Improving Video Game Recommendations: Addressing Challenges and Opportunities in E-Commerce
Have you ever browsed through what seemed like Steam’s entire catalog trying to find a game you might like? Have you spent more time on Netflix trying to decide what you want to watch as opposed to watching something? These are problems most of us have faced or are facing, and are a result of several factors, especially lack of personalization.
Personalizing your products for your customers is critical in today’s world. One can see an average increase of 20% in sales when using personalized experiences. Companies want their customers to be aware of how they are catering specifically to user needs. For example, if you’ve played a shooter game, you might be recommended to play another just because other people who liked the first game you played also enjoyed this. Your reasons for playing the former might differ from the others and thus, the latter might not be a good fit for you. The lack of good video game recommendations reduces trust in these companies to provide a good service experience.
From the companies’ end, good customer experiences help in generating income, as well as differentiating themselves from the competition. An e-commerce company needs to focus on three things to thrive:
- Increase the acquisition rate of new users
- Increasing conversion rates of your users
- Ensure that users don’t leave (reduce churn)
The Importance of Recommendations
Personalization has become a major factor in the success of e-retail companies. Whether it is addressing customers by name in communications or providing them with special offers based on their interests, online stores are increasingly focused on improving this.
Recommendations are the deepest level of personalization and are a necessary feature to be added to their portal. They are integral for both customers and the companies which cater to them for a multitude of reasons we will look at. For a customer, they provide the following benefits:
- A significantly better user experience
- A sense of being understood and seen
- More personalized benefits and deals
For companies, the benefits are just as tangible if not more so:
- Improved customer engagement
- Significantly increased customer retention
- Larger levels of web traffic
- Better sales and revenue
Below are some examples of companies that thrive based on their recommendation systems.
Case Studies
Let’s take a look at one of the world’s most successful companies, Netflix. While Netflix started as a movie rental service, today, they stream movies and have over 200 million paying customers across the globe. A key part of this evolution is their personalized recommendation system.
Netflix’s recommendation systems have been developed over years by hundreds of engineers after analyzing millions of users. When a new subscriber joins, Netflix asks them to pick shows/movies they like, and as they watch more over time, the suggestions are powered by these as well as some additional factors like:
- Viewer history
- Viewer ratings for prior shows
- Information like title, genre, category, and more
- Other viewers with similar preferences and taste
- Time an episode/movie lasts vs time duration of a viewer watching a show
- The time of the day you’re watching
- The device on which you’re logged in
Closer to home, we have Steam, which is a digital game distribution system, with more than 120 million monthly active users and a catalog of over 50,000 games. It is also home to a powerful video game recommendation system that helps gamers find games they will love.
They recommend games based on your played games, purchase history, store browsing history, and games that other players with tastes similar to yours love.
However, neither of these do a perfect job. Let’s look at why.
The Problems with Recommendations Today
We’ve looked at the importance of personalization and the role recommendations play in this. However, despite online stores realizing how vital a good quality recommendation is, they still haven’t perfected the art of suggesting the right products. Here are some of the common problems faced by customers while trying to find what they need.
Wrong recommendations: Thanks to imperfect algorithms or lack of high quality data, sites can often suggest irrelevant or incorrect recommendations. These reduce customer trust, engagement, and overall, is a waste of a good opportunity.
Impersonal communication: We all buy products and services for a variety of reasons. However, distributors still use generic and non-engaging messages most of the time while communicating with users. Messages such as “You might like Item X” without mentioning why you might like it can turn your customers off.
Choice overload: Too much choice can be a detriment to your customers. A recent consumer report discovered that more than half (54%) of consumers have stopped purchasing products from a brand or e-retailer website because choosing was too difficult, with 42% admitting to abandoning a planned purchase altogether because there was too much choice. These problems are a result of sub-optimal recommendation systems on websites.
Behavior Vs Motivation
The reason for inadequate online recommendations is that these mechanisms are primarily driven by behavior as opposed to motivation.
If several people play the same game, they might do so for different reasons. Let us take one of the most popular games which came out in June 2020, Valorant, as an example. Valorant is a 5v5 tactical first person shooter (FPS) where the characters you play as (agents) all have unique abilities. It has a monthly player base of at least 12 million throughout 2021, making it one of the most popular current FPS titles. Let’s analyze the different possible motivations that drive people to play Valorant:
Satisfying the urge to compete, dominate, and win: A large number of people play video games to compete against other skilled players and dominate the leaderboards for a sense of achievement. Valorant has this in spades with its highly competitive online multiplayer nature and detailed rank progression.
