Eric Seufert is the head of marketing & user acquisition at Grey Area Labs, a Helsinki-based mobile gaming studio best known for its inventive location-based title, Shadow Cities.

Working for a start up in the mobile gaming space, I'm constantly reminded of how much people dislike payment walls and forced upgrades in free-to-play games.

They view these tactics as extortion – players invest their time into a game that later becomes unplayable unless a fee is paid.

The first time a player experiences this interruption to the gameplay she has enjoyed up to that point, she becomes suspicious; she wonders how many times this will happen again and if she wouldn't be better served by playing a different game that might not try to coerce her into parting with her money.

I believe the freemium business model is one of the greatest commercial innovations to reach appreciable scale in the past 100 years – but achieving financial success with the model requires a strong conceptual grasp of why the model works.

The basics

The freemium model is predicated on the idea that a free thing will reach a greater audience than a thing that costs money when the transaction medium is frictionless, the thing can monetise through a number of usage scenarios and the marginal cost of distribution is nearly zero.

The model is obviously an appropriate fit for apps, where platforms such as Google Play and the App Store are free to distribute from, and the cost of selling one more copy of Free Fart of the Day is nothing (today's was particularly entertaining, in case you were wondering).

Users can download a freemium app and access its basic functionality for free, but advanced features and add-ons must be paid for.

A small number users for whom the app is most appealing will spend money on the advanced features and add-ons, an even smaller subset of that group will spend a lot of money, and the average revenue per user (ARPU) will settle at a level higher than it would have otherwise been.

The caveats

Understanding this model and adapting it to mobile gaming doesn't seem to require any intellectual acrobatics – after all, games are delivered and consumed in the same way as all apps.

But the freemium model does require some theoretical adjustments when implemented into games, because games are a unique sub-species of app that exhibit characteristics not shared with the rest of the species.

Firstly, games are expensive to develop. Free Fart of the Day could be developed by a college kid at home over winter break; games require art, music, server and client-side programming, and business overhead to develop.

The barrier to entry in gaming is high – much higher than for developing apps. This means that the profitability threshold for a game is much higher than for the average app and can only be achieved through massive scale, because...

Secondly, very few freemium players monetise. As opposed to an app, which accomplishes something, a purchasable in-game item doesn't provide a practical solution to a problem in return for the money spent on it – it provides a virtual crystal, potion, or coin which has no real-world value.

Conversion rates in mobile games are hauntingly low – only about 3-5% of mobile game players monetise – and this phenomenon also contributes to the necessarily massive scale a game must achieve to reach profitability, which is very difficult because...

Thirdly, mobile games are expensive to market. The distribution mediums for mobile games are difficult to use for discovery of new apps because they promote a virtuous cycle environment.

This is problematic because the market has matured to the point where the established developers can buy enough downloads through mobile acquisition networks to propel their games to the top of the rankings, thus making them more easily discovered by new players and alleviating the need for further paid marketing.

The amount of money needed to summit Mt. App Store has increased steadily over the past year as the large developers buy up mobile marketing (paid install) inventory. Small studios without established user bases are competing on a CPI basis with massive, sometimes public corporations with seemingly infinite marketing budgets.

The data bit

These three points paint a picture of an industry dominated solely by behemoths, but fortunately, that's not the case. Studios (even small studios!) are applying the free-to-play model to make highly-profitable games that are still fantastically fun to play.

And they're doing so through the full utilisation of big data.

Back to payment walls. If you're seeing a payment wall in a game and have no intention of buying anything, the game isn't making use of your full behavioural history to predict your predilection to buy.

That's a failure on the game developer's part: when a player sees a payment wall, they get suspicious and might stop playing the game – which is bad, even if they're unlikely to ever monetise.

It's bad because while they might never buy anything in the game, the friends they tell about the game might, or the guy who catches a glimpse of their smartphone screen while clandestinely trying to smell their hair on the metro might.

And if the game is multiplayer, their mere existence in digital form provides cannon fodder for the hardcore players who definitely will spend money in-game.

Keeping the non-payers around is important, and strong-arm payment tactics on users that will most likely never monetise are really bad for that.

The enthusiasm

The free-to-play model doesn't require a high level of conversion to deliver profitability; it requires that the users who will pay for items enthusiastically be given every possible opportunity to enhance the gaming experience through purchasable items.

Broadening the base of paying users is great – and that's accomplished through developing a fun game. Engaging the power players who are happy to exchange their money in return for in-game goods requires a robust analytics platform and sophisticated player segmentation algorithms that can spot these users behaviourally from very early on.

To achieve high levels of monetisation from the very most engaged users, a free-to-play game must command a diverse product catalogue of purchasable items – not because power players will buy all of them, but because the game should cater to the individual tastes of the power players.

A prediction algorithm trained on a massive volume of player events data is a powerful thing: it allows a game to not only predict whether or not a user will buy but what exactly they want to buy.

Those items can then be presented most prominently; if a user has shown that he loves blowing up buildings, why show him flower pots and seeds first in the product catalogue?

The fun

For a free-to-play game to be fun, it must utilise big data; it must recognise when players are willing to purchase in-game goods and allow those users to do so as often and to as great of an extent as they want.

Likewise, it must recognise when players are unlikely to pay and still deliver an engaging experience to them – not because they're entitled to it, but because even people that don't buy items in a game can still be influential brand evangelists.

Most free-to-play games take a generic approach toward gameplay, which runs counter to the core of the business model: more people downloading and playing the game provides more data, which allows those users who do pay to be catered to specifically.

Free-to-play games don't fail because their players don't understand the business model – they fail because the developers don't.
You can follow Eric on Twitter or keep up with his thoughts on big data over on his blog. To find out more about Grey Area, visit the company's website.