Andy Carvell

For cart abandonment, leverage limited incentives like one-time offers and discounts. This can be used for users abandoning checkout but also interacting with the paywall (clicking through different packages and prices).

Figure out the sweet spots for discounts. You’re leaving a ton of money on the table if you’re treating everyone equally.

When getting started on pricing or when reconsidering pricing, survey users to know what they would be willing to pay. Do this per market as you can see big differences based on geo.

If you have users that are willing to pay a lot, give them a platinum option. It can be an expensive lifetime option for example. This also makes other plans look cheaper.

A lot of the subscriptions happen very early on in the lifecycle (often the first 24h). Discounts and coupons can be a very valuable way to get more people across the line if they don't convert early.

In your free tier, have several CTAs inviting to use a feature only subscribers have access to and clearly mark it (example: lock icon with "Subscribe to unlock"). This way you'll be able to identify the users that showed interest.

You can create custom upsell pages (paywalls) for different segments of customers.

Trying to avoid the app stores' tax is probably not worth it if you are not a big player. You need to figure out how to reduce the friction it creates and how not to get kicked out of the store. You also have to handle payment processing, refund, etc.

There are different indications of intent from users to become a paid subscriber based on their activity in the app: indirect intent, direct intent and abandoned cart.

To increase your subscription conversion rate, segment and personalize your paywall messaging depending on the intent level that they've shown: \- High intent → show them stuff tied to what they've shown interest in \- Indirect intent only → less signal so pitch the most popular stuff (unless you ran an onboarding survey or questionnaire) (Check out this slide)

When it comes to in-app messages something important is to have trigger points, not just target based on a segment. But keep in mind the smaller reach you might get if you go with something too narrow: you want to get statistical significance.

You can use a push notification or email to have users come into the app, before showing the in-app message adapted to the segment. That way you chain these campaigns together to leverage the strengths of these different channels.

If you are building slightly more advanced in-app messages with javascript you can ping results of surveys straight back into CRM system to enrich user profiles and start immediately segmenting them.

Low-hanging fruits for in-app messages: work on the onboarding flow. Figure out your key features/screens in the app (something tied to better activation rates), target users that haven't used/seen these features and educate/deep link to them.

Best use cases for in-app messages: on-boarding, upselling, content recommendation and surveys.

Data taxonomy is critical: everything rests on how clean and well-structured your data is. That starts by documenting what you’re currently collecting, where it sits, how it’s encoded, etc. From there, you can identify gaps.

A common concern is that sending another push notification will make users opt-out of push, or even uninstall the app. This is often overblown, and there is usually a lot of untapped upside especially if you have relevant messaging.

You want to know not just who subscribes but also who cancels and who asks for a refund (and when) because it makes a very big difference as to how valuable a user appears when you make your calculations (e.g. LTV) (check out this slide)

Use "abandoned cart"-type tactics (personalized approach, discount, etc.) for users that interacted with the paywall (i.e. swipes, selecting a price package, etc.) even if they did not try to start the subscription (i.e. hitting "subscribe now" button).

Simplified "propensity to purchase": build a kind of intent model. Look at the signs of intent to purchase and sort them by level of intent: click on a feature that is behind the paywall, interaction with the paywall, starting a free trial, etc.

Working on retention means different things depending on the growth stage:

• Early days (pre product-market fit): focus on having a solid product (good design, gut feel) and set analytics to measure retention. You won’t be able to run A/B tests because you don’t have enough volume, but you can get a qualitative understanding of how people use your product (even with wireframes, in alpha, in beta).
• Later on (post product-market fit and at scale): CRM with push notifications, emails, and in-app messages can be great boosters on top of the foundation. They typically require large cohorts to play with.

Do not wait until users go dormant. If you can't/don't build a fancy machine-learning churn predictor, keep an eye on your analytics to observe signs of dropping off to create a segment of users whose activity is declining so you can target them (potentially with more drastic tactics).

Simplified "propensity to churn" model: at SoundCloud they looked at the whole userbase to define how many days users should be inactive and mapped out the graph of when they should contact them to reactivate them.

Your onboarding needs to ensure continuity with the expectations you set in the App Store. A lot of onboarding problems come from a misalignment between the promises made or the expectations set in the acquisition process or the app stores.

There’s no unique path to purchase. Based on this, think about how you can personalize your approach on optimizing subscription revenue.

“Signal testing” is about identifying the early signals (within 24 hours) that indicate a user’s likelihood to become a subscriber (beyond free trial starts) in order to optimize your acquisition through “feeding back” the ad networks’ algorithms. Phiture uses the AIC framework (bucketing events in Acknowledgement, Interest, Conversion) to help with this.

Try to tackle one (maximum two) strategic objectives per quarter (e.g. free trial starts, free trial to paid CVR, increasing subscriber lifetime value, upselling, reducing churn). Don’t try to tackle everything at once.

People are very willing to answer surveys and seem to actually like being asked for their opinion about things, including during onboarding.

You know your product inside out, so you assume subscribers can find your features or content. But that’s not always the case. You must guide subscribers and help them find what you want them to do: adding a premium onboarding can help!