Strategizing for the win: Gamers enjoy certain games because they involve a great deal of planning and strategizing to be victorious. With its deeply tactical nature, Valorant satisfies this motivation.
To play with friends or meet people: A significant portion of players like games for their socialisation aspect. Whether it is being able to play with your buddies, meeting new like-minded strangers you can have fun with, or working as a team, Valorant fills these socialization shoes very well.
For an adrenaline rush: Gamers often get motivated by the rush of adrenaline or dopamine they get as they play games that excite their senses, and this is what keeps them coming back to the game as well. Valorant certainly fits this criterion.
Aggression: Some people like playing video games for the violence and ferocity that come as a part of the game, especially shooters and hack & slash games. Valorant satisfies this urge.
The behaviour here in common is people playing Valorant. However, as you can see, their motivations may be completely different. For instance, in terms of story and lore, Valorant is found lacking compared to Overwatch, another popular competitive multiplayer title. Thus, people who play Overwatch because they like its lore and narrative aspects might not be as interested in Valorant.
How can you Improve Video Game Recommendations?
Gamer motivations are a culmination of their emotional and psychological makeup while also covering traits like values, personality, and life situations. To revolutionize video game recommendations, you will need to start by understanding the games you’re recommending, and why people play them. Next, look at your user base and try to understand each individual at a fundamental level. Finally, once you have an understanding of the games as well as your user, see why people play what they do, and use that to provide a video game recommendation. As a result of this, you will:
- Provide fewer recommendations: This will keep you from overloading your customers with choice.
- Give better recommendations: When you understand your users’ motivations, you can suggest games that are aligned with their motivations every time.
- Personalized recommendations: Each of your recommendations will effectively communicate why a particular game is right for your user, as well as address their needs.
Apart from the above, you can improve e-retail personalization in general by:
- Refine your search pages. You can use metadata to improve product descriptions and make it easier for your algorithms to match products to customer preferences and needs.
- You can use referral bonuses to improve signups and good email marketing that conveys personalized deals and offers to your customers to increase retention.
- Ensure your home page, product pages, and promotional offers are tailored to your customers’ needs based on data you’ve collected and their preferences.
Intelligent machine learning algorithms combined with high quality data are your best friends. The next section will go into detail about recommendation models you can use in conjunction with them.
Recommendation Models
Below are the models most commonly used by e-commerce companies:
Popularity-based: These are products that are best-selling currently. For example, Among Us blew up in 2020 and was a game that popped up on Steam’s bestseller list. These also include games that have been popular for a long time, such as Counter-Strike: Global Offensive. It is meant primarily for new users on the website.
Quality based: The games which have a high number of positive reviews and ratings show up here based on this model and are recommended to users. However, this might not be the best method as peoples’ tastes can drastically differ, and a game might have ‘boosted’ reviews. Also, newer games might not have enough reviews to show up, despite possibly being something your user might love.
Content-based: This model recommends products based on their similarities with other products. It leverages the description and content of items and an understanding of the user’s consumption history. For example, Valorant is recommended to players who love Overwatch and Counter-Strike: Global Offensive, since it has similar characteristics to both these games.
Collaborative Filtering: In the newer, focused sense, collaborative filtering is a method of making automatic predictions (filtering) about the interests of a user by collecting preferences or taste information from many users (collaborating).The system generates recommendations using only information about rating profiles for different users or items.
Of course, hybrid recommendation systems which use a mix of these models are your best bet to provide personalized recommendations to your customers. Going back to Netflix, they make recommendations by comparing the watching and searching habits of similar users (i.e., collaborative filtering) as well as by offering recommendations that share characteristics with content that a user has rated highly (content-based filtering).
Metadata is crucial to fuel understanding of your products. This will help you organize your product database, as well as categorize it better. High quality and comprehensive metadata gives personalization algorithms more data to train on. If you want to know more about the importance of video game metadata and managing it, this blog might help you.
Conclusion
Personalizing recommendations is the best way for e-commerce companies to improve revenue as well as stand out among their competitors. When it comes to video games, understanding the motivations as to why people play the games they play is integral to making good suggestions. Gameopedia’s quality-checked and extensive metadata as well as our intelligent sentiment analysis tool can help with optimizing your content and website for better personalization and improving video game recommendations. Contact us to learn more about what we can do for you and your business.