Predicting your app’s monetization future

Post 2 of 2: Practical lifetime value calculations for five popular monetization models

In my first post on predictive analytics I introduced you to a simple formula for calculating lifetime value (LTV) and explain how it can be used to assist with planning customer acquisition.

Now, I’m going to take a more practical look at LTV by examining how it might be calculated for some of the most common app monetization models: premium, subscriptions, freemium, ad-funded, and hybrid. Other business models, notably retail and shopping, financial services, and apps that are used as other channels of revenue are not considered, just for simplicity. I’ve also included best practices for each business model, mainly based on conversations with different developers, on how you might improve your LTV.

Premium apps

Calculating LTV for premium apps is probably the easiest of all the monetization models. Premium apps are those in which the user pays to download the app from a store. As a result, all users who download the app are paying customers and the price of the app will be its LTV.

LTV = Price of the app

In the case of premium apps, lifetime is not included in the calculation as it’s not directly correlated to converting users into paying customers. However, lifetime is still very relevant as engagement and retention of existing users are critical to attracting new ones.

Best practices for premium apps

  • Discounts and promotions should be tested to see whether the increase in sales or conversions adequately compensates for the decrease in profit per user. Consider segmenting these tests by country or region, as not all locations may behave similarly.
  • Recommendations and reviews are essential to success. This is particularly true for premium apps, where the user will be buying an app without being able to test it. Therefore, recommendations in the app store pay a significant part in convincing users to download the app. So, figuring out when to ask users to leave a review is important. It’s advisable to wait a while after the install. Also, when asking for a review, check whether there is a Wi-Fi connection — in an Android app, for example, you can do this using the connectivity manager API. But, do not insist, doing so could easily put the user off.
  • Engagement and retention, achieved through adding new features, the onboarding experience, and similar are critical to encouraging users who eventually will become ambassadors for the product and drive word-of-mouth.
  • Localizing, regarding both price and distribution in the store and of the app itself, is essential to expanding across regions and countries.

Subscription apps

Many companies, from verticals as diverse as media and IT services, rely on subscriptions to monetize their services. The main reason why they do: subscribers usually represent a stable and constant revenue stream. “Churn“, the ratio of users unsubscribing from the service at every renewal, is critical to the success of the company.

Using the LTV equation I defined in the first post:

LTV (period) = Lifetime x ARPU

The variables are expressed as follows:

  • LTV period and Lifetime equal average subscription length. This is because subscribers usually churn at the renewal date, which is when the user has to decide whether to continue or stop using the service.
  • Average subscription length is driven mainly by the churn rate, or its inverse ratio the renewal rate (1-churn ratio). Usually, this is expressed as:
Average subscription length = 1 / churn ratio
  • Calculating LTV for a certain period is also possible, but most developers just use the average subscription length.
  • ARPU is usually the subscription fee. To calculate the monetization variable simply calculate the average price of a subscription for the period. To do so, consider all types of subscriptions or segment by cohorts (for example location, subscription length, and alike.)

The importance of churn calculation in subscriptions

The way churn is calculated will have a direct impact on the expected lifetime of a user and hence the LTV calculation. As such, many companies are extremely careful about how this ratio is calculated. For example, how many periods are taken into consideration when calculating the average churn, the last 12 months or just the last 3 months?

As the churn might vary depending on the assumptions made always consider relevant factors, such as:

  • Account for recent app changes or updates. Has new functionality reduced churn dramatically?
  • Exclude free trials or test periods. These subscriptions usually have a higher churn rate.
  • Segment users by cohorts. Usually, long time subscribers will have a lower churn ratio compared to recent users.
  • Defined milestones to calculate churn. Some developers calculate LTV only for those users who have reached the breakeven point.

The negative churn paradox

Sometimes, developers aim for a negative revenue churn. That is, they look to expand the revenues generated by the remaining users to compensate for the impact of user churn.

To illustrate this, imagine an app has two users, Max and Jakob. If Max churns from the app, 50% of the revenue churns too. However, if the remaining customer, Jakob, could be upsold to a super extra premium package, which increases the subscription fee by +120%, it would compensate for the impact of Max leaving. The result is “user churn” of 50% and a negative “revenue churn” of -20%.

Negative churn is highly sought after by subscription companies such as telcos, SaaS, or media providers. These companies constantly offer their users upgrades to current subscriptions, and these might be included in their LTV calculations too.

Best practices for subscription apps

  • Discount future cash flows where subscription services might have a lifetime longer than a year.
  • Consider cross-platform services, where content or services might be accessible on smartphones, laptops, TVs, and wearables. In this case consider assigning a value to the app based on a percentage of use, acquisition of users through the platform, or another suitable method.
  • Analyze users by cohorts, particularly analyze trends between segments. For example, the average churn of users onboarded in January compared to February.
  • Discount initial months or free trial periods as users tend to churn very quickly, and ratios might not be representative.
  • Set free trial durations with care, to maximize retention. For example, in a media app make sure the free period isn’t so long that the user can watch all the content that interests them most during the trial.
  • Use one time promotions (such as 80% off the first month) as they usually work better than perpetual discounts (such as 10% off every month) but check the bottom line impact of such heavy discounts.
  • Run price elasticity tests across regions and user types (for example students versus professionals).
  • Avoid passive churn, where the user doesn’t actively cancel the subscription, as much as possible. Expired credit cards, missing information, or insufficient funds are some of the typical reasons for passive churn.

Freemium

In freemium apps, the user downloads and enjoys the app for free but to access certain functionality or accelerate their progress in a game has to purchase digital goods. This model has become extremely popular among certain verticals, most notably gaming. The key to a successful freemium app is finding the right balance between free and paid items.

