LiveOps at Behaviour: In-Game Sales
Germany’s Making Games magazine recently profiled Live Operations at Behaviour Interactive and we are republishing that series here with permission. In this third article, our Head of Live Operations, Gregory Feuz, looks at the three main elements of In-Game Sales strategy.
Gregory leads the Live Operations team at Behaviour Interactive, which consists of Product Managers, Marketing, Data Analysts, Live Event Managers and Community and Customer Support Managers. He also leads Behaviour’s Mobile Live Operations team. A 10-year veteran of the video-game industry, Gregory started his career at Atari Europe and then moved on to work for Bwin.Party before joining Behaviour in 2015.
There are five key elements required for any game to be successful: a great product, comprehensive technology infrastructure, efficient live content pipeline, great User Acquisition capabilities and Advanced Sales and Offers merchandising. Sales and Offers is the one that has the most direct impact on revenues generated during LiveOps.
At Behaviour, our In-Game Sales Strategy is straightforward and can be broken down into three main elements:
- Offers — What to offer and at what price?
- Timing — When to initiate sales?
- Segmenting/Targeting—who to target with what bundle?
Offers consists of four variables: items, pricing, discount, and creatives. Whenever we create a sale offer, the goal is to come up with the best offer possible to drive conversion and revenues. In terms of items, we ask ourselves if we want to offer one exclusive item or bundle a set of items. We also always ask ourselves if an item will provide significant value for the player who buys it because it is extremely important that the player gets a return on their investment (value versus willingness to pay). However, we also always make sure the item(s) is not game-breaking in terms of in-game progression or economy.
For Pricing, it is all about finding the sweet spot where a user will be incentivized to make a purchase. In terms of discounts, the trick is figuring how much of a discount to offer. AB testing can help but trial and error is really the best approach. We assess our sales and offers performance regularly so that we can optimize them over time. On any given game, we are running hundreds of different offers and we are constantly optimizing them.
Finally, when it comes to assets, we try to create the best-looking creatives possible because they tend to perform better (we treat sales creative as if they are an in-game creative). Our ability to AB test our offers in terms of creatives has been very helpful.
It is all about when to offer a sale/bundle to a user. We use Timing in three different ways. First, we have offers that are based on a Fixed Timing. For instance, every Friday we used to offer a “TGIF” (“Thank God it’s Friday”) limited-time offer on one of our Social Casino mobile games. This offer would always be made on Fridays at the exact same time. Knowing that our players are playing the most during the weekend, it represented a great opportunity to drive sales and revenue as they were more likely to make a purchase just before the weekend.
Secondly, we have offers based on Trigger Timing. This is especially useful for so-called “Progression Offers.” A good example is a Starter Pack offer. This pack obviously needs to come up early in the player’s journey. We usually offer it based on Level progression so that it comes after a feel-good moment and enhances the chances of conversion.
Finally, we also take advantage of the calendar. This type of offer and sale is what we call “seasonal offers.” Dead by Daylight is the game from which we have benefited the most from Seasonal Offers. As a multiplayer horror game, Halloween is the most important holiday of the year and presents a great opportunity to run special bundles.
The last lever of our sales strategy is Segmenting/Targeting. This is critical because each game has unique players with different behaviours and needs. Therefore, the ability to target users with the right offer has become extremely important. Our targeting and segmenting strategy is based on two levers: Frequency (how engaged is a user? How often do they play? What level are they?) and Lifetime Value or LTV (how much a user has already spent in the game). This is the backbone of our sales targeting. This is how we can offer the right sale at the right time to maximize conversion. For instance, a new user who has never spent time in the game will be offered a lower price point bundle while a user who has spent a large amount of money and has been playing for more than three months will be offered a totally different bundle, most likely with a higher price point.
Where it gets interesting is when we are looking at behavioural patterns in our player base. On one of our mobile games, we were trying to understand if there were any patterns that we could identify from a player’s first day playing that would point to that player becoming a highly engaged and monetizing player. We looked at all players that were still playing three months after installing the game and had spent at least X amount of money and purchased at least X number of bundles. We realized that these players had unique playing patterns on their first day, meaning they were constantly spending their in-game currencies to the maximum as soon as they started playing the game. This insight allowed us to create a predictive model which would segment these “potentially high-value, highly engaged” users and offer them specific bundles suited to their needs and aspirations. We saw a significant uptick in our game performance and KPIs improvement across the board thanks to that segmenting strategy.
The Future of Sales
Now that we’ve looked at our approach to sales, let us briefly look at how sales could evolve in the near future. Many game companies (Behaviour included) are now investing in their machine-learning capabilities (our sales strategy is still determined by human actions). We believe the next milestone is when sales are fully managed by algorithms. We envision this is as follows: The Product team will still create hundreds of special bundles but instead of these bundles being offered based on predeterministic actions decided by the Product team, the machine-learning algorithm will take over and decide which bundle is offered to the player. The algorithm will be mixing and matching two dimensions: bundle price (based on previous purchases by the player) and bundle content (what content is most appropriate based on the player’s in-game behaviour). With such an approach, AI will be entirely responsible for deciding what content at what price has the best chance of conversion. This will allow game companies to be even more efficient as it will be 100 per cent player-specific and remove the need to segment players into groups. Each player becomes their own unique segment.
Head of Live Operations, Studios