This article builds on previous articles I've written and talks I’ve given. In particular, it is a follow-up to my recent article on Dynamic Pricing, Personalized Offers, and Modern Gaming. That article was an extended discussion of the recent uproar around randomized price-points in Zynga’s CSR 2 and a survey of how prevalent sophisticated data-driven merchandising techniques already are in mobile gaming.
Readers of that article objected to dynamic pricing as a practice, using words like “shady” or “shameless” and claiming that dynamic pricing, if adopted widely, would destroy the gaming industry. This FAQ helps to set the record straight by providing explanations and addressing the underlying concerns associated with dynamic pricing.
What is Dynamic Pricing?
Dynamic pricing is the practice of altering prices for goods or services in real-time without altering the goods or services (e.g. “just changing the price”). Different users, or possibly even the same user at different times, will see different prices for the same good or service.
That’s a broad definition, and encompasses many different pricing strategies. For example:
Auctions. Every time someone bids, the price changes for everyone. This isn’t usually thought of dynamic pricing because the seller isn’t fully in control (the price changes depend on bid activity), but it shares some aspects and it’s hard to come up with a definition of dynamic pricing that rules out all types of auctions.
Surge pricing. Many companies, most famously ride sharing company Uber will respond dynamically to changes in supply in order to price goods and services differently. For Uber this frequently happens on stormy nights (when more people take cabs, and they are more desireable). Storms, more generally, are a great source of examples—hoardable goods, like flashlight batteries, often become more expensive. Also note that taking advantage of duress to change prices for staple goods is illegal in many areas of the world.
“Airline Pricing” Airlines are famous for charging different prices for different classes of fares, and for offering different prices at different points in time for the same seat (this article is very good). It can safely be said that the airlines, and more generally the hospitality industry, pioneered modern pricing science.
Adaptive Retail Pricing. Retail chains like Walmart and Target charge different prices at different stores. You can wander around my native San Francisco and visit different CVS Drugstores—they will charge different prices for the same good.
Online Price Testing. Ever since the pioneering work of Rothschild in the early 1970’s, retailers have yearned for a way to estimate demand and elasticity. Unfortunately, it’s very hard to do so in traditional retail environments (because price tags are hard to change effectively). With the advent of web and app-based retail, however, the field of price testing has been revolutionized. Go to, for example, Camelcamelcamel and watch how many prices Amazon changes each and every day (or go to your local university and peruse the back issues of The International Journal of Electronic Commerce).
Online Retail Personalization. The next step after price testing is price customization and personalization. In much the same way that personalizing content has been a web best-practice for a long time, personalizing bundles and offers is now a best practice in online retail (see here and here in the Wall Street Journal).
That’s a lot of examples! Perhaps too many, but the point I want to make is simply this: prices change all the time, for a wide variety of reasons. In fact, the idea of a “fixed price” that is stable and available to all buyers is the exception, not the norm in commerce.
But if you only take away one thing from this first section, it's this:
The idea of a “fixed price” that is stable and available to all buyers is the exception, not the norm in commerce.
Do Games Price Goods and Services Dynamically?
Yes, they do. Dynamic offer management is a best practice in gaming as well. My article Dynamic Pricing, Personalized Offers, and Modern Gaming covers this in depth, but the key point is simply that big data and machine learning make it possible to tailor ads, prices, and bundles to individual users with a high degree of effectiveness. This has been true in advertising for a long time (see, for example, this overview from 2010) and is now equally true in bundle management, offer management, and pricing.
In fact, dynamic pricing for digital goods inside a web-page or from an application can be incredibly effective as a general technique. This relies on two observations:
Because the price is displayed by a software application, the price can be changed “on the fly” as information becomes available.
Because the marginal cost of production of a digital good is often close to zero (bits are easy to manufacture), you can easily lower prices for users who don’t value the good or are otherwise not inclined to purchase it.
When you marry this to modern data science, you have an amazing ability to price goods to the demand curve.
Wait. You’re Telling Me this is Done Today? In Video Games?
I'm not just telling you. I’m telling EVERYONE. As an industry, we need to get past the dismissive rhetoric (“shady” and “shameless”) and apocalyptic predictions (“The result is harm to ALL F2P developers”) and move on to a discussion of when and where dynamic pricing is appropriate.
In a world where most video games lose money (especially indie games), most games studios constantly struggle to stay in business, and the cost of user acquisition reaches record highs each year, rejecting best practices from other industries is irresponsible.
But Isn’t it Wrong to Dynamically Price Goods and Services?
In some cases, yes absolutely. The classic case is usually phrased in terms of a life-saving medicine and the scenario goes along the lines of:
There are a limited number of suppliers of the medicine.
An affluent person needs the medicine.
The suppliers know that the person has no choice but to buy the medicine, and exploit the situation to charge an exorbitant amount for a single dose.
This is a classic example of “predatory” pricing (as are the examples above where batteries get more expensive during storms).
But, while these use-cases are easy to remember and dramatic, and great examples of what the philosopher Daniel Dennett has termed an intuition pump, they’re not all that common.
Consider the following example instead. A software publisher decides to sell bags of gold coins inside their game. They’re not really sure what the demand curve is, but they think it looks something like Figure 1 below
Based on this, they pick a point on the curve that seems pretty good, and pretty fair (“I’d pay that much”). They wind up with the prices and revenue illustrated in Figure 2
But then someone on their data science team realizes: there are complex predictive models that will reveal which users won’t buy at that price. And that it is possible to offer these price-sensitive users, who won’t buy at the initial price, a second and lower price. F