Also, while game users generally understand the digital economy (buy a life to continue playing the game), it’s more of a challenge to explain to app users the value of purchasing digital items.

For freemium the three variables from our simplified LTV equation (LTV period = Lifetime x ARPU) are defined as follows:

  • LTV period is usually calculated for 180 days, 365 days, or 2 years, although it will depend on the average user lifetime and type of app.
  • Lifetime is frequently evaluated from a retention and engagement perspective. Retention is often calculated, as I mentioned in the previous post, using survival or decreasing curves models, based on the number of days since the install for certain periods such as 1, 7, 28, 90, and 180 days.
  • Engagement will commonly include both the number of daily or monthly active users as well as the average length of sessions. Session length is particularly relevant here, as the more time spent in the app by a user the higher the chances that they will convert within the app.
  • ARPU calculations depend on the time units selected when calculating Lifetime. The most frequently used is days, and therefore the calculation could be average revenues per day divided by daily active users (DAU).

Best practices for freemium apps

  • Test and optimize the timing and content of messages and calls to action, to ensure they reach users effectively and explain the value of digital goods well. Timing may be about creating a sense of urgency (for example, an offer to expire after a certain time, such as 48 hours). Communicate value by highlighting the additional features that the in-apps purchase offers (for example, the ability to see the profiles that have checked yours recently in social and dating apps).
  • Use “sachet marketing techniques” — small-ticket in-app purchases — that might be particularly effective in emerging markets.
  • Encourage user actions that trigger retention and engagement. For example, users that create a profile during the onboarding process may be found to have a higher than average retention, and therefore it might be worth experimenting with incentivizing users to do this.
  • Calculate LTV for different cohorts, notably age or gender might have different ARPUs or lifetimes.

Advertising

Monetization through advertising is a very popular option among certain verticals and regions. It’s also frequently used in combination with other models (subscription or freemium) as it enables non-paying users to be monetized.

As with all other cases, engagement and retention are critical. The longer spent per visit, the higher the chances of showing relevant ads and improving ad targeting to deliver more value to the advertisers.

For advertising the three variables from our simplified LTV equation (LTV period = Lifetime x ARPU) are defined as follows:

  • LTV period will depend on the type of business. Some newspapers, which usually rely on advertising and other models too (mostly subscriptions), might calculate LTV for 1 year, as this allows them to compare returns on investment for both models.
  • Lifetime is usually calculated in a similar way to the freemium models, based on retention and use of the app. It may also be helpful to factor in the duration of visits, with suitable cohorts, as longer visits to the app will generate more ad requests and therefore opportunities to monetize.
  • ARPU is commonly calculated as the average revenues per daily users (also known as ARPDAU) or the average revenues per monthly users (ARPMAU).

More so than other models, it may be worth evaluating LTV across different cohorts given the variables that affect revenue, such as:

  • Sales teams: direct versus programmatic sales. Be aware of any differences in selling method, for example, CPD compared to CPM, and track back appropriately. Also, remember third parties might average CPM.
  • Ad formats used: interstitials, incentivized video, and alike.
  • Type of content: opportunities to monetize. For example, in a news app the financial sections usually have better CPM than general news sections, Video consumption may have higher CPM compared to display banners in text articles.
  • Regions or countries: the differences in price between mature and emerging markets.
  • Type of audience: gender, age of users (for example, apps for children might be subject to COPPA advertising restrictions, and therefore CPM or coverage will be significantly lower).
  • Brand: well-known brands compared to others.
  • Number of screen views: every screen view is, in theory, an opportunity to place a new banner or interstitial ad after the screen view.

Best practices for advertising apps

  • Cluster users based on loyalty. For example, use the frequency at which users open the app to create groups for daily, weekly, or monthly visitors. Advertising can then be tailored to these groups, for example, showing less aggressive ads to loyal, daily visitors.
  • Test ad positions and formats, reducing the ads that underperform. One developer found that they should reduce 320x50 ads at the bottom, as they saw these generate low CPM, but increase the bigger 300x250 ads in other positions, because they offered a better CPM.
  • Identify key moments within the app that will trigger good placements, for example, feed placements, before an action occurs, post-flow placements, and others.
  • Use multiple networks through a mediation platform, to increase average CPM.
  • Monitor the app’s ratings and reviews and watch for any negative effects from the use of advertising.

Hybrid business models

Combining two or more monetization models is a common approach. This is because most apps appeal to a range of users, who respond differently to each of the monetization strategies.

Calculating LTV in the hybrid model is a case of adding together the different LTV values for the various monetization models used. By calculating the LTV for each model, the models can be compared and the right solution for each user can be determined. While changing behavior, such as convincing a user that never purchases in-app purchases to do so, is usually very hard, most developers will try to foster certain behavior depending on the user. For example, a user more likely to make in-app purchases will receive offers to buy in-app products while a user less likely to do so, might see more ads.

Conclusions

Predictive analytics and more specifically LTV are becoming more widely used among developers. While these techniques help understand an app better from a business perspective, care needs to be taken to avoid some common pitfalls, such as making overly optimistic calculations or centering a strategy around LTV optimizations.

Use of LTV can be introduced with a simple calculation based on estimating it for a certain period, understanding the average lifetime for this period, and the unit value generated per user.

When speaking with various developers about optimizing monetization, it’s clear that both engagement and retention are key: whether it’s a subscription business that needs to reduce its churn ratio or an app selling in-app purchases where higher engagement means an increased opportunity to monetize users.

Further reading

If you’re interested in learning more about predictive analytics and LTV, you might be interested in the following:


What do you think?

Do you have questions or thoughts on predicting lifetime value? Continue the discussion in the comments below or tweet using the hashtag #AskPlayDev and we’ll reply from @GooglePlayDev, where we regularly share news and tips on how to be successful on Google Play.