In this exclusive Q&A session done at the very end of the App Growth Summit, Thomas Petit (growth consultant) answers questions about subscription apps and the corresponding user acquisition.
The free trial event has the 2 characteristics that you need to make app event optimization work:
- It happens early enough
- It happens often enough
One of the problems of going deeper than a free trial for AEO is if that the % of completion of the event is too low. This depends on the market you're operating on, on the free trial itself, on your budget, and on the typical cost of an event. Make sure you can reach the strict minimum of 10 events per day (the more the better).
After a while, look at how you can send a better event than the free trial. For example by delaying the free trial event to only send the event to Facebook for users that have not canceled the free trial.
To find what you should optimize for if you do not have a free trial, run a regression analysis on every single event you have in your onboarding and first day of usage and see what is correlating with purchase/revenue (or even long-term revenue). Example: completing a first meditation.
Sometimes the data/event is very specific to your app so you can't necessarily bring learnings from other apps. Example: looking at the duration of signup flow to exclude users that go to it very quickly can work for an app but these users might be the most valuable for another app.
If you can't find correlation between events and your goal, you might want to iterate on your onboarding to have personalization questions (even if doesn't really customize the experience) that allow you to optimize acquisition.
Sometimes the right event for AEO is a combination of a lot of small things. It's often keeping the free trial as main event but filtering it a little bit. For example: only if the user has answered that way to this question, and only if he hasn't cancelled the event straight away OR if he hasn't completed the free trial but completed 3 activities on day 1 + invited a friend. This kind of mix can help you getting to a higher % of completion.
If you have the structure to run multiple subscriptions in parallel and a lot of tests, present subscription options sequentially. You can have a high priced yearly plan (higher revenue, money upfront) when you first show the paywall. You will see a drop off from people that are sensitive to price, but you can later offer them a different subscription or even discounts.
Challenge with subscriptions (vs. IAP for gaming):
- There is a spend "ceiling" on subscriptions (e.g. everybody is paying $50/year). You might be underselling to some users!
- There is a spend "floor" because subscriptions put a barrier to entry where games don't (e.g. smaller IAP, users monetized just with rewarded videos, or even leveraging users for virality).
Think about how you can get value out of the part of your audience that can't/won't pay, without charging them for a subscription. Virality might be the answer: "invite 3 friends and I give you a free month".
Thomas once tested "only yearly" and "only monthly" and the LTV ended up being the exact same, but because with yearly you get the money upfront it can make a huge difference. Some companies might even be interested in getting a cut on LTV for better cashflow.
For a lot of apps, hitting users with the paywall for free trial almost immediately works fantastically. Especially for apps not building for long term, focused on acquisition, need early data, etc. But over time you're also paying it elsewhere: retention, reviews, etc.
In a test where some users were paying through Stripe in-app, the difference in conversion from asking users to add their credit cards (friction) was exactly compensated by the gain of avoiding platform fees. It can still be a positive because you control the transaction (refunds, etc.), data model is better, etc.
The free trial event has the 2 characteristics that you need to make app event optimization work:
- It happens early enough
- It happens often enough
One of the problems of going deeper than a free trial for AEO is if that the % of completion of the event is too low. This depends on the market you're operating on, on the free trial itself, on your budget, and on the typical cost of an event. Make sure you can reach the strict minimum of 10 events per day (the more the better).
After a while, look at how you can send a better event than the free trial. For example by delaying the free trial event to only send the event to Facebook for users that have not canceled the free trial.
To find what you should optimize for if you do not have a free trial, run a regression analysis on every single event you have in your onboarding and first day of usage and see what is correlating with purchase/revenue (or even long-term revenue). Example: completing a first meditation.
Sometimes the data/event is very specific to your app so you can't necessarily bring learnings from other apps. Example: looking at the duration of signup flow to exclude users that go to it very quickly can work for an app but these users might be the most valuable for another app.
If you can't find correlation between events and your goal, you might want to iterate on your onboarding to have personalization questions (even if doesn't really customize the experience) that allow you to optimize acquisition.
