Guest post by Julian Runge, Assistant Professor at Northeastern University and Principal Scientist at Game Data Pros

I went out and asked game and app developers how they price their starter packs, and to find out why they do so. Read on to learn about the results. Good starter packs are indispensable for effective freemium monetization. But, in the heat of marketing and live operations, too strong a focus on conversion can lead you to price your starter packs too low. Carefully designed experiments and personalization strategies can be a solution!

Starter packs are an essential component of the free-to-play monetization toolbox. The idea is to make players an attractive offer to get them to become paying customers quickly. I have long had a suspicion that starter packs get priced too low in the day-to-day struggle for conversion. Personalization at the starter pack stage is difficult as there is virtually no behavioral data available. So, it’s just too appealing to convert as many players as possible as early as possible – which commonly happens by making them an appealing offer at a low“no-brainer” price point.

To understand if my suspicion has merit, I recently distributed a survey on the Deconstructor of Fun slack, the Mobile Dev Memo slack, and on LinkedIn (the complete survey with results can be found at the end of this post). My aim with the survey was to find out how people price their starter packs and, if true, why they may price them too low. 54 people working on free-to-play and freemium apps participated. Out of those, 16 were product managers / designers, 14 data scientists / analysts, 14 marketing managers, and 10 respondents chose “Other” as a response.

Almost everyone (96.3%) believes that promotional offers are an essential strategy in free-to-play monetization. So far so good. What really surprised me: 96.3% of respondents say they would set a price below $10 for a starter pack, and 59.2% would set a price below $5. These price points seem low given that promo packs of $99.99 and more are commonly part of effective free-to-play price promotion schemes.

Now, what might lead people to price their starter packs so low? When asked to choose between focusing on “an attractive offer to get [users] to make a purchase quickly” versus “an offer at the highest possible price [users] are willing to pay”, 85.2% of respondents choose the former. A clear majority of 61.1% further favors an offer “at a price lower than what a user is willing to pay”, and less than 40% deem it “crucial to not set too low a price”. There may hence be a widespread belief that once a player has made a purchase, it will be easy to sell them more and bigger packs. I however believe this thinking to underestimate the power of anchoring in strongly habit forming products such as apps and games (more on that here or here). In these settings, players’ initial purchase might majorly (co-)determine their future monetization behavior.

The speed and asymmetric information of day-to-day operations can introduce biases and misconceptions in managerial decision-making. Common occurrences are risk aversion and myopia. As applied analytics often don’t reach beyond simple descriptive ratios, averages, and aggregates, a scenario with comparatively higher observed (not predicted!) conversion may provide a sense of lower risk. To get an indication if this may be happening in practice, the survey asked respondents for their preferred monetization configuration (assuming advertising revenue is the same) among the following options:

• A. Average lifetime value of users in the app is 10 USD; 1% of users spend 1000 USD each.

• B. Average lifetime value of users in the app is 10 USD; 10% of users spend 100 USD each.

• C. I’m indifferent between both options.

The risk neutral option is C, but only 13% of respondents choose it. 5.5% choose A. And 81.5% choose the risk averse option B. It is sensible that B would be the preferred option as, without further information on the distribution or on longer-term outcomes, revenue seems to be supported by a substantially larger number of paying customers. I believe that this “simple analytics fallacy” can help explain why low-price starter packs are preferred: Low-price packs drive up the number of paying customers and give a sense that revenue is more robust – even if the lower price adversely impacts repeat purchase and longer-term monetization behavior among high-value users. 

A number of factors may further amplify the presence of such a bias in favor of conversion in free-to-play operations specifically:

• The intensity of day-to-day performance marketing where early conversion of cohorts is generally considered a major signal of quality.

• The pressure of live operations where conversion is also a major signal of success.

• The difficulty of personalizing player experiences right after app download when no player behavior data have been collected.

Take all these factors together, and you have the perfect setup to end up pricing your starter packs too low. When driving up conversion becomes the sole aim, $2.99 offers get sent to players who would actually be more interested in a large pack at $19.99 or more.

Such a bias in favor of conversion can be particularly present when companies only use simple descriptive analytics, but the importance of early conversion as a predictor can be overwhelming even in predictive analytics models. Carefully designed experiments that assess the longer-term effects of starter packs can be a solution. These experiments and their careful analysis can inform effective personalization strategies that make players an offer with the content they want, at the price they want.

The survey I ran indicates that, while 70% of respondents believe that offer and price personalization is an essential tool to succeed in free-to-play game publishing, only about half (53.7%) have tried it. Looks like there’s room for growth and optimization! Reach out if you have tried it and want to chat about it, or if you would like to learn more about how to make players offers they enjoy and cherish.

