Monopoly lego

I’ve been thinking a lot about my new side project lately (“thinking” including writing down an insane amount of things in various Google Docs along with too much code to keep my sanity), a large part of it being whether or not going freemium is a good idea.

Freemium is a very powerful way to find new client when well done. Poorly executed, freemium will cost you a lot of money and make you miss clients. The past 7 years, I’ve been working in 2 companies doing freemium, helped 2 others, and analyzed countless of freemium stories. Here are a raw dump of my thoughts for my next project as I think it may interest someone else.

Is your product as addictive as heroin?

Freemium is your usual heroin dealer: the first shoot is free because addiction starts the first time. Whether you want to rely on subscription or in app purchase, this is the first thing you need to know: will your clients use your product regularly, on a yearly scale?

Freemium is not offering a free test, but allowing a limited use for free. That’s very different because the concept of test itself implies a one shot use. If a product can be used for a one shot ride, going freemium is a terrible idea, because no one will take a subscription for it.

The subscription model is very interesting. It means recurring revenues, easier forecasting, and at some point, captive clients. This model, and the in app purchase one, which is of the same kind does not fit every service.


Let’s imagine Uber going freemium. That would be a terrible idea, so let’s go the worst way possible. People would get a free ride first. Then, they would pay a yearly subscription to be allowed to use the service again, and would pay each and every ride. From a short sight point of view, this is an awesome model: since clients have to pay a subscription, they indeed use the service a lot.

This would indeed be a total failure. Uber is a service you can use once and then forget for months or years, and here’s what would happen then:

  • People would use the service once for a free ride. Those free rides would cost Uber an insane amount of money, because they would to pay the drivers anyway.
  • 99% of them would never come back, because they only need a private chauffeur only once or twice a year.
  • The 1% taking the subscription would not use it enough during the year to make it interesting and would cancel at the end.
  • The subscriptions and rides would not cover the free ride costs.
  • Uber would quickly fill for bankruptcy.

Buffer and Candy Crush Saga are 2 great examples of heroin shot done right.

Candy Crush

Candy Crush is free, the gameplay is awesome, and the difficulty progressive enough so you would not imagine anyone stop playing before they reach level 30 or 40. This is the heroin shot. Then, the game business model relies on a “wait or pay” model:

  • If you’re stuck,and running out of lives, you can pay to get some help through the next level, or wait until your life counter refill.
  • If you want more level, you need to go through a few levels. Between every level, you need to wait 24 hours, or pay to get your new levels.

To gain customers, Candy Crush plays with a human, powerful string: frustration. Once you start traveling through the levels, it’s very hard to stop. You can compete with your friends on Facebook, so you get updates every time they reach a new level, and you didn’t because you didn’t pay… yet. Isn’t waiting 72 hours to get 10 new levels frustrating? Indeed it is. Frustration is the reason why people pay to play a free app.

Buffer is an awesome service to push scheduled updates on your favorite social networks. Buffer target is a B2B SAAS service., and for most people (just like me), the free model is cool: limited accounts, limited updates per day, no detailed analytics, but once you’ve starting using it for real, there’s no way back, because it saves you lots of time.

I’ve studied Buffer a lot, which is rather easy since they promote transparency to the max. The premium plan is very affordable (and there’s only 1, for a 10 USD a month) and plays on both your ego and your business needs:

  • If you want more than a few accounts, you have to pay.
  • If you want multiple schedules (like one for the week-end), you have to pay.
  • If you want detailed analytics, you have to pay. Detailed analytics are mandatory to explain that paying someone to be on Twitter and Facebook all the day is not a complete waste of time and money.

How much does a user cost?

Users cost

In a freemium service, you have 10 types of people: clients, who pay, and users, who don’t. Since they don’t pay, users cost you money, sometimes lots of money, and knowing how much a user costs, and how many of them you need to turn one of them into a client is critical. Having tools to track conversion rates on a daily basis as soon as you launch is critical.