Sometimes the right event for AEO is a combination of a lot of small things. It's often keeping the free trial as main event but filtering it a little bit. For example: only if the user has answered that way to this question, and only if he hasn't cancelled the event straight away OR if he hasn't completed the free trial but completed 3 activities on day 1 + invited a friend. This kind of mix can help you getting to a higher % of completion.
If you have the structure to run multiple subscriptions in parallel and a lot of tests, present subscription options sequentially. You can have a high priced yearly plan (higher revenue, money upfront) when you first show the paywall. You will see a drop off from people that are sensitive to price, but you can later offer them a different subscription or even discounts.
Challenge with subscriptions (vs. IAP for gaming):
- There is a spend "ceiling" on subscriptions (e.g. everybody is paying $50/year). You might be underselling to some users!
- There is a spend "floor" because subscriptions put a barrier to entry where games don't (e.g. smaller IAP, users monetized just with rewarded videos, or even leveraging users for virality).
Think about how you can get value out of the part of your audience that can't/won't pay, without charging them for a subscription. Virality might be the answer: "invite 3 friends and I give you a free month".
Thomas once tested "only yearly" and "only monthly" and the LTV ended up being the exact same, but because with yearly you get the money upfront it can make a huge difference. Some companies might even be interested in getting a cut on LTV for better cashflow.
For a lot of apps, hitting users with the paywall for free trial almost immediately works fantastically. Especially for apps not building for long term, focused on acquisition, need early data, etc. But over time you're also paying it elsewhere: retention, reviews, etc.
In a test where some users were paying through Stripe in-app, the difference in conversion from asking users to add their credit cards (friction) was exactly compensated by the gain of avoiding platform fees. It can still be a positive because you control the transaction (refunds, etc.), data model is better, etc.
The free trial event has the 2 characteristics that you need to make app event optimization work:
- It happens early enough
- It happens often enough
One of the problems of going deeper than a free trial for AEO is if that the % of completion of the event is too low. This depends on the market you're operating on, on the free trial itself, on your budget, and on the typical cost of an event. Make sure you can reach the strict minimum of 10 events per day (the more the better).
After a while, look at how you can send a better event than the free trial. For example by delaying the free trial event to only send the event to Facebook for users that have not canceled the free trial.
To find what you should optimize for if you do not have a free trial, run a regression analysis on every single event you have in your onboarding and first day of usage and see what is correlating with purchase/revenue (or even long-term revenue). Example: completing a first meditation.
Sometimes the data/event is very specific to your app so you can't necessarily bring learnings from other apps. Example: looking at the duration of signup flow to exclude users that go to it very quickly can work for an app but these users might be the most valuable for another app.
If you can't find correlation between events and your goal, you might want to iterate on your onboarding to have personalization questions (even if doesn't really customize the experience) that allow you to optimize acquisition.
Sometimes the right event for AEO is a combination of a lot of small things. It's often keeping the free trial as main event but filtering it a little bit. For example: only if the user has answered that way to this question, and only if he hasn't cancelled the event straight away OR if he hasn't completed the free trial but completed 3 activities on day 1 + invited a friend. This kind of mix can help you getting to a higher % of completion.
If you have the structure to run multiple subscriptions in parallel and a lot of tests, present subscription options sequentially. You can have a high priced yearly plan (higher revenue, money upfront) when you first show the paywall. You will see a drop off from people that are sensitive to price, but you can later offer them a different subscription or even discounts.
Challenge with subscriptions (vs. IAP for gaming):
- There is a spend "ceiling" on subscriptions (e.g. everybody is paying $50/year). You might be underselling to some users!
- There is a spend "floor" because subscriptions put a barrier to entry where games don't (e.g. smaller IAP, users monetized just with rewarded videos, or even leveraging users for virality).
Think about how you can get value out of the part of your audience that can't/won't pay, without charging them for a subscription. Virality might be the answer: "invite 3 friends and I give you a free month".
Thomas once tested "only yearly" and "only monthly" and the LTV ended up being the exact same, but because with yearly you get the money upfront it can make a huge difference. Some companies might even be interested in getting a cut on LTV for better cashflow.
