Quantitative analysis of freemium game purchases
November 28, 2011 8:48 PM   Subscribe

How to approach a quantitative analysis of consumer purchasing decisions in the freemium games market?

Freemium games, for those unfamiliar with the term, are video games that are free to play but entice the users to purchase in-game perks--$2 for a fancier sword, for example. Game companies using this business model want to structure the incentives, prices, and related issues (e.g. periodic sales) to maximize revenue, and they have a vast wealth of usage data to inform these decisions.

I don't have any real-world experience in any kind of business analytics, but I do have a solid academic foundation in econ, stats, modeling, and data visualization. Turns out this academic foundation is not as useful as one might hope; I can talk about Marx's labor theory of value but give me real-world questions like,

- What discount should we offer during our next sale?
- How can we learn from the data from our last sale?
- Here's a bunch of data about a popular but under-performing game--how can we make it more profitable?

...and I don't know where to start. To complicate this further, the freemium games market has some significant differences compared to most kinds of retail price-setting--all sorts of minutely detailed usage data are available, so one must develop more sophisticated models to make use of that; additionally, some erstwhile staple concepts like marginal cost are suddenly irrelevant.

I'm trying to learn as much about this as possible within the next week or so. I've read some wikipedia pages, read the Flurry blog, and I've skimmed through papers like this (big PDF) looking for relevant bits. My Googling hasn't turned up much else, and so now I turn to you. Metafilter, can you direct me to other things I should study? I'm looking for all of the following, with bonus points for the bolded:

- web resources, papers, book recommendations, course syllabi, or just concepts or search terms,
- on the subjects of consumer data analysis, modeling purchasing behavior, the current video game market (especially freemium or mobile games), real-world price-setting in any industry, or other related topics I should know about,
- with a focus on either applied and quantitative approaches or theory.

Anything even remotely relevant to this query is welcome; I'm happy to sift through it all, and it's quite likely I don't know the correct terms to describe what I'm trying to learn. I have a second interview coming up for a job I really want. Help me impress them! Thank you!
posted by kprincehouse to Education (3 answers total) 2 users marked this as a favorite
How Valve experiments with the economics of video games

I know nothing about economics. But a refresher in optimization algorithms might be helpful, with a side of clustering and classification algorithms. Although you could probably get away with using these algorithms as plug-and-play black boxes for the job you describe.
posted by qxntpqbbbqxl at 9:32 PM on November 28, 2011

Perhaps think of the usage data as choice data from options, where there are tradeoffs & costs. Analyze what players choose and what they avoid; what they are willing to pay. Choice-based conjoint might be used to model how people value features and willingness to pay, and create simulators to predict behavior when players are offered new options. This is normally done by setting up experiments (or questionnaires with choice tasks) but might be doable with existing data if the data can be properly structured.
posted by lathrop at 7:11 AM on November 29, 2011

This probably won't be exactly what you're looking for but I would check out Valve's Team Fortress 2. Players spend tons of money in the game's store but also in an 'underground economy' where items are bought and sold for hundreds of dollars. Pretty interesting.
posted by Funky Claude at 9:35 AM on November 29, 2011

« Older What is this strange feeling and why?   |   Please tell me I haven't missed the blogging boat!... Newer »
This thread is closed to new comments.