re more likely to hear a steady stream of “one more thing” requests.

The problem is, too many features can make it hard for people to learn how to use your product. Simply put, there’s a threshold of user tolerance for features. Having too few or too many features will make your product unpleasant to use. Above that threshold, the product is a pleasure to use. Kathy Sierra coined the term “suck threshold,” to mark this delineation.

Extending the suck-threshold dynamics (see sidebar below) with Malcolm Gladwell’s concept of a Tipping Point, where things discontinuously change, there will also be some threshold by which your product is so pleasurable to use, that people will feel compelled to share it.

In Pinch Media’s iPhone application analysis, we see that 90% of the applications didn’t appear to tip, so I’ll show the default tipping point as being out of reach (at least until you do something about it). Clearly, your investments in user interaction (and added capabilities) can increase levels of user happiness with your product until it exceeds this viral tipping point.

Modes of viral product propagation

Remembering the two primary human-nature mechanisms by which a product will propagate virally—altruism and selfishness—we can draw some additional suck-threshold conclusions. If your product falls below the suck threshold, I don’t believe you can sustain any form of viral growth. People will be discouraged from (maybe even embarrassed about) telling others about your product.

Sharing a product recommendation builds on trust; sharing something that people won’t like erodes that trust. I believe this is a self-correcting behavior, and what little sharing might occur will be short-lived.

  • Leveraging altruism as a viral mechanism. A product that gets shared because of altruism needs to not only be better than good, it has to be so good that you’ll go out of your way to tell people about it-—with no explicit benefit from sharing. If you make your product so good that people feel compelled to tell their friends about it (or blog or tweet about it), you’ve got a great product.

    Product management decisions to achieve this feat are “easy” in that you only have to make the product fantastic for your users. At the same time, your product management decisions are difficult, because you have to make your product good enough to cross the viral tipping point.

    The iPhone, Synergy, Benjamin Moore paint (when the first local store opened in Austin, the owner was stunned by how much brand-loyal demand was out there), Tweetdeck (a Twitter client), and GMail are all examples of products that have tipped. Additional users/customers don’t make the experience any better for the current users, but people still rave about it to their friends and associates.

  • Leveraging selfishness as a viral mechanism. There are two ways to leverage people’s inherent selfishness when developing products. The first (and harder) is to define capabilities or features for the product where the customer’s experience is better when more people use the product.

    Twitter (and Facebook and other social media applications) take advantage of this. If you’re using Twitter as a broadcast medium, then the more people who listen to you, the more value Twitter has for you. So you encourage people to use it. On the reverse side, if you’re looking to Twitter as a source of good information, the more people who are sharing information on Twitter, the more valuable Twitter is to you.

    The second (and easier) way to reward customers for encouraging other people to use your product is to explicitly reward them. Affiliate programs, finder’s fees, account credits, or other compensation can be given to existing customers in exchange for signing up new customers. A SaaS variant of this would be a program that rewards you with credits to your account for every customer you refer, for as long as both of your accounts are active. People can get your product for free if they encourage enough other people to buy.

To be sustainable, either of these selfishness-model variants would need to be leveraged to promote a product that is actually good-—at least above the suck threshold. But the product does not need to be above the viral tipping point.


You can create viral messages or videos that spread awareness of your product tangentially…or you can create compensation programs that encourage people to promote your product…or (best of all) you can create products that promote themselves.

As a product manager, it is far better to create a viral product that people love than to cross your fingers and hope that a marketing message creates viral exposure. Simply put, why not be intentional about viral product management rather than hope to get lucky?


In my blog, Tyner Blain, I’ve written an article about featuritis called Goldilocks and the Three Products, which builds on some great suck-threshold ideas from Kathy Sierra.

These diagrams focus on the holistic ��How good is your product from a user perspective?” question and take a Kano-analysis approach to looking at ways to improve your product.

You can improve the curve for any particular product by improving the user’s experience with “more is better” features. You can either improve usability or improve performance or improve both to change the shape of the curve above. This change increases the likelihood that your product will be above the suck threshold.

Categories: Pricing Requirements
Scott Sehlhorst

Scott Sehlhorst

Scott has been helping companies achieve Software Product Success since 1997, and started Tyner Blain in 2005. Scott is a product management and strategy consultant, and a visiting lecturer for DIT's Product Management program. Scott has managed teams from 5 to 50, and delivered millions of dollars in value to his customers. You can reach Scott at, or join in the conversation on the Tyner Blain blog.


Looking for the latest in product and data science? Get our articles, webinars and podcasts.