For a lot of apps, hitting users with the paywall for free trial almost immediately works fantastically. Especially for apps not building for long term, focused on acquisition, need early data, etc. But over time you're also paying it elsewhere: retention, reviews, etc.
In a test where some users were paying through Stripe in-app, the difference in conversion from asking users to add their credit cards (friction) was exactly compensated by the gain of avoiding platform fees. It can still be a positive because you control the transaction (refunds, etc.), data model is better, etc.
Notes for this resource are currently being transferred and will be available soon.
The majority of subscriptions happen before people even see the product.
[💎 @35:05] Sometimes the right event for AEO is a combination of a lot of small things. It's often keeping the free trial as main event but filtering it a little bit. For example: only if the user has answered that way to this question, and only if he hasn't cancelled the event straight away OR if he hasn't completed the free trial but completed 3 activities on day 1 + invited a friend. This kind of mix can help you getting to a higher % of completion.
Events will vary based on the nature of your app. What's mentioned above mostly works with "motivation" subscription apps, not "content" apps like Netflix or SoundCloud.
🔒 Check [RESOURCE #71] Subscription apps fundamentals (RESOURCE #71 - Subscription apps fundamentals with Thomas Petit) for more information on the different kinds of subscription apps (GrowthGems.co membership might be needed).
You need access to the events to be able to find the right events and optimize acquisition for them. If not, you can not do your job.
A workaround could be to ask for an anonymized dataset (channel 1/2/3, adsets a/b/c) that can be exported to a tool like data studio for example.
[💎 @44:40] If you have the structure to run multiple subscriptions in parallel and a lot of tests, present subscription options sequentially. You can have a high priced yearly plan (higher revenue, money upfront) when you first show the paywall. You will see a drop off from people that are sensitive to price, but you can later offer them a different subscription or even discounts.
Note: Hiding the "skip"/close option on the paywall is usually not a good tactic.
The most evolved subscription models have a lot of different options but not necessarily all visible to everybody. They do it sequentially and in a personalized way.
Example of sequence: $100 → $70 → $5/month
Testing only on the paywall, you will find local maximums. Going sequential you might be able to expend your paying user group.
Think about the floor and ceiling for user spending when it comes to subscription. 3/4 of money is made by gaming and a lot of this money is coming from whales.
Games have maximized the ceiling for user spending to infinity.
[💎 @50:08] Challenge with subscriptions (vs. IAP for gaming):
Ads can be an answer to lowering the floor, especially for geos where users don't have much money. But some apps can't show ads (e.g. app for kids).
[💎 @51:44] Think about how you can get value out of the part of your audience that can't/won't pay, without charging them for a subscription. Virality might be the answer: "invite 3 friends and I give you a free month".
Example: an app (maybe Memrise) does this well. If you don't pay, after 1 or 2 weeks they give you one of the features for free. When you've tested it, they remove it.
Monthly plan on mobile is not necessarily a must. A lot of people say that the shorter periods (weekly, monthly) are the best for LTV. But for "motivation" apps (cf. [RESOURCE #71] Subscription apps fundamentals) the motivation doesn't last for long.
[💎 @54:46] Thomas once tested "only yearly" and "only monthly" and the LTV ended up being the exact same, but because with yearly you get the money upfront it can make a huge difference. Some companies might even be interested in getting a cut on LTV for better cashflow.
Don't look at only LTV and payback.
Be careful of short term gains that become long term losses
Some apps even have a "hard" paywall: either you pay, or you can't do anything.
Win-back campaigns are hard. Grace periods can help, you implement CRM tactics
If you're at a big scale, a grace period affecting 2% of users can make a big difference. If you're at a small scale, concentrate on big wins (grace period won't be one).
Andre Kempe shared that LinkedIn did this kind of "one time deal". Repercussions in reviews/ratings is something to consider if you have a big brand.
LinkedIn actually tells you that if you re-subscribe you get 50% off, which is different from the refund/wire mentioned above.
Example of Basecamp.
