More Related Content
Similar to Maximizing Monetization - Casual Connect SF 2013 (20)
Maximizing Monetization - Casual Connect SF 2013
- 1. Maximizing Monetization Casual Connect SF 2013 Emily Greer, Co-Founder & COO
- 2. A little Kong background • Open platform for free browser-based games – Flash, Unity, HTML5, Java, etc. • 15M monthly uniques, core gamers • ~300 games selling virtual goods • Revenue from ads (15%) & virtual goods (85%) • Now a mobile publisher of free-to-play games, first titles launching globally August 2013
- 3. Some of Kongregate’s Existing Developer Partners
- 5. It’s not just genre
- 6. It’s not just quality
- 8. Econ 101 In perfect competition the market price is set where demand & supply are equal. Real life example: the stock market
- 9. Imperfect Competition Perfect competition assumes that goods are homogenous, i.e. that there’s no difference buying from one supplier or another. But nobody can sell a good that’s useful in your game but you. (Ignoring gold farmers)
- 10. Your game is a monopoly
- 11. But I’m surrounded by competition! Yes, and the competition for player ATTENTION got so fierce that it dropped the game price to free. The market for in-game goods is separate: players are not price-shopping packages of gold in two different games and deciding which to buy.
- 12. Since players can leave your game/market for goods freely your monopoly is (very) insecure. What it does mean, though, is that you look internally to your game to set prices, not externally.
- 13. Monopolies can set the price freely, deciding whether to sell fewer units at a higher price or more at a lower price. Marginal revenue is the change in total revenue from a change in price. Monopoly Revenue Maximization Example: 5 units at $5 = $25 7 units at $4 = $28 Marginal Revenue = $3 Total revenue is maximized where marginal revenue = $0
- 14. In an econ class the professor would give you a formula and you’d calculate a derivative. In the real world you need to deduce it from trial & error: set a price, change it and see what happens Uh, how do I figure out where MR=$0?
- 15. It’s all about elasticity When a good is elastic, quantity decreases rapidly with a price increase and total revenue drops. When a good is inelastic, quantity decreases slightly with a price increase but not enough to compensate for the change in price and total revenue increases. Gasoline is a classic example of an inelastic good.
- 16. Now in graphs! Area of the box = total revenue
- 17. So which are virtual goods? Mostly inelastic.
- 18. Immensely popular series by Ninjakiwi, BTD4 introduced virtual goods and was the first big single-player success. Sold 20 items ranging from $0.30 - $10 BTD5 launched last year, selling nearly 40 items from $0.60 - $100 – on average 70% higher on comparable items. Bloons Tower Defense 4 vs 5 Results: 92% increase in ARPPU, -1% decrease in conversion, +88% ARPU Player freakout? Nope. Rating is slightly higher, revenue much higher than BTD4
- 19. Mind the drop Skyshard Heroes is a competitive kingdom-builder with a steampunk theme from Synapse games. They A/B tested dropping the price of their heroes 40% on cohorts of new users, expecting that it would help conversion. Results: +21% in conversion but -25% in total revenue
- 20. Quick Math Break Linear regression is a standard statistical method for modeling the relationship of two variables. The trendline through a scatterplot is the predicted value of variable y given that value for x – the farther the points are from the line, the less predictive x is of y. That error is measured with by the R2 value, where 0 is no relationship and 1 is perfect correlation.
- 21. Curve Line Linear regression doesn’t work very well with exponential functions because they are curves, not lines. But because power curves are scale invariate you can safely transform them to a linear relationship by taking the logarithm of the values. I’m going to be using this technique a fair amount to look at correlations between monetization variables. The Richter Scale is a logarithmic scale of an exponential function, earthquakes.
- 22. Min Price & ARPU
- 23. Correlation Check: Min Price & ARPU
- 24. Max Price & ARPU
- 25. Correlation Check: Max Price & ARPU
- 26. Low entry prices don’t even help conversion rate in any clear way Min Price & % Buyers
- 27. % Long-Time Players & % Buyers So what correlates with increased conversion if it isn’t price? Time in game!
- 28. More on demand curves A demand curve is really the aggregation of individual player demand curves. Factors that shape the individual demand curves: • Desirability (utility) of goods for sale • Income/ability to pay
- 29. Demand curves change with time A player won’t desire goods from your game until they care about their status & progress in the game. The longer someone spends playing a game the greater their investment and emotional attachment, and therefore their willingness to spend. Demand goes up, price elasticity down.
- 30. Conversion rate by lifetime plays
- 31. Gameplays before 1st Purchase
- 32. Retention = Security The more time a player invests in a game, the more they value their status and progress, the higher the switching costs to another game/hobby become.
- 33. % Long-Time Players & ARPU
- 34. Correlation Check: % 100 Plays & ARPU
- 35. More on demand curves The shape of the game’s aggregate demand curve is determined by: • Total players in the market • The shape of the individual demand curves, weighted by their total demand If a market has a few people with high, inelastic demand that can make the whole curve inelastic.
- 36. Big Spenders are a Big Deal Four of the top five games get the majority of revenue from those spending $500+ Every top ten game gets the majority of revenue from players spending $100+
- 37. ARPPU & ARPU
- 38. Correlation Check: ARPPU & ARPU
- 39. Correlation Check: % Buyer & ARPU
- 40. Transaction Size & ARPU
- 41. Correlation Check: Trx Size & ARPU
- 42. Number of Transactions & ARPU
- 43. Correlation Check: Num of Trx & ARPU
- 45. So what do players want? Permanent Upgrades! Items that give real and permanent advantage in the game. Permanent Upgrades > Consumables & Convenience > Cosmetic Items & Content Consumables tend to be in the 10-30% range of sales. Impermanence reduces the value to the player – it feels like money dripping away rather than a sound investment.
- 46. Cosmetic-only items sell poorly though cool looks can help functional items sell. Real-life example: I buy a coat to keep me warm, but am willing to spend more on one that looks good on me. Content is a tough proposition in a world of free, only appeals to those who have finished game. What do players want?
- 47. If powerful items sell, a winning item will sell even better! Maybe temporarily, but if you break your game players will lose interest and leave, even those buying wins. Items can be powerful if they also require skill to use well and are acquirable through a very large amount of play. Balancing advantage vs pay-2-win
- 48. Be careful in designing permanent upgrades/items as it is easy to cap the amount a player can or wants to spend. Ideal: players should be able to invest continually with (close) to linear benefit – players will spend only as long as the in-game benefit is real. Beware of diminishing returns
- 49. Wartune, a very deep MMORPG from R2 Games, has over 25 categories of items for sale, with dozens of items within most of the category. No one category dominates – each adds, and raises the ceiling of spend. More is more
- 50. Back to monopolies In imperfect competition companies can use price differentiation by customer segment to maximize revenue. Classic example: student & senior movie tickets.
- 51. Well-designed upgrade paths for in-game items create natural price differentiation by allowing both power & prices to scale organically within the game. Fusing new items out of old items adds another dimension by ensuring that even low-powered items remain valuable to the player, increasing the value of extensive collections. Having systems like this on a diverse set of items extends the opportunities for purchase exponentially. Upgrade to Infinity
- 52. Get Lucky! Luck-based purchases add an element of excitement while again creating price differentiation – an elastic player will pay a low price once at a chance of an item they’d like, an inelastic player will keep going until they get the item they really want.
- 53. Dawn of the Gacha Dawn of the Dragons from 5th Planet sells “Expeditions” which award a random item with each hard currency purchase. Any one purchase is quite inexpensive ($2-$3) but it can cost $70-$100 to clear the board and get all the rare items. Expeditions drive nearly half the revenue in the game, and when you add in Chests (another chance-based item) the clear majority comes from gacha items.
- 54. Events & Sales Events & sales are very, very powerful – players react emotionally as well as rationally, can drive huge (5x) spikes in revenue. But powerful things can also be destructive: since demand is inelastic you may just be moving sales earlier and at a lower price. The best events serve two purposes: 1. Energizing committed (inelastic) players; giving them more to do and spend on 2. Incenting non-buyers/lapsed buyers with elastic depand to spend (and keep on spending) An event that does both spikes revenue, then resets to a higher base.
- 55. Unleashing Tyrant Tyrant is a great CCG by Synapse Games that’s been one of Kongregate’s top games since it launched in March 2011. Retention and buyer % have always been exceptional but its ARPPU is mediocre, 70% below the average for top games. We’re publishing a new mobile version, which has given us a chance to work with Synapse to significantly revamp the monetization system.
- 56. Tyrant Capped On the web version cards are sold in packs like most CCGs, but strong cards are sold directly on a limited time basis. There are four levels of cards but you’re limited to one legendary per deck. Multiples of cards have limited utility. Sales are dependent on new card releases, creating spikes that fall off quickly.
- 57. Exponential Tyrant • Cards can only be acquired through pack purchases and drops, no direct purchases • Any card can be gained through soft currency purchases or drops • Cards are upgraded through the salvaging of inactive cards • Each card can be upgraded with escalating costs, up to 6x for top-level cards • An effective 5th tier of card rarity is created through fusing specific combos of cards Result? Early test market data shows strong retention, higher conversion, and 2X the ARPPU relative to the web version. PvP has been restructured to focus more on asynchronous combat against well-matched foes, stealing resources for card crafting, and month-long event cycles.
- 58. Mileage will vary! The right strategy will vary dramatically game to game based on retention and base monetization design. If retention is poor (and unfixable) then shorter-term monetization is probably the best you can do. If your audience is younger and/or poorer, demand may be more elastic. If your base monetization is capped, then events and sales may be less productive. Remember that the most important thing is to make a fun game that people really care about. Without that there’s nothing to leverage.
- 59. Thank you! For more talks & this data visit developers.kongregate.com For web games contact us at [email protected] If you’re interested in mobile publishing it’s [email protected] Follow us on Twitter: @EmilyG & @KongregateDevs
Editor's Notes
- #2: Generally the assumption is that maximizing monetization and player satisfaction are contradictory goals, but I really don’t think that’s true.
- #3: So let’s take a look at some of that data. Before I dive in I want to give a few definitions so that it’s clear what I’m talking about.
- #4: The advantage of having a platform is that we can do cross-game comparisons across a wide set of games (Eastern & Western, big companies & small studios, MMOs & social games) to better understand what drives game success
- #5: This chart takes all games selling virtual goods on Kongregate and ranks them by ARPU – as you can see it is a classic power curve. Most games don’t do very well at all (the median ARPU is $0.08), but the top-performing games make a LOT of money (top ARPU is 100x the median). ARPU is obviously not the same thing as revenue, but it is a pretty good proxy – most of the variance from ARPU to revenue has to do with the time a game has been on the site. So what’s driving this? Some obvious explanations have to do with the genre of the game, multiplayer elements, etc.
- #6: I limited the data to just one multiplayer genre we had a pretty good sample set on: collectible card games. All 20 of these game have multiplayer PvP and monetize primarily by selling cards/card packs and yet you can still clearly see the power curve in action.So how about the quality of the game? Maybe the poor monetizing games just aren’t very fun or appealing.
- #7: Nope. In this chart I’m adding in the rating of the game, which on Kongregate is a pretty good proxy for the initial impressions and fun level of a game – our users take rating quite seriously. Anything over 3 is a good game, anything over 4 is really strong, but there is almost no correlation with monetization.
- #8: Well there’s a lot going on, but for now I’m going to step back and look at the base economics at play (warning, I nearly went to grad school in Econ before starting Kongregate so I may geek out a little here).
- #9: negative slope means qty demanded goes up as price goes downpositive slope means qty supplied goes up with pricePrices stabilize where marginal revenue = marginal cost
- #10: I don’t care (or even know) who I’m buying a stock from, but it doesn’t matter if another game has an awesome sword to buy, it doesn’t work in the game that I’m playing.
- #13: You’re not a cable provider or the like who can charge high prices and provide poor service and the player is stuck. Players should be treated well and will leave if you don’t.
- #14: Marginal revenue is the additional revenue
- #15: You’re not a cable provider or the like who can charge high prices and provide poor service and the player is stuck. Players should be treated well and will leave if you don’t.
- #19: Marginal revenue is the additional revenue
- #20: Marginal revenue is the additional revenue
- #21: I’m going to start diving into some statistical correlations
- #22: I’m going to start diving into some statistical correlations
- #24: I’m going to start diving into some statistical correlations
- #25: Now I don’t think adding a $1,000 currency bundle is going to magically alter the monetization in your game – having larger price points reflects the designer’s expectations of what someone might want to spend as much as anything, and the designer allowing and encouraging deeper spend is what makes the difference
- #26: Now I don’t think adding a $1,000 currency bundle is going to magically alter the monetization in your game – having larger price points reflects the designer’s expectations of what someone might want to spend as much as anything, and the designer allowing and encouraging deeper spend is what makes the difference
- #30: I’m going to start diving into some statistical correlations
- #31: By 900 plays it’s approaching 50% on average.
- #32: Almost nobody buys on their first play. The median # of gameplays before a player’s first purchase is 28, and at 100 plays you’ve still only converted about 70% of buyers who will eventually purchase
- #33: I’m going to start diving into some statistical correlations
- #34: Because of the power curve of ARPU it’s a bit hard to see the trend with just the values
- #35: This is a really strong correlation for an element (plays) that has no direct relationship to revenue
- #36: That’s what I believe is happening here.
- #38: Now these of course are going to be correlated, since ARPU = ARPPU x % Buyers, and it’s visible on this chart.
- #39: This is as tight a correlation as you’re likely to see and implies that % Buyer is much less of the driver of high monetization than high ARPPU
- #40: Still pretty good as you would expect, but much lower correlation to ARPU than ARPPU. A good % conversion is a pre-condition to good monetization, but not sufficient.
- #41: Let’s break ARPPU into it’s two components: Avg Transaction Size and Number of Transaction.
- #42: AvgTrx $ matters as much to ARPU as conversion.
- #43: Most games average around 3-4 transactions per buyer, but the best games do much better.
- #44: The number of transactions matter WAY more. Big spenders are not about Saudi princes dropping $1k on a single purchase (though that does happen occasionally) but about steady investment in reasonable-size purchases over time.
- #45: We all tend to obsess over games/companies showing exponential growth but more often that not that’s followed by a crash. Linear may be less sexy and headline-catching, but the tortoise generally does win out. Top games still making as much or more as at launch as they hit 2- and 3-year anniversaries on Kong.
- #46: Energy systems are a retention mechanism, not a revenue-driver.
- #48: Players want to achieve in the face of challenge, over-powered items are boring.Advantage doesn’t have to be substantial to be compelling. In a competitive environment a 1-2% (check) improvement can be plenty to make a purchase tempting.
- #49: Example: selling a weapon that is effective at all times/parts of the game, even worse if you can only use one weapon at a time. Why buy another?The cost doesn’t need to be linear, though.
- #50: Thanks very much to R2 for allowing us to share this data. to Some sharp-eyed people will notice that clothing is the second largest category – clothing in Wartune is not strictly cosmetic as it gives you slight upgrades.
- #51: Example: selling a weapon that is effective at all times/parts of the game, even worse if you can only use one weapon at a time. Why buy another?
- #52: Crystal Saga does this well. Wings and pets can be upgraded by combining with tokens (available through loot drops or sold for $0.30-$0.50). Each upgrade costs more tokens and involves more luck, scaling costs significantly with items that are cheap on their own. That luck element creates both drama and frustration as the player doesn’t know if an investment is going to be a total waste.
- #53: Luck systems in general are important parts of monetization, though I think most effective where luck can get you something better than what you expected rather than negative experiences of failure. Collectible Card Games automatically have this by selling cards in packs, with more expensive packs having a higher chance at rare cards. These types of mechanics are often called “Gacha”, which is the Japanese term. I think most effective where luck can get you something better than what you expected rather than negative experiences of failureUpgrades, fusion, luck mechanics all make purchasing more interesting – they’re ultimately part of the gameplay, which makes them more compelling.
- #54: One of the things I like about this implementation is that it is extremely well-labeled and clear – each item is labeled with the precise chance of receiving it, and whether you can receive duplicates. The new item pack only awards one of each item, so you are guaranteed to eventually get the item you want. Misleading luck-based systems may drive sales in the short-term, but long-term they alienate players who feel frustrated and taken advantage of, and make them more likely to churn. Dawn has exceptional retention.
- #55: Month long event cycles are good, go to frequent and you create numbness, go to rarely and players get bored. Vary the events as much as possible, create natural peaks and valleys. Limited time items can be great.
- #57: Some cards you can only purchase directly with hard currently. One’s investment in cards is quite linear, and by capping the # of legendaries in a deck diminishing returns set in quickly.
- #58: Focus in the mobile was in increasing the depth of the card collection systemsThis simultaneously reduces the feeling of pay-to-win while increasing the luck and fun of collectingThe salvaging of cards works well in concert with a pack-only purchasing system: even the cards you don’t want are of guaranteed usefulnessOne salvage point for a base common cardA fully-upgrade legendary takes 750 salvage pointsFused cards are then upgraded again, for a full-upgrade cost of more than 2,000 salvage points.
- #60: This is about how you think about your business.