Steam Dev Days – In-Game Economies in Team Fortress 2 and Dota 2

In-Game Economies in TF2 and DOTA2

Focus on making your product better.

Use Micro-transactions and economic systems to improve the customer experience. Just using them to extract value will fail.

Part 1 – Lessons Learned & Recommendations

Recommendation #1 – Focus on Persistent Customer Value.

They have to be able pass the “regret” test. If it comes at the cost of customer happiness, don’t do it – even it means that someone might not become your customer. You’ll get them later once you figure out how…

Regret Avoidance Tools

  • Communicate clearly up front (Store front & checkout)
  • Maintain that value over time (Game Design – i.e. Trade Systems)
  • Metrics tracking of customer usage (Back-end systems)

Regret Generation Tools

  • Artificial barriers in the game. Clearly there to extract revenue and provide no service to the customer – aka every appointment-based F2P game ever.
  • Virtual Currencies which obfuscate value.

Recommendation #2 – Positive Externalities

More players spending more makes for a more positive gaming experience. Systems which cause you to have more fun, because someone else spent money. Improve the quality of the game for everyone. Used to evaluate existing designs and as a starting point for new designs.

Recommendation #3 – Make Everything Tradable

Trading makes every item and system in existence more valuable to more people. Every system which interacts with trade becomes more valuable as a result. Trade can become a positive thing for everyone involved – because two customers are interacting together.

Recommendation #4 – Distribute Value Randomly

Random distribution is another tool to generate player engagement and concrete value. Other games use “static” distribution systems with fixed reward schedules based on parameters like time played. That limits what you can offer as a designer. Via random, item values can be dramatically different.

Recommendation #5 – Let Users Make Value for Each Other

Enlist the Internet to fight to make your product even better. Some are amazingly good – use the players of your games. [Valve does this really well…] If you don’t allow them to do so, they’re going to improve someone else’s. [From a hiring perspective, a game company should aspire to be the “reference” standard for portfolio pieces that students produce in school. Valve is well on their way here…]

Reward people financially relative to the value that they contribute and/or generate.

List of possible targets:

  • Comics
  • Movies
  • Animation
  • Tutorials
  • Community
  • Leagues
  • Crafting
  • Trading
  • Moonbase (?)
  • Modeling
  • Gifting
  • Maps
  • Levels
  • Mesh Content
  • Organize groups of like-minded people

Part 2 – Individual Case Studies

Case Study #1 – TF2 Gifts Data (as of 2/1/14)

  • Sent Gift: 1,067,399 accounts
  • Received Gift: 1,841,051 accounts [ratio is closer to 1:1 than I would have expected…]
  • Big gift bundle: 10th highest lifetime revenue generator
  • Current leaders: 12,355 gifts given
  • Given over 1,000 gifts: Over 140 accounts


  • Impetuous – what is something that you can buy which causes me to start celebrating when you make that purchase? What is a system which can generate positive externalities (Recommendation #2)
  • One time consumable.
  • Everyone on the server gets value from your action.
  • You get no direct value.
  • You get social recognition.

Case Study #2 – TF2 Crates & Keys


  • Impetuous – Let players opt-in to random distribution system. This was an experiment to find out if players where interested in spending their money to do this.
  • Most negative feedback around store launch
  • Perception problems:
    ** Capitalize on poor judgement
    ** Maximize revenue extraction
    ** The community ascribed intentions to their actions. Opposite of Recommendation #1.

Follow-on Actions Taken

  • Remove ability to hard-lose real currency on open [What this means is that the items you could get from the crate were worth less than the what the key cost to open. This provoked more negative emotion than positive because the loss generates more than gain. Basic Psychology 101 error… Established a clear rule that the items had to cost more than the key – eliminated the possibility for loss to occur (because trading exists.)]
  • Put users in a situation where they had random chances – not gambling [in the Vegas sense.]
  • Add variety: more types, more contents
  • Add value: common items from crates add rarity-agnostic services. (i.e. crate guns provide gameplay tracking statistics that regular ones don’t.)
  • Participation is higher. Increased community engagement. If they don’t ship crates fast enough, users get angry.
  • “Better is not “done”

Conversation Rate

  • 13% Purchased
  • 15% Opened
  • 75% Owned Item
    This isn’t that useful and potentially damaging. 3 out of 4 people are generating demand and consuming content which comes out of the crate system. The trade system allows Value to effectively monetize free players, because they generate demand for goods which other’s purchase – even if they never spend any money. Everyone along that chain is happier as a result of completing those transactions.

Random Distribution and Trading

Use valuation differences to generate revenue for you and your players, all the while making players happier. Trade combines with random distribution systems in a way which is beneficial for everyone. “Incredibly positive.”

[OK SERIOUSLY.] “Users are potentially trading for keys, because they don’t have credit cards or Valve has poor payment services.” [NOT FOR THE LACK OF TRYING. Geese.]

[Need to analyze the DOTA2 value chain.]

Case Study #3 – Community Content

  • Over 90% of item content from community:
    ** Models, maps, and much more
    ** Marketing (Users make animated videos, amazing art, etc.)
    ** Evaluation (community does this through the Steam Workshop)

Contributor Payout

  • 2010: $590,900 for 106 items by 63 creators
  • 2013: $10,215,796 for 2,349 items by 661 creators
    The community is evaluating the user created content and “voting” with their own money what that content is worth.

Community Content

  • Entirely community-made:
    ** Marketing
    ** Movie
    ** Comic
    ** Organization
  • Then-highest single day revenue in TF history
    Community did their own complete patch of TF2. All of those people collected revenue based on their work.

Case Study #4 – Trade – Item Visibility/Demand

  • “Trade” is a basic idea:
    ** More consumers add value
    ** More uses add value
  • Increase visibility/demand:
    ** Trade as a Steam feature
    ** Community Market
    ** Still growing
  • Partner games benefit in the same way
    If you have an item you can link to it from everywhere. You can trade across games. Steam Marketplace allows users to trade items for Steam wallet funds. The value of ever item in the economy as more games join the economy – there’s no fall-off and it’s reciprocal.
    Daily revenue of some games have gone up by over 50%

Case Study #5 – DOTA2 Leagues

  • Funds split with league organizers, used for prize pool, production, etc.
  • Average 1:8 league viewers traded for ticket, up to 1:4
    Leagues work similar to pay-per-view. Watch the games live. Historical access. Follow a particular player’s mouse movement. 100% spectator.
    Because these tickets were tradable – these leagues all have bigger audiences. Monetizing free players – while making everyone happier.

Case Study #6 – DOTA2 Battle Boosters

  • Goal: “I celebrate when the guy next to me buys one” – Same place as the TF2 Gifts design.
  • Rise of positive, lowering of negative comments at round start. Sea-change level of reaction.
  • Increate the rate at which all players get random drops – even if an opponent buys one.
  • Another free player monetization strategy because the free players are some of those making the positive comments.
  • Iterated to clarify value proposition – because they want more of this.

Case Study #7 – DOTA2 – The Interactive Compendium

  • Positive externalities:
    ** Stretch goals (Kickstarter-like thing.)
    ** Battle Booster (Which effected you, even if you didn’t buy the book.)
    ** Prize pool
  • Results:
    ** 484,768 Sold
    ** ~$1.2m added to prize pool (Single largest eSports event ever.)
    ** Community rallying cry
    DOTA2 tournament that Valve runs. Interactive program with historical game. Fantasy “football” like mini-game. Vote for All Stars. [Think every sports fan participatory thing ever.] Doesn’t fit into any existing model of Micro-transactions. It was a surprising new thing.

Recommendation #6 – Explore!

There are so many ideas out there which don’t fit into existing models. This space is relatively new. Allow everyone to be able to make you’re games more interesting – artists, programmers, experienced players, EVERYONE.

Success Example: TF2 Today

  • 17m accounts owning items
  • 500m total items
  • 4B actions performed on items
  • 5x monthly players (600k -> 3M)
  • 4x daily free items dropped
  • 9x daily items sold
  • 6x daily revenue
    Game continues to grow – which means that the decisions they’re making are keeping their customers happy.

Everyone can win

Value rejects the premise that micro transaction systems must come at a cost of customer happiness.

[It is possible for everyone to make money and come away happy. You just need to think it through – have that as a goal – and it can be done. That is the net best thing for your, and everyone else’s, game.]

Follow Up

Kyle Davis: robin at