(not sure this is still true now after the whole Hey "scandal") If you sell on other platforms (e.g. web) with people log in the app afterwards is allowed. But you can't tell people "hey, this feature is available if you go there". You can not say it. The only ones that can pull it are huge brands like Netflix, Spotify.
It's easier to do this if you have a big fanbase/audience outside of the app, and when the app is kind of a "side-product" (like Basecamp).
[💎 @01:12:00] In a test where some users were paying through Stripe in-app, the difference in conversion from asking users to add their credit cards (friction) was exactly compensated by the gain of avoiding platform fees. It can still be a positive because you control the transaction (refunds, etc.), data model is better, etc.
Apple also offers the convenience of paying easily, giving credibility and trust, etc.
Fitness apps adding features to fight for the renewal and keep users longer (sounds, nutrition, etc.) and on the other hand meditation apps are adding fitness features.
Adding content is not the way you get renewals. Improving the habit of your existing content can yield better impact on retention and renewals than adding content.
What headspace is doing is market expansion because they are reaching a plateau and therefore moving to a surrounding market.
When considering unbundling (e.g. language apps for each language, sport app for each club, etc.) keep in mind that Apple doesn't want you to have a portfolio of apps. You might also dilute the brand if you have too many apps, although BetterMe is doing well with only 4 apps. It also requires more resources.
The most successful apps only do 1 thing, like Whatsapp. So bundling is not necessarily the solution either.
8fit was one of the first app to have fitness and nutrition in one place, which was their value-added. But it was not what consumers wanted and results were worst when trying to sell the combo vs. each feature to different audience.
An app that does too many things usually end up doing none at all
If you're not the category leader, be careful about expanding your market.
Apple might catch you for something small but not catch you for something big. There's a random factor.
If you're trying to build something sustainable, play by the rules.
The majority of subscriptions happen before people even see the product.
[💎 @35:05] Sometimes the right event for AEO is a combination of a lot of small things. It's often keeping the free trial as main event but filtering it a little bit. For example: only if the user has answered that way to this question, and only if he hasn't cancelled the event straight away OR if he hasn't completed the free trial but completed 3 activities on day 1 + invited a friend. This kind of mix can help you getting to a higher % of completion.
Events will vary based on the nature of your app. What's mentioned above mostly works with "motivation" subscription apps, not "content" apps like Netflix or SoundCloud.
🔒 Check [RESOURCE #71] Subscription apps fundamentals (RESOURCE #71 - Subscription apps fundamentals with Thomas Petit) for more information on the different kinds of subscription apps (GrowthGems.co membership might be needed).
You need access to the events to be able to find the right events and optimize acquisition for them. If not, you can not do your job.
A workaround could be to ask for an anonymized dataset (channel 1/2/3, adsets a/b/c) that can be exported to a tool like data studio for example.
[💎 @44:40] If you have the structure to run multiple subscriptions in parallel and a lot of tests, present subscription options sequentially. You can have a high priced yearly plan (higher revenue, money upfront) when you first show the paywall. You will see a drop off from people that are sensitive to price, but you can later offer them a different subscription or even discounts.
Note: Hiding the "skip"/close option on the paywall is usually not a good tactic.
The most evolved subscription models have a lot of different options but not necessarily all visible to everybody. They do it sequentially and in a personalized way.
Example of sequence: $100 → $70 → $5/month
Testing only on the paywall, you will find local maximums. Going sequential you might be able to expend your paying user group.
Think about the floor and ceiling for user spending when it comes to subscription. 3/4 of money is made by gaming and a lot of this money is coming from whales.
Games have maximized the ceiling for user spending to infinity.
[💎 @50:08] Challenge with subscriptions (vs. IAP for gaming):
Ads can be an answer to lowering the floor, especially for geos where users don't have much money. But some apps can't show ads (e.g. app for kids).
[💎 @51:44] Think about how you can get value out of the part of your audience that can't/won't pay, without charging them for a subscription. Virality might be the answer: "invite 3 friends and I give you a free month".
Example: an app (maybe Memrise) does this well. If you don't pay, after 1 or 2 weeks they give you one of the features for free. When you've tested it, they remove it.
Monthly plan on mobile is not necessarily a must. A lot of people say that the shorter periods (weekly, monthly) are the best for LTV. But for "motivation" apps (cf. [RESOURCE #71] Subscription apps fundamentals) the motivation doesn't last for long.
[💎 @54:46] Thomas once tested "only yearly" and "only monthly" and the LTV ended up being the exact same, but because with yearly you get the money upfront it can make a huge difference. Some companies might even be interested in getting a cut on LTV for better cashflow.
Don't look at only LTV and payback.
Be careful of short term gains that become long term losses
Some apps even have a "hard" paywall: either you pay, or you can't do anything.
Win-back campaigns are hard. Grace periods can help, you implement CRM tactics
If you're at a big scale, a grace period affecting 2% of users can make a big difference. If you're at a small scale, concentrate on big wins (grace period won't be one).
Andre Kempe shared that LinkedIn did this kind of "one time deal". Repercussions in reviews/ratings is something to consider if you have a big brand.
LinkedIn actually tells you that if you re-subscribe you get 50% off, which is different from the refund/wire mentioned above.
Example of Basecamp.
(not sure this is still true now after the whole Hey "scandal") If you sell on other platforms (e.g. web) with people log in the app afterwards is allowed. But you can't tell people "hey, this feature is available if you go there". You can not say it. The only ones that can pull it are huge brands like Netflix, Spotify.
It's easier to do this if you have a big fanbase/audience outside of the app, and when the app is kind of a "side-product" (like Basecamp).
[💎 @01:12:00] In a test where some users were paying through Stripe in-app, the difference in conversion from asking users to add their credit cards (friction) was exactly compensated by the gain of avoiding platform fees. It can still be a positive because you control the transaction (refunds, etc.), data model is better, etc.
Apple also offers the convenience of paying easily, giving credibility and trust, etc.
Fitness apps adding features to fight for the renewal and keep users longer (sounds, nutrition, etc.) and on the other hand meditation apps are adding fitness features.
Adding content is not the way you get renewals. Improving the habit of your existing content can yield better impact on retention and renewals than adding content.
What headspace is doing is market expansion because they are reaching a plateau and therefore moving to a surrounding market.
When considering unbundling (e.g. language apps for each language, sport app for each club, etc.) keep in mind that Apple doesn't want you to have a portfolio of apps. You might also dilute the brand if you have too many apps, although BetterMe is doing well with only 4 apps. It also requires more resources.
The most successful apps only do 1 thing, like Whatsapp. So bundling is not necessarily the solution either.
8fit was one of the first app to have fitness and nutrition in one place, which was their value-added. But it was not what consumers wanted and results were worst when trying to sell the combo vs. each feature to different audience.
An app that does too many things usually end up doing none at all
If you're not the category leader, be careful about expanding your market.
Apple might catch you for something small but not catch you for something big. There's a random factor.
If you're trying to build something sustainable, play by the rules.
The majority of subscriptions happen before people even see the product.
[💎 @35:05] Sometimes the right event for AEO is a combination of a lot of small things. It's often keeping the free trial as main event but filtering it a little bit. For example: only if the user has answered that way to this question, and only if he hasn't cancelled the event straight away OR if he hasn't completed the free trial but completed 3 activities on day 1 + invited a friend. This kind of mix can help you getting to a higher % of completion.
Events will vary based on the nature of your app. What's mentioned above mostly works with "motivation" subscription apps, not "content" apps like Netflix or SoundCloud.
🔒 Check [RESOURCE #71] Subscription apps fundamentals (RESOURCE #71 - Subscription apps fundamentals with Thomas Petit) for more information on the different kinds of subscription apps (GrowthGems.co membership might be needed).
You need access to the events to be able to find the right events and optimize acquisition for them. If not, you can not do your job.
A workaround could be to ask for an anonymized dataset (channel 1/2/3, adsets a/b/c) that can be exported to a tool like data studio for example.
[💎 @44:40] If you have the structure to run multiple subscriptions in parallel and a lot of tests, present subscription options sequentially. You can have a high priced yearly plan (higher revenue, money upfront) when you first show the paywall. You will see a drop off from people that are sensitive to price, but you can later offer them a different subscription or even discounts.