From my operational point of view, tracking a user’s cost is about:

  • How much running the product costs.
  • How much the marketing actions for awareness cost.

I’m probably missing a lot of things, I’ll add them as my project goes. Then, I’m counting on a weekly basis.

  • The more users onboarding, the less a user costs.
  • If the number of new clients don’t cover the number of new users, there’s a problem. The user cost needs to be lowered, as well as the fixed costs of running the product and acquiring the new users.
  • If the new and existing clients can’t cover the users cost, there’s a problem in the model. Maintaining existing users should have a very low cost, that cost being calculated on a weekly basis.

After running a freemium for a while, you’ll probably reconsider your model if the users cost still can’t be covered by the money paid by the clients.

What’s your killer way to attract new clients?

Free beer... tomorrow

Over the years, I’ve been studying lots of companies running a freemium business. Attracting new clients without spending all your money is the hardest part of the job. Keeping them using your product so they renew their subscription is almost as hard as onboarding them

Realization is the key, and it starts with a real go to market strategy. No bullshit there: if you don’t know how you’re going to get your users, don’t even start a freemium (or even a business).

From all the companies I have studied, the worst freemium I’ve ever seen was the one done by my own employer, and one of the best by our then competitor: Yammer. 3 years after going freemium, they were acquired 1 billion USD by Microsoft.

Yammer started like a Twitter for enterprise. I’ve been working for an enterprise social network for 6 years, and the hardest part in that job is to get people use your product. Enterprise social networks need lots of people from the same company being onboard to work well, as it’s overlapping many existing tools. Here’s how Yammer did.

Anyone could open a Yammer account for the company. It was free and incredibly easy, everything you needed was an email Then, you would invite your colleagues having the same email root domain to join, freely. There was no user limit, only an email domain name one. Every new user was able to invite his colleagues, this was not reserved to the administrator.

What’s important there is how Yammer started recruiting the company employees from the bottom, completely bypassing the management and hierarchy. Once half of the company started to use Yammer as a work tool, it was very hard for the company to refuse to buy it.

That’s where Yammer had a hard time: keeping users coming past the first onboarding.

The other company I loved studying is Github.

Github onboarding model is stupid simple: they host your code for free as long as it is open source. And they provide you feature to manage it: ticketing system, issues, wiki pages and organizations. If you have a single line of code to hide, you have to pay.

Most important open source projects have moved to Github. The Linux kernel is hosted on Github. Ruby on Rails is hosted on Github. FreeBSD is hosted on Github…. Today, I think almost every developer and system administrator using open source software has a Github account. Even though you don’t own an open source project, you can fork other people project so you have your own repository.

Github model is genius, because when one of these developper joins a company, and there’s a need for a fast to setup, reliable and cheap source code management tool, they will think about Github paying plan, since they’re already using it.

Do you have key feature you can hide behind a paywall?

That’s the final question, but it’s as critical as the previous ones. To go freemium, finding the right balance between uselessness and frustration is key. If a product offers the feature most users want in the free version, no one will pay for it. But without enough feature to make it useable, a product won’t attract anyone first.

Finding the right limitation cursor is key.

Looking back at the company I mentioned in this article, the way they setup their cursor is very interesting.

Candy Crush plays with frustration by letting you play enough to make you addicted, but leaving most of the game levels behind a paywall.

Buffer gives you basic feature for free, but keeps the business critical ones behind the paywall. Buffer free version is perfectly useable below a certain level.

Yammer model was relying on offering an agora for the company, but private discussion groups were in the paying plans. So it would become an annexe to the coffee machine before a working tool, which led to frustration.

Github hosts code for free, if it’s open source. Companies will usually want to hide a few things from their competition.

That’s all for tonight folks, I indeed have found a few answers to my questions (except the users cost as I don’t have a running product yet, even though I have a very pessimistic forecast). I hope this article was helpful to you. Wishing you a nice week-end!

Perry the Platypus wants you to subscribe now! Even if you don't visit my site on a regular basis, you can get the latest posts delivered to you for free via Email: