(To learn how Game Data Pros can help you optimize your games, contact us)
In 2024, we held the first-ever Revenue Optimization in Games Mini-Summit at GDC. We did it because we didn’t like that there aren’t many revenue optimization talks at GDC and that, in general, the idea of “Game Revenue Optimization” doesn’t seem to get much, if any, mindshare at industry conferences.
So, instead of grousing, we organized our own summit in 2024. The feedback we got was incredible — the attendees loved the event, they thought the talks were amazing, and, more generally, they spent the next year asking us if we were going to do it again.
Spoiler Alert— We did. We rented the same venue (the incredible American Bookbinders Museum), ordered a few thousand dollars worth of goat-cheese tarts and coconut macaroons, invited the world, and put on a show.
And what a show it was!
First and foremost, we had a set of world-class talks
After a brief introduction by Pallas Horwitz, the day’s emcee, the talks began at 2:15. We had five speakers.
Our CEO, Bill Grosso, opened the show with “10 Reasons MMMs Are More Interesting Than You Think” — an overview of how Generative AI combines with open-source libraries like Meridian to make building a good and useful MMM much more accessible to small companies than it was even 5 years ago (slides).
Then Ryo Shima, CEO of JetSynthesys Japan, presented “How Game Revenue Optimization is Different in Japan” — an in-depth discussion of the behavioral differences between Japanese and Western gamers, and how that impacts monetization strategies (slides).
Tiffany Keller, one of the superstars at Liquid and Grit, followed Ryo and gave a talk on “Advanced Hybrid Monetization.” This was the graduate seminar version of the roundtable she held last February and was a comprehensive overview of the state of the art in hybrid monetization.
And, finally, Joost Van Dreunen and Julian Runge closed the presentation part of the day by presenting a sweeping overview of the future of brand engagement with gaming (slides).
Speakers, from Left: Pallas Horwitz, Bill Grosso, Ryo Shima, Tiffany Keller, Julian Runge, and Joost Van Dreunen.
Second, we had an amazing audience
Like last year, we were slightly nervous about this — the room only holds 105 people, and we had 340 people registered. Ultimately, we decided to issue 180 tickets. 75 people came, most stayed for the entire summit, and the event turned into a caffeine-fueled group conversation about revenue optimization.
The combination of the incredible speakers and the amazing audience meant that the happy hour was more than just an excuse to have salmon brioches and artichoke salad while downing plastic glasses of red wine. The food was good, but the conversations were excellentand lasted until the museum closed.
In the first blog post in our series on the history of revenue optimization, The Origins of Revenue Optimization, we explored the early practices of revenue optimization in the airline industry, where dynamic pricing and sophisticated forecasting models revolutionized pricing for airfares, hotel bookings, and other travel-related services.
In the second article in this series, A Brief History of Revenue Optimization: Airline Fares and Price/Time-to-Flight Curves, we examined one key aspect of airline pricing strategies: price/time-to-flight curves map changes in ticket prices as the departure date nears. These curves demonstrate a simplified version of the algorithms that airlines use to strategically price tickets to maximize revenue while filling as many seats as possible.
In this article, we’ll take a closer look at the data historically needed for successful revenue optimization efforts and some of the challenges and advancements that have beset data-focused revenue optimization efforts. When Robert Crandall developed his early revenue optimization techniques — which he called yield management — the idea was to maximize revenue by managing the inventory of airline seats based on demand forecasting.
From the beginning, successful revenue optimization efforts depended on careful data analysis. And data analysis relies foremost on collecting useful data (comprehensive and correct data sets). Many businesses collect data but aren’t particularly worried about noise or the occasional incorrect value. That’s a huge problem in gaming. If you’re not validating the data you collect, are you really collecting data?
Computerized Booking and Data Capture
Revenue optimization in the travel and hospitality industries, including airlines and hotels, relies heavily on rich, long-term data sources to enhance decision-making processes. These data sources help organizations develop strategies that maximize revenue by predicting demand, optimizing pricing, and managing inventory efficiently.
Airline revenue optimization efforts began in the 70s and heavily leveraged the emerging development of computerized central reservation systems (CRS) that had started becoming available in the mid-1960s. American Airlines and IBM had collaborated on the first widely successful CRS, the Semi-Automatic Business Research Environment (SABRE), which went online in 1960. Other airlines followed suit, leading to a number of competing CRS systems, including Delta’s DATAS, TWA’s PARS, and United Airlines’ Apollo.
For context, many of the modern data capture, storage, and analysis tools we’ve come to appreciate were not yet ready for prime time or even available at all. Prior to the early 1970s, reservation systems relied on manual record keeping and tabulation, microfilm, or flat data files. The relational data model paper by Edgar Codd was published in 1970. SQL (then SEQUEL) was initially developed at IBM during the mid-1970s. And commercial relational database management systems (RDBMS) weren’t available until 1979.
When Ken Littlewood wrote his 1972 paper Forecasting and Control of Passenger Bookings, BOAC’s BOADICEA system was a key element. It provided accurate, historical booking data for analysis and prediction that could later be correlated against actual flight manifest data.
The BOADICEA system’s Flight Transaction History File is a good example of the data available for early airline revenue optimization analysis. The Flight Transaction History File represents a historical compilation of passenger name records (PNRs) extracts dumped from daily flight departure reports.
“Each line represents a transaction associated with the flight and shows the number of days before departure and the time (in minutes past midnight) at which the transaction was made. The sector booked, the class and the number of seats sold are also recorded.”
Codes provided further metadata about a booking, such as cancellations, no-shows, upgrades or downgrades, and departed loads by sector and class.
The features available for analysis (flight, number of days, time, sector, class, number of seats sold, price) were chosen based on subject matter expertise. There were a very limited number of attributes used to build models because of computational restrictions. In the modern world, we can handle hundreds or thousands of covariates, and we have much more advanced statistical techniques available to help us choose and evaluate them.
The data was rudimentary, but it was enough for Littlewood to perform critical analysis, such as forecasting demand by day and sector and comparing the forecasted demand to actual customers in seats at flight time. While most airline optimization research was based on maximizing the number of passengers per flight, Littlewood was able to use the limited data on hand to also generate a successful model for maximizing revenue by controlling low-yield fare allocation versus high-yield fares.
As similar digital booking capabilities and data storage and analysis tools became increasingly available across the travel industry, more sophisticated revenue analysis and optimization efforts usually followed.
Bill Marriott Jr.’s work to formalize a practice in the hotel business of ensuring high occupancy while maximizing revenue per available room (RevPAR) coincided with the emergence of digital property management systems (PMS) such as Springer-Miller’s now ubiquitous SMS Host and SABRE expanding access to hotels and rental car agencies in the 1980s.
Capturing More Detailed Data About Bookings and Customers
Historical booking data has been at the core of airline and travel-related revenue optimization since the very beginning. This data includes records of previous reservations, which provide valuable insights into overall demand patterns based on several factors:
Booking rates provided high-level information about how flights or rooms were booked during specific periods. This helps airlines and hotels forecast future demand. No-show and cancellation rates monitored trends related to customers who cancel or fail to show up. This provides critical data to adjust inventory, overbooking strategies, and pricing.
Flight patterns identified popular routes and peak travel times. This information allows airlines to optimize seat pricing and availability for high-demand flights. In the broader travel industry, hotels also adopt this approach by capturing data on occupancy rates, length of stay, and lead times for bookings, allowing them to tailor room availability and pricing strategies based on patterns observed in past reservations.
Moving into the 1980s, the widespread availability of computerized reservation systems in both airlines and hotels marked a significant shift in data collection and analysis. These systems recorded every booking transaction in real time, enabling companies to store, track, and analyze large volumes of data. Reservation trends—tracking bookings across time—provided insight into periods of high and low demand, enabling better forecast accuracy. This is an advanced approach to early booking rates, aided by more sophisticated software and larger datasets to enable more detailed analysis of reservations.
The ability to capture more detailed data across time meant that customer histories and demographics became more extensive. Collecting passenger or guest information provided airlines and hotels with the ability to “segment” their customer base, identify key target groups, and tailor pricing and offers to specific audiences. The key insight leading to the early success of SMS Host was “to build a complete PMS around the guest history profile instead of the more typical rooms-based approach.”
As the airline industry shifted from direct phone sales and travel agent bookings to web-based bookings, the ability to collect even more direct data about customers exploded. Analyzing online search data, booking times, and the decision-making process allows companies to identify the factors influencing booking decisions and adjust their pricing accordingly. Understanding how users navigate an airline’s or hotel’s website enables optimization of the booking funnel, ensuring that customers are offered the right product at the right price and increasing conversion rates.
Today, airlines capture many orders of magnitude more data about bookings and customer preferences than Littlewood had available in 1972. The standardized IATA Airline Industry Data Model (AIDM) used across the industry contains over 60 fields just for its Passenger Bookins dataset and over 40 fields for route-specific data. Each airline keeps additional proprietary datasets of booking, loyalty program, and demographic data that it can use for analysis and optimization.
While the details of these datasets may be closely kept, it’s possible to extrapolate the details captured from products offered by third-party data brokers. For example, Techsalerator’s Airline Passenger Data Products is “primarily collected from airlines and their associated systems, including reservation systems, passenger service systems (PSS), and customer relationship management (CRM) platforms. These systems capture data during the flight booking process, check-in procedures, boarding, and other interactions with passengers.” Their People Lifestyle and Behavior Data alone contains over 100 fields of data about individual customer lifestyle and purchase details.
The War for Customer Data
Direct online sales provided improved access to customers and data but came with some new challenges. The rise of online fare aggregation sites such as Expedia, Kayak, and Skyscanner significantly impacted the way airlines and hotels collect data about customer bookings. These platforms act as intermediaries, allowing consumers to compare prices across various providers and make bookings without directly interacting with the airline or hotel’s booking system. As a result, airlines and hotels lose access to valuable data about booking behavior, preferences, and demographics.
To counter this shift, airlines and hotels have increasingly turned to loyalty programs as a way to regain direct relationships with customers. Frequent flyer programs and hotel rewards memberships incentivize customers to book directly with the company by offering points, discounts, and exclusive perks. Through these programs, airlines and hotels regain the ability to capture direct, detailed information about their members, including travel patterns, preferences, and frequency of bookings. This first-party data is crucial for enhancing revenue management strategies and delivering a more tailored customer experience.
Airlines leverage their loyalty programs, websites, and branded credit cards as powerful tools to capture detailed customer data. Websites and apps collect browsing behavior, booking preferences, and spending patterns over time, as well as responses to advertisements and personalized offers. Branded credit cards add another layer, offering insights into non-flight purchases such as dining and retail spending. This rich dataset feeds into advanced data models, enabling the prediction of customer preferences and fine-tuned marketing strategies.
Instead of sending travelers to a hotel website or third-party service, airlines position themselves as one-stop travel hubs. They can cross-sell or bundle these services with flights, creating seamless booking experiences that further increase customer loyalty. And rather than handing off both booking revenue and customer data to Expedia, Kayak, or Skyscanner, the airline itself captures both a higher share of the traveler’s overall spending and a wider data model of overall travel preferences from which it can further tailor future offers and optimizations.
Data Privacy Concerns for the Travel Industry
The growing reliance on data for revenue optimization in the travel industry has brought with it significant concerns regarding data privacy. The collection and use of large amounts of customer data, which fuels modern pricing, inventory management, and demand forecasting models, must now be balanced with the responsibility to protect consumer information. Several key data privacy issues have directly impacted the data available for revenue optimization efforts in the travel industry.
One of the most significant impacts on data collection and usage practices has been the introduction of stringent data privacy regulations globally. These regulations aim to safeguard consumer privacy and ensure that companies handle personal data responsibly. Major regulations include General Data Protection Regulation (GDPR), GDPR is one of the most comprehensive data privacy regulations. It mandates strict guidelines on how personal data is collected, processed, and stored.
For airlines and hotels operating in or with customers from the EU, GDPR compliance means informed consent, the right to be forgotten, and data minimization.
Detailed data about customers’ search and booking behavior, travel preferences, and loyalty program participation are valuable for segmentation and personalization. However, if customers choose not to share this data or if they opt out of tracking mechanisms like cookies, airlines and hotels lose access to this important revenue optimization input.
Historically, airlines, hotels, and third-party vendors such as travel agencies have shared customer data to improve services and optimize pricing. Data privacy regulations, customer concerns, and competitive forces now impose limitations on these practices. For example, GDPR’s data-sharing rules require companies to justify the need for shared data explicitly and to secure explicit consent from customers, which complicates data flows and limits collaborative revenue optimization efforts.
The traditional model of airline revenue optimization has long revolved around maximizing yield on different seat categories. Airlines would predict demand for various seat classes, then allocate seats accordingly, pricing them to maximize revenue. This approach relied on increasingly larger datasets of historical data and demographic insights to determine how many seats of each category should be offered and at what price point to achieve the optimal balance between capacity and demand.
The modern revenue optimization model has evolved significantly, expanding beyond simply pricing airline seats. Today, airlines also offer a broad range of unbundled additional services and products to generate revenue. When booking through an airline’s website, customers are now presented with an integrated experience that includes hotel accommodations, car rentals, and even restaurant reservations. This diversification of the product line allows airlines to generate additional streams of revenue by packaging travel-related services alongside the flight itself, thus creating a one-stop shop for travelers.
In addition to these ancillary services, the concept of what constitutes a product for sale during the flight itself has transformed dramatically. Rather than viewing the airline seat as a single purchase, airlines have unbundled aspects of the individual flight experience into distinct, purchasable items. For instance, customers can now opt to buy meals or checked luggage separately. Seats themselves have become more varied, with airlines offering different options such as extra legroom or “premium economy,” priced according to demand. Other services, such as in-flight Wi-Fi, entertainment, or priority boarding, are also priced independently, allowing airlines to offer specific pricing for each feature based on customer preferences and market demand.
This shift in product offering is closely related to the “Free to Play” model seen in the gaming industry, where the core experience is available at a base price (or even for free), but additional features and enhancements can be purchased à la carte. In the airline industry, this approach allows for greater flexibility and personalization. Customers can pick and choose which elements of the travel experience they wish to pay for, and airlines can adjust pricing for each service dynamically based on real-time demand. By re-bundling options into personalized offers—such as flight and hotel packages or discounted car rentals with flight bookings—airlines can appeal to a broader range of customer preferences while encouraging customers to make higher-value purchases.
The evolution from the basic, seat-based pricing models of the 1980s to today’s highly diversified and complex pricing strategies is remarkable. Where airlines once operated with a limited set of parameters for determining seat prices, the modern approach involves hundreds of factors that influence both seat prices and the pricing of a wide array of ancillary services. Advances in technology, big data, and real-time analytics allow airlines to fine-tune their offerings to meet the precise needs of their customers, capturing more value from each flight than ever before.
Ready for Takeoff
Revenue optimization in the airline and travel industries has evolved from relying heavily on limited historical data and basic forecasting models to employing sophisticated real-time data, external market information, and advanced machine learning algorithms. The increasing ability to leverage diverse data sources, including booking data, competitive information, and customer behavior analytics, lets companies make informed, dynamic decisions that optimize pricing and maximize revenue. As technology continues to advance, the role of data in revenue management will only grow, providing even more precise and effective optimization strategies.
Improvements in algorithms and changes to the product mix are the other components of a revenue optimization strategy. Revenue optimization in the airline and travel industries has evolved from relying heavily on limited historical data and basic forecasting models to employing sophisticated real-time data, external market information, unbundled products, and advanced machine learning algorithms (In particular, while I haven’t focused on algorithms in this article, but papers like Dynamic Pricing for Airline Ancillaries with Customer Context are fascinating.)
At every step along the timeline of revenue optimization, new technologies have enabled businesses to collect more detailed data and create more sophisticated pricing and inventory management models. Moving from manual record keeping to CRS and PMS applications allowed airlines and hotels to track bookings and cancellations, providing foundational data for early forecasting models. Over time, advancements in data storage and analytics made it possible to analyze larger datasets and incorporate external factors, such as competitor pricing and market demand, into revenue management decisions. From the first relational databases to today’s cloud-based data lakes, more information enables businesses to personalize offerings and maximize profits with unprecedented accuracy.
In future posts, we’ll continue exploring the history of revenue optimization in games, focusing on the complicated and shifting landscape in which game and digital entertainment publishers must operate and highlighting the unique challenges and opportunities in pursuing digital revenue optimization strategies.
In previous posts, we introduced the concept of revenue optimization, a strategy that has reshaped industries by harnessing data-driven decision-making to maximize profits. We traced its origins and success in various markets, highlighting characteristics that make traditional techniques particularly effective. For example, in our post “The Origins of Revenue Optimization,” we explored the foundational principles and key milestones in the evolution of revenue management across industries like hospitality and airlines. Our subsequent article, “A Brief History of Revenue Optimization: Airline Fares and Price/Time-to-Flight Curves,” examined how airlines have fine-tuned their pricing models, demonstrating a clear fit for revenue optimization in markets with predictable demand and strong historical data.
However, as any gaming industry veteran might argue, the landscape of mobile gaming is a different beast altogether. Unlike the more traditional sectors where revenue optimization has found its foothold, the mobile gaming market is marked by rapid evolution, high variability in user behavior, and a fiercely competitive environment. And that rapid evolution is occurring along three dimensions: the underlying technologies, the phone and market infrastructure, and gameplay/game mechanics. The phones keep changing, the app stores keep changing, the business models keep changing, and the definition of what a game is keeps changing.
Naturally, a gaming-oriented reader might wonder: “How could these strategies possibly apply to the chaotic world of mobile gaming?”
This post addresses that question by taking a look back at how revenue optimization took shape within mobile gaming. We’ll explore how the industry was on the verge of adopting sophisticated techniques similar to those in other sectors, only to be derailed by several major events:
The onset of COVID-19.
Apple’s sweeping changes to the advertising ecosystem.
The increasing importance of compliance with privacy-oriented regimes.
COVID-19 upended market dynamics and intensified the race for market share, while Apple’s policy shifts effectively dismantled the advertising infrastructure that many gaming companies relied on for revenue generation and profitable distribution. An increasing focus on data privacy and emerging compliance requirements restricted the uses of data, particularly the ability to share datasets with third parties.
However, this is part of a larger exploration of revenue optimization in games, so let’s first step back and take a higher-level view of the emerging mobile gaming and ad network environment of the 2010s and what that meant for game optimization.
A High-Level Overview of Optimization Areas for Games
The mobile gaming industry is a complex ecosystem with multiple avenues to optimize for revenue and player experience. Each optimization area has its own specialists and tools, catering to distinct but interrelated aspects of game success.
At a high level, we can summarize six key areas for optimization in games:
Advertising
In-Game Ad Monetization
Pricing and Offer Management
Retention and Engagement
Re-engagement and cross-promotion
Live Operations (Live Ops)
Advertising optimization is all about discovery. With millions of games vying for attention, getting more and better users into your game is crucial. The focus here is on identifying the most effective channels and strategies for acquiring new players. This involves everything from crafting compelling ad creatives to targeting the right audience segments to analyzing ad performance to maximizing return on ad spend (ROAS). The goal is to bring in users who are likely to both install the game and engage with it long-term.
Once players are in the game, the next challenge is in-game ad monetization. This strategy focuses on deriving revenue from players via ads. Ad monetization is particularly relevant for games that don’t have in-app purchases (IAPs). In-game ad monetization involves strategically placing ads within the game that maximize revenue without detracting from the player experience. Tools like rewarded video ads, which offer in-game rewards in exchange for watching ads, are commonly used to achieve this balance.
It’s critical to understand the progression from the rise of ad networks starting in 2010-2011 to the seismic change that followed. Initially, ad networks enable developers to monetize their game by selling in-game ad inventory while also purchasing ads to acquire new players. The introduction of player identification through mobile device attribution, particularly through tools like Facebook’s and Google’s ad networks and Apple’s Identifier for Advertisers (IDFA), made it easier to track user behavior and optimize campaigns.
The depreciation of IDFA in 2021 by Apple forced a re-evaluation of these models. Advertisers struggled with limited access to user-level data, resulting in a fundamental shift in how mobile advertising works today. This marked a turning point where game publishers had to innovate and adapt to revenue optimization in a new privacy-centric environment.
Pricing optimization is a critical area for games that rely heavily on IAP. This involves determining what to sell, at what prices, and how to present these offers to players. Pricing optimization also considers player segmentation—tailoring specific offers (either by altering prices for different classes of users, like Apple’s pricing tiers, or by creating specific products and bundles for specific groups of users). The objective is to maximize revenue per user while maintaining a fair and attractive value proposition.
Game publishers need to consider important intersections between advertising optimization and pricing optimization. For instance, games that rely on in-app purchases often do not engage in significant ad monetization, leading to a natural division in the industry between the “buy side” (games that purchase ads and sell IAPs) and the “sell side” (games that sell ad inventory). However, this divide is not absolute, and some games choose to blur the lines between these two approaches, opening additional revenue opportunities. For example, there is a general belief that rewarded video, done right, improves IAPs (by giving people small amounts of currency and “teaching” them how to play the game with money).
Keeping players happy, engaged, and progressing through the game is essential for long-term success. Retention and engagement optimization focuses on crafting experiences that encourage players to stay in the game. This is often seen as the realm of game designers, focusing on game elements such as level design, playability, reward systems, and social features. However, the same analysis and optimization strategies used for ad and pricing optimization—based on statistical analysis, segmentation, experimentation, automated subgroup analysis, and personalization—can be used to reduce churn and increase the lifetime value (LTV) of each player.
Re-engagement and cross-promotion strategies play a vital role, particularly for games that have a large player base or that operate within a broad portfolio of games and other digital entertainment products. Tactics can include targeted notifications, special offers, or personalized content designed around player interests and preferences. Special offers and personalized bundles require messaging strategies that can then also be repurposed for cross-promotion when predictive analytics suggest that it is a better monetization strategy. This is why cross-promotion is sometimes called cross-selling (in analogy to upselling). Re-engagement and cross-promotion are gaining importance as the cost of acquiring new users continues rising and publishers build investments in IP franchises. Web storefronts are becoming an increasingly important consideration for re-engagement and cross-promotion strategies.
That last optimization strategy, cross-promotion, illustrates that there are two distinct lenses through which optimization can be viewed:
Optimizing within a single game
Optimizing across an ecosystem (most revenue optimization strategies occur here)
The former focuses on maximizing the potential of an individual title, while the latter looks at strategies that might span multiple games or platforms, considering the overall health and profitability of a developer’s portfolio or even the gaming ecosystem as a whole.
Ecosystems also have accounting entailments. If you cross-promote, you’re using ad inventory in one game to promote another game. This is a great use of the inventory in the source game (modulo predictive analytics saying it makes sense). But if the two games have different P&Ls and wind up being different ledgers with different owners, you need to have a way to divide the revenue.
Live operations, or Live Ops, refers to ongoing management and updates to a game after launch, with an eye to keeping players engaged and driving long-term revenue. Live Ops involves a sustained practice around regular content updates, special events, limited-time offers, and player-driven experiences like tournaments or seasonal activities. Live Ops enables game developers and publishers to respond to player behaviors in near-real time. Emerging Live Ops strategies employ remote configuration and the concept of “touchpoints,” which allow adjustments in optimization strategies out-of-band from game releases, transforming games into continuously evolving services rather than one-time purchases.
In future posts, we’ll dive deeper into each of these optimization areas, exploring how they evolved, the tools and techniques that have been developed, and the impact of recent market disruptions on their effectiveness.
In this post, we’ll explore ad monetization in games by reviewing the first major chapter in mobile game revenue optimization: the rise of free-to-play (F2P) games and how they reshaped the gaming landscape. We’ll examine how F2P models fundamentally shifted how publishers monetized games, relying less on upfront purchases and more on ad-based revenue. Alongside this shift came the rise of ad networks and the use of data analytics to optimize revenue. In an upcoming post, you’ll see how these ad-based strategies, after initial success, eventually faced significant challenges.
The Rise of Free-to-Play Games
Starting in 2010, F2P mobile games revolutionized the mobile gaming industry, shifting the business model away from the traditional premium content approach. Before this transformation, game developers relied on selling their titles upfront, with each player paying a one-time fee to access the full game. This model was simpler in some respects—if advertising helped acquire a player, the game developer immediately recouped their investment through the purchase. Developers didn’t need massive audiences to generate revenue because every paying customer was valuable.
The trend toward F2P mobile games was foreshadowed by the success of browser-based Facebook games. If you look at the top five Facebook games in 2010, you see the charts dominated by Zynga, which later became a mobile powerhouse. “The saying goes that all good things come to an end,” notes the YaninaGames team in The Rise and Fall of Old Facebook Games, “and this happened to these popular Facebook games eventually. Back in 2015, there was a revolution in the gaming world. One reason for the fall of old Facebook games was due to the rise of mobile games.”
The rise of F2P was transformative for mobile gaming, marked by rapid technological advancements, widespread smartphone adoption, and innovative game design. Early mobile games were relatively simple, often leveraging the novelty of touchscreen interfaces and the ubiquitous nature of smartphones.
Titles like “Angry Birds” and “Fruit Ninja” exemplified this era. The rapid user acquisition of mobile game titles like Angry Birds and Fruit Ninja was driven by their simple mechanics and quick play sessions, which appealed to a broad audience. These games required minimal learning time and provided instant gratification, making them addictive and easy to pick up and play. This simplicity allowed these games to attract millions of players globally, turning into billion-dollar franchises. The economic model of these games shifted towards free-to-play (F2P) with in-app purchases, monetizing through massive user bases despite only a small percentage of paying users.
By the mid-2010s, games like “Clash of Clans” and “Candy Crush Saga” introduced more intricate gameplay mechanics, social features, and monetization strategies, laying the groundwork for new revenue models.
The rise of F2P games turned the game monetization model on its head. In a typical F2P game, around 98% of players don’t spend any money. Only a small fraction of users makes IAPs or pays for premium content. As a result, the success of an F2P game depends on reaching vast audiences, hoping that even a small percentage of them will convert into paying customers. This business model brought its own challenges—acquiring players is no longer directly tied to immediate revenue, and developers must keep their audiences engaged long enough to eventually convert non-paying players into spenders and keep them engaged to leverage repeat purchases.
Since at least 2013, the top-grossing mobile games have been overwhelmingly dominated by F2P titles. Games like “Clash of Clans,” “Candy Crush Saga,” and “PUBG Mobile” have built massive user bases by offering the game for free while monetizing through IAPs. Even games that have historically been paid, such as “Minecraft,” now feature IAPs to further boost revenue, despite an initial purchase price of approximately $6.99 (in the Apple App Store). This hybrid approach highlights how deeply F2P mechanics have embedded themselves into the industry.
“When looking at the top charts [in 2015], things have moved even more clearly in the direction of F2P dominance.
The current top 100 grossing chart on iOS is 99 F2P games, and Minecraft. It has been pretty consistently like that for the past 2 years.”
A turning point came around 2011 when premium content—where players paid upfront for full access to a game—was increasingly replaced by the F2P model that now dominates the industry. This shift altered how games were monetized but also influenced design decisions, leading to the rise of gameplay mechanics built around engagement, retention, and monetization through microtransactions.
I identified this shift at Scientific Revenue, as memorialized in the following slide from a joint presentation between SuperData and Scientific Revenue. During the three years from 2012 to 2015, we could clearly see a revolution unfolding in the game economy: game developers and publishers quickly pivoted from premium-priced mobile games to a market dominated by FTP games funded primarily by advertising and IAPs. (And don’t miss the other surprise here: an expansion from primarily US/EU markets to a global consumer base!)
This shift in the mobile game industry from premium to F2P didn’t just show up in slides and leaderboards, it hit the bottom line as well. Stuart Dredge reported in an April 2014 article in The Guardian that:
Clash of Clans developer Supercell made $892m in revenues (£529.6m – so four times Rovio’s total) from its two mobile games in 2013, ending the year with a headcount of just 132 staff. Meanwhile, Candy Crush Saga developer King’s 2013 revenues were $1.9bn….
‘It’s pretty clear that free-to-play as a model monetises the best, but no matter what model you use, you have to make great games,’ Rovio’s marketing boss Peter Vesterbacka told The Guardian.”
The movement to FTP changed game publishing and distribution and even fundamental approaches to game design. But a common thread in the evolution of successful revenue optimization in mobile games has been the increasingly sophisticated use of user data and analytics.
A Common Thread: Data-Based Optimization
While the early days of buy-side and sell-side ad networks were effective in reaching large audiences, these early strategies lacked the fine-tuned data insights necessary to optimize either ad sales or player acquisition. Competition between gaming companies drove the evolution of a more sophisticated view of the marketplace. Even the early mobile blockbusters relied on deep insights into the game ecosystem and user preferences to optimize their success.
‘”We saw on the iPhone that paid content works,” Vesterbacka says. Consumers pay for the initial download and Rovio keeps the game fresh with updates. On Android, they saw that paid content wasn’t working, so went with an ad-supported model. It now earns them more than £600,000 monthly. In December they introduced the Mighty Eagle, a bird you can buy in-app that clears any level. Priced at 89p, it has been downloaded two million times and cost Rovio next to nothing.’
“In the beginning of the game you have decent amount of free gems (in-game currency, which can be bought with real money). After few hours of playing you finally run out of free gems, because you have spent all your gems to gold and elixir (two soft-currencies, which can be cheaply bought with gems). At this point your psychology about gems is already formed; gold and elixir are cheap and you have purchased those with gems. Now you need to get more gems and you would get those by purchasing via the same “Shop” where you spend all your free gems. The threshold to purchase first gems with real money is low. Clash of Clans is optimized for the first purchase with real-money.”
Developers increasingly relied on the sophisticated targeting offered by platforms like Facebook and Google to optimize user acquisition and retention. Facebook called this App Event Targeting (AEO) and Value Optimization (VO). Google had its own Universal App Campaigns (UAC). The idea was interesting: aggregate certain common events and player metrics across games and then allow game marketing teams to optimize for players who tend to do those things in other games.
Aggregating players with similar behavior in all games lowered the bar in terms of the amount of traffic required for any one game to be able to reach the critical mass to optimize with confidence. In fact, a new game doesn’t have to have any traffic at all (yet) to benefit from this. The Facebook and Google platforms could then charge a higher price to send these users to games since they are theoretically more valuable. User acquisition teams got good early results with these tools and bet hard on them.
The downside of this approach was that Facebook and Google were optimizing for outlier behavior, not normal human behavior. This made a good “jump start” for early UA results, but the good results were typically not sustained for more than a few weeks unless supported by other best practices. Teams that spent their entire marketing budgets this way, rather than diversifying their spending, couldn’t sustain the results because they lacked the data and tools to build a long-term strategy around.
The rise of more carefully constructed predictive LTV models allowed developers to estimate a player’s potential revenue contribution early, enabling real-time ad campaign optimization. This shift towards using user data increased advertising efficiency and reduced fraud by allowing developers to filter out low-quality traffic.
Overall, the successful monetization of mobile games relied heavily on leveraging rich user data and advanced analytics to create personalized, efficient, high-return advertising strategies. Mobile game revenue strategies leaned heavily on data-driven optimization. Developers started using data analytics to understand player behavior, preferences, and spending patterns. Players could be segmented based on their activity, spending behaviors, and engagement levels—even device and location. Personalized offers, initiatives, and cross-promotions could be tailored to specific player segments.
Revenue Optimization strategies were an important part of the mobile F2P landscape even in the early days of naive implementations. As time passed, better data and better algorithms made it possible to do more sophisticated and effective revenue optimization. But they also enabled better measurement (and thereby enabled faster progress because it became easier to tell which techniques worked best for which populations).
Next Steps
The rise of F2P games transformed mobile game monetization. The shift from premium, paid-up-front games to a F2P model opened the door for a wide range of new revenue strategies, with advertising and IAP at the forefront.
Over the past decade, successful mobile game publishers have increasingly relied on sophisticated data analytics to optimize these revenue streams. In future articles, I’ll take a closer look at the movement from simple ad networks to advanced predictive models developed by platforms like Facebook and Google. Leveraging user data has become a cornerstone of modern mobile game monetization. But new challenges, including Apple’s deprecation of IDFA and a growing emphasis on user privacy meant the industry would face an entirely new set of challenges.
The future of mobile game revenue optimization will depend on how effectively companies can adapt to these new constraints while continuing to use data-driven insights to drive user engagement and monetization. In future posts, we’ll dive deeper into the areas mentioned, from in-game ads to Live Ops, and discuss how they can be combined to provide a more complete perspective on revenue optimization.
In the competitive airline and travel industries, revenue optimization is critical to maximizing profitability. This strategy, commonly referred to as Revenue Management (RM), uses advanced analytics to predict consumer behavior, optimize product availability, and adjust prices to the ideal levels at the right times.
The concepts of dynamic pricing and revenue optimization are not limited to the airline and travel industries. They are also crucial to industries including gaming and digital entertainment. Just as airlines adjust prices based on demand and timing, game developers can optimize pricing for digital content and in-game purchases by understanding consumer behavior and market trends.
In a previous blog post, The Origins of Revenue Optimization, we explored the early practices of revenue optimization in the airline industry, where dynamic pricing and sophisticated forecasting models revolutionized how tickets were priced and marketed. Other travel-related industries have since adopted this approach to optimizing revenue. The evolution from simple pricing strategies to complex algorithms that consider a multitude of variables, such as consumer behavior and competition, is now used across multiple sectors.
In this article, we’ll examine one key aspect of the strategy used by airlines for pricing fares: price/time-to-flight curves map changes in ticket prices as the departure date nears. These curves demonstrate a simplified version of the algorithms that airlines use to strategically price tickets to maximize revenue while filling as many seats as possible.
Understanding Price/Time-to-Flight Curves
In the context of airline ticket sales, price optimization involves adjusting the price of tickets over time to maximize revenue. Price/time-to-flight curves are graphical representations that illustrate how the price of an airline ticket changes as the departure date approaches. Here’s a broad-brush breakdown of what these curves represent—frequent fliers and expert fare chasers will know that, in practice, there is more complication to actual airline seat categorization and pricing practices.
The early booking period typically presents initially low prices. Right after the tickets are released, prices are usually low because airlines target price-sensitive customers who plan their trips well in advance to secure their seats at the lowest expected price.
The mid-booking period adopts price stabilization. During this time, prices fluctuate moderately as airlines gather data on demand and adjust prices accordingly. If demand is lower than expected, airlines might keep prices lower to stimulate sales. Conversely, if demand is high, prices may increase.
The late booking period features gradual price increases. As the flight date approaches, prices generally increase as available seats become more limited. Airlines capitalize on last-minute travelers who are often willing to pay higher prices for convenience or out of necessity or business travelers who are generally less price sensitive. Occasionally, if there are many unsold seats, airlines might offer last-minute discounts to fill the plane, resulting in a brief dip in prices before the final surge.
Prices are frequently at their highest on the day of departure, which targets travelers who need to book a flight urgently.
Factors influencing the curve include demand forecasting, competition, seasonality, and events. Airlines use historical data and current booking trends to forecast demand and adjust prices. Prices can be influenced by the pricing strategies of competing airlines on the same route. Prices may vary significantly depending on the season (holidays and prime vacation months) and specific events (sports events or conferences).
However, while these examples cover the general case, each flight’s fare pricing over time may move in different directions due to dynamic attributes of the prior bookings, connecting airports, time of year, and other factors.
Adding Some Real-World Complexity to Dynamic Fare Pricing
So far, we’ve just looked at a generalized picture of price/time-to-flight curves. In the real world, dynamic pricing for airline tickets involves numerous data points and factors that help airlines maximize revenue while efficiently managing seat availability. Key elements influencing pricing strategy and price/time-to-flight curves include the airline’s internal seat or fare classes, frequent flier programs, competitor pricing, customer segmentation, overall economic conditions, and real-time data analytics.
Seat or fare classes are categorized into different types, such as economy, premium economy, business, and first class, each with its own pricing strategy based on demand, amenities, and target customers. Airlines dynamically manage the number of seats available in each fare class, increasing or decreasing availability based on booking patterns. Different fare classes have varying levels of flexibility, with more expensive classes offering benefits like refunds or changes, leading to price differentiation within these classes.
Frequent flier programs incentivize loyalty by offering points or miles that can be redeemed for flights, upgrades, or other benefits. Airlines set aside a certain number of seats for redemption with frequent flier miles, and the availability and points required can change dynamically. Additionally, frequent fliers often have access to exclusive discounts or early booking windows, which can affect overall pricing strategies.
Understanding when different customer segments book flights is crucial for tailoring pricing strategies. Leisure travelers often book in advance and are price-sensitive, while business travelers book closer to the departure date and are less price-sensitive. This results in different price adjustments over time. Airlines analyze past booking data to forecast demand and adjust prices accordingly, using machine learning models to predict booking curves and optimize pricing.
This comprehensive approach highlights the airline industry’s complexity and sophistication of modern revenue management strategies. By integrating these factors into their dynamic pricing models, airlines can optimize their revenue while efficiently managing seat availability and responding to market changes.
An Example of a Price/Time-to-Flight Curve
A price/time-to-flight curve graphs a predicted or actual airfare over time, where the x-axis represents the time to flight (days, weeks, months) and the y-axis represents the price.
Based on our theoretical model for airfare pricing so far, consider the following possible pricing curve for seats on flight #123, where the price is reflected on the Y-axis against time-until-flight on the X-axis:
Is this correct? The answer is, of course, “it depends.” To truly understand the proper pricing curve shape, you’d need to consider various factors:
Historical Fill Rates: Does this flight tend to fill up historically? Understanding past trends helps set a baseline for expectations. If the flight typically sells out, higher prices can be sustained longer.
Booking Patterns: When do travelers usually book their seats? If data shows that most bookings occur well in advance, the pricing strategy might involve higher early prices with occasional discounts to stimulate early sales, followed by a steady increase as the flight date approaches.
Priors and Trust: How reliable are these historical patterns? Do we trust these priors, or is there significant variability? The pricing strategy can be more confident if the historical data is consistent. If not, more dynamic adjustments may be needed.
Deviations and Covariates: For times when historic booking curves deviated from the “norm” (whatever that is), what covariates can be identified that correlate to those deviations, and which of these are “in play” with regard to the current flight bookings? Key covariates include:
Time of Year: Seasonal variations can significantly impact demand. Holidays, school vacations, and other seasonal factors can cause spikes or drops in bookings.
Events at the Origin or Destination: Local events, such as festivals, conferences, or major sports events, can drive up demand and allow for higher pricing.
Alternative Destinations: If other destinations are more appealing due to events or promotions, demand for this flight might decrease, necessitating a different pricing approach.
Day of the Week: Business travelers typically book more on weekdays, while leisure travelers might prefer weekends. Understanding this mix can help set optimal prices.
Economic Factors: Broader economic trends, both global and local, impact consumers’ willingness to spend on travel. During economic downturns, more aggressive pricing might be needed to stimulate demand.
There are many others, some of which are nearly impossible to anticipate.
Understanding these curves helps airlines optimize their pricing strategy to maximize revenue while ensuring high seat occupancy. By continually analyzing and adapting to the factors mentioned, airlines can refine their dynamic pricing models to better match supply with demand, ultimately leading to a more profitable operation.
Real-world pricing tends to be much more variable, sometimes demonstrating price/time-to-flight curves much different than our theoretical model. Some of the frequent flier blogs provide examples of airfare time analysis that demonstrate the complexity of predicting dynamically priced fares. For example, godsavethepoints.com shows pricing for a flight to Hawaii that peaks just before day of flight, as predicted by our simple model.
What can you take away from this? Revenue optimization is challenging and encompasses a wide range of variables and strategies.
Applying the Lesson of Price/Time-to-Flight Curves to Digital Entertainment
Airlines maximize their revenue by adjusting prices dynamically based on factors such as demand, seat availability, and booking patterns while efficiently managing seat occupancy. This method is relatively straightforward—it simply tries to adjust to the optimal overall revenue for seats on a flight, seats that expire once the flight pulls away from the gate.
In contrast, the revenue optimization challenges faced by the game and digital entertainment industries are much more complex. The revenue streams are more diverse and influenced by a wider range of variables, including user engagement, in-game purchases, subscription models, and content updates. Unlike airline ticket sales, where the product is a single, non-repeatable purchase, digital entertainment products require tailored strategies for multiple monetization points.
Even within airline fare price optimization, the price/time-to-flight curve represents only one part of a much more sophisticated system. Advanced revenue optimization for airlines incorporates machine learning algorithms, real-time data analysis, and complex forecasting models to adjust prices dynamically. Factors such as loyalty programs, ancillary revenues from baggage fees and onboard services, and strategic partnerships also play critical roles in the overall pricing strategy.
Revenue optimization for games and digital entertainment involves a significantly more complicated array of data inputs and considerations. These include user engagement metrics, audience profile segmentation, in-game purchase behaviors, subscription models, content updates, and a constantly evolving market landscape.
In future posts, we’ll examine the history and background of revenue optimization in games, focusing more on the complicated and shifting landscape in which game and digital entertainment publishers must operate and highlighting the unique challenges and opportunities in pursuing digital revenue optimization strategies.
(To learn how Game Data Pros can help you optimize your games, contact us)
The legendary newsman Scoop Nisker, when confronted by people who didn’t like his reporting, would tell them, “If you don’t like the news, go out and make some of your own!”
Inspired by Scoop, we did exactly that. We didn’t like that there were very few revenue optimization talks at GDC and that, in general, the idea of “Game Revenue Optimization” doesn’t seem to get much, if any, mindshare at conferences.
So, instead of grousing, we organized the first annual Game Revenue Optimization Mini-Summit. We rented the American Bookbinders Museum (a really amazing space close to Moscone), ordered a few thousand dollars worth of crab cones and coconut macaroons, invited the world, and put on a show.
And what a show it was!
First and foremost, we had a set of world-class talks
Talks began at 2:15. We had four speakers.
Our CEO, Bill Grosso, opened the show with a presentation about recent developments in Web Storefronts entitled “Setting the stage for Game Revenue Optimization.”.
Then Brett Nowak, CEO of Liquid and Grit, presented “The Future of Monetization” — a talk about new and interesting monetization techniques that his team has recently written in-depth research reports on.
Joost Van Dreunen, SuperJoost, followed Brett and presented a sweeping overview of the gaming universe entitled “Life after live services: The next era in game marketing and monetization.”
And Julian Runge closed the presentation part of the day by presenting both a theoretical framework and a set of illustrative case studies entitled “Building the Basis for Monetization: Optimizing Player Engagement.”
Speakers clockwise from upper left: Bill Grosso, Brett Nowak, Joost Van Dreunen, and Julian Runge.
Second, we had an amazing audience
We were slightly nervous about this — GDC has a reputation for “One hundred people registered. Eight showed up.” In our case, 170 people registered, we issued 140 tickets, and 80 people showed up (which, since the room only held 100 people, was perfect). The audience included, among other luminaries, at least 4 PhD’s in Statistics and the entire staff of the Game Economist podcast (note that the Game Economist did a special GDC podcast including an interview with Julian as he headed to the airport— give it a listen and hear him school the economists on the true value of cross-promotions).
Third, the happy hour was delightful
“That was fun! I thought the crowd was really engaged and knowledgeable. And it sure was nice to spend a few hours away from the circus to have a more functional networking event.” — Joost Van Dreunen
The combination of the incredible speakers and the amazing audience meant that the happy hour was more than just an excuse to have reuben croquettes and duck mole tacos while downing plastic glasses of red wine. The food was good, but the conversations were excellentand lasted until the museum closed.