[Thanks to Bill Grosso, Chris Pierse and David Nixon for comments on earlier versions of this article, and to Unsplash for the images.]

Survey questions and aggregate responses:

Question 1: How many years have you worked with mobile apps?
Response: 7.9 years (median: 4.4)

Question 2: What types of apps have you worked on?
Response: 90.8% have worked in games; 66.7% have worked only in games. 9.3% have worked on other apps (ride hailing, news, lifestyle, social media, traveling) only.

Question 3: Did the freemium apps that you worked on offer one-off purchases, subscriptions or both?
Response: 53.7% of respondents worked on apps that combined one-off purchases and subscription(s), 42.6% on apps that only offered one-off purchases, the rest on apps that only offered subscriptions.

Question 4: How many years of professional marketing training do you have?
Response: 44.4% have zero years professional marketing training. The rest has an average of 3.2 years of such training (overall average 1.8 years).

Question 5: Which role did you work in most?
Response: 16 respondents worked mostly in product design or management, 14 worked mostly in marketing, another 14 mostly in data science or analytics. 10 respondents did not provide an answer.

Question 6: Monetizing and retaining app users can be challenging. In your opinion, to successfully monetize a freemium app’s user base, do you need to offer promotions and deals to users?
Response: 96.3% of respondents pick this over not using promotions and deals.

Question 7: When new users download an app, do you think it is more important to…
Response: 85.2%: focus on making them an attractive offer to get them to make a purchase quickly.
14.8%: focus on making them an offer at the highest possible price they are willing to pay.

Question 8: Building on the previous question, which of the following statements is more correct in your opinion:
Response: 61.1%: It is safer to sell a premium upgrade at a price lower than what a user is willing to pay – you can sell them more later.
38.9%: It is crucial to not set too low a price as a low price can impact what users are willing to pay in the future.

Question 9: If you had the choice, which monetization configuration would you prefer in an app (assuming advertising revenue is the same):
Response: 5.5%: Average lifetime value of users in the app is 10 USD; 1% of users spend 1000 USD each.
81.5%: Average lifetime value of users in the app is 10 USD; 10% of users spend 100 USD each.
13%: I’m indifferent between both options.

Question 10: Which price (after discounts) do you think is appropriate for an initial promotion, i.e., an offer to new users, in the app?
Response: <= 3 USD: 25.9%
3.01 to 5 USD: 33.3%
5.01 to 10 USD: 37.1%
> 10 USD: 3.7%

Question 11: Retention and engagement are crucial to build an app’s user base. In your opinion, when a user makes a purchase,…
Response: 18.5%: it is because they are engaged with the app.
81.5%: it is because they are engaged, and it increases their engagement with the app.
0%: it increases their engagement with the app.

Question 12: Many apps sell premium upgrades at different prices in different countries. Price personalization is the practice of setting different prices for individual users based on further characteristics, e.g., the device they downloaded the app on. In your opinion, such price personalization is…
Response: 70.3%: an essential tool in freemium apps to increase user monetization.
27.8%: more risky than useful in freemium apps.
1.9%: N/A

Question 13: Have you used price personalization beyond country-based pricing (either through promotional offers or in-app purchase prices directly) in the apps you worked on?
Response: 44.4%: No
53.7%: Yes
1.9%: N/A

Question 14: If you like, please share further thoughts:
Response: 
- offer personalization – yes; price personalization – yes, but the impact was hugely negative within the community.
- We use user behaviour to personalise the IAP prices during the special events.
- I think that it worth to an test price segments per country, even though on my experience it wasn’t successful.
- At my previous job, we had a bigger user base on Android than on iOS. But they generated roughly the same revenue. Therefore we experimented with lower prices (which wasn’t possible anyway in the App Store with their fixed price tier system) and it increased the number of buyers and later the total revenue as well.
- It can work well.
- Machine leaning can be used for classifying user types for price personalization as segmentation based tools (mainly used for adjusting prices in the aviation industry as for the case of discounters like Ryanair) might be unsuccessful in modeling the user’s preference.
- I have never seen country based pricing generate a higher average revenue per user. I don’t deem it risky, I just don’t see the value in the extra work as you get the same return. However it is a way to gain featuring, which then justifies the additional effort.
- Do not discriminate players based on weird selection, offer good value to everyone.
- I think price personalization is essential. Some demographics have less to spend (think students, 65+, families, etc) and tailoring prices to them helps retaining users. Most of these concepts are already socially accepted, so people tend to accept them (think lower prices to museums, public transportation, etc)
- We set prices uniformly for all users, in all countries.
- If you value growing a healthy community around your app (which is very important for games), my bet is that you have more to lose than gain from price personalization, simply due to the fact that players talk to each other and it’s (in my opinion rightly) perceived as unfair.