Note: Hiding the "skip"/close option on the paywall is usually not a good tactic.
The most evolved subscription models have a lot of different options but not necessarily all visible to everybody. They do it sequentially and in a personalized way.
Example of sequence: $100 → $70 → $5/month
Testing only on the paywall, you will find local maximums. Going sequential you might be able to expend your paying user group.
Think about the floor and ceiling for user spending when it comes to subscription. 3/4 of money is made by gaming and a lot of this money is coming from whales.
Games have maximized the ceiling for user spending to infinity.
[💎 @50:08] Challenge with subscriptions (vs. IAP for gaming):
Ads can be an answer to lowering the floor, especially for geos where users don't have much money. But some apps can't show ads (e.g. app for kids).
[💎 @51:44] Think about how you can get value out of the part of your audience that can't/won't pay, without charging them for a subscription. Virality might be the answer: "invite 3 friends and I give you a free month".
Example: an app (maybe Memrise) does this well. If you don't pay, after 1 or 2 weeks they give you one of the features for free. When you've tested it, they remove it.
Monthly plan on mobile is not necessarily a must. A lot of people say that the shorter periods (weekly, monthly) are the best for LTV. But for "motivation" apps (cf. [RESOURCE #71] Subscription apps fundamentals) the motivation doesn't last for long.
[💎 @54:46] Thomas once tested "only yearly" and "only monthly" and the LTV ended up being the exact same, but because with yearly you get the money upfront it can make a huge difference. Some companies might even be interested in getting a cut on LTV for better cashflow.
Don't look at only LTV and payback.
Be careful of short term gains that become long term losses
Some apps even have a "hard" paywall: either you pay, or you can't do anything.
Win-back campaigns are hard. Grace periods can help, you implement CRM tactics
If you're at a big scale, a grace period affecting 2% of users can make a big difference. If you're at a small scale, concentrate on big wins (grace period won't be one).
Andre Kempe shared that LinkedIn did this kind of "one time deal". Repercussions in reviews/ratings is something to consider if you have a big brand.
LinkedIn actually tells you that if you re-subscribe you get 50% off, which is different from the refund/wire mentioned above.
Example of Basecamp.
(not sure this is still true now after the whole Hey "scandal") If you sell on other platforms (e.g. web) with people log in the app afterwards is allowed. But you can't tell people "hey, this feature is available if you go there". You can not say it. The only ones that can pull it are huge brands like Netflix, Spotify.
It's easier to do this if you have a big fanbase/audience outside of the app, and when the app is kind of a "side-product" (like Basecamp).
[💎 @01:12:00] In a test where some users were paying through Stripe in-app, the difference in conversion from asking users to add their credit cards (friction) was exactly compensated by the gain of avoiding platform fees. It can still be a positive because you control the transaction (refunds, etc.), data model is better, etc.
Apple also offers the convenience of paying easily, giving credibility and trust, etc.
Fitness apps adding features to fight for the renewal and keep users longer (sounds, nutrition, etc.) and on the other hand meditation apps are adding fitness features.
Adding content is not the way you get renewals. Improving the habit of your existing content can yield better impact on retention and renewals than adding content.
What headspace is doing is market expansion because they are reaching a plateau and therefore moving to a surrounding market.
When considering unbundling (e.g. language apps for each language, sport app for each club, etc.) keep in mind that Apple doesn't want you to have a portfolio of apps. You might also dilute the brand if you have too many apps, although BetterMe is doing well with only 4 apps. It also requires more resources.
The most successful apps only do 1 thing, like Whatsapp. So bundling is not necessarily the solution either.
8fit was one of the first app to have fitness and nutrition in one place, which was their value-added. But it was not what consumers wanted and results were worst when trying to sell the combo vs. each feature to different audience.
An app that does too many things usually end up doing none at all
If you're not the category leader, be careful about expanding your market.
Apple might catch you for something small but not catch you for something big. There's a random factor.
If you're trying to build something sustainable, play by the rules.
