Phil Carter

AI has fundamentally changed the cost structure of consumer subscription apps, making multi-tier pricing a necessity rather than a luxury. The emerging standard is a basic tier with limited AI access and a premium tier with full AI capabilities, mirroring B2B SaaS pricing models. If you have meaningful AI features, structure your pricing to protect your unit economics while using the basic tier as a funnel into higher-value plans.

AI apps face a unique trial design challenge: longer trials convert better but can be financially catastrophic when power users rack up massive compute costs and then churn. The industry trend toward shorter trials is being driven by AI cost economics, not conversion data. Carefully model your AI compute costs per trial user before setting trial length, and consider capping AI feature usage during trials rather than simply shortening them.

Prices going down for subscription apps make sense because:

1. There's more competition/supply
2. It's now more cost-effective and faster to build software with AI

Expand conversion among younger users by

1. Being strategic with discounts (e.g., activity-based pricing or 2nd offers)
2. Providing alternative payment methods like PayPal or Venmo, since many younger users don't have credit cards.

AI-powered apps face significant marginal costs for LLM compute, unlike traditional subscription apps with near-zero marginal costs. Web-to-app funnels become critical for AI apps because avoiding the 30% app store fee helps offset AI compute costs and maintain healthy unit economics.

There's an "accordion" between:

• Finding new growth opportunities with low-hanging fruits leading to a high experiment win rate
• Not being able to get as many winning experiments and being in the "micro-optimization" zone which indicates you need to find new growth opportunities/areas

Customize the paywall for different entry points: the channel users came in through or other signals you have on users’ behavior. For example, if you enter Calm via Matthew McConaughey content, you will be displayed a paywall related to Matthew McConaughey.

A/B testing for pricing is fast and effective but is a blunt instrument. Surveys like Van Westendorp or Conjoint analysis can give you deeper insights when you do them correctly.

Promotions are a tool to better align the value of your product with the willingness to pay of an individual customer so you can capture more consumer surplus. You have to be targeted and thoughtful about promotions. The companies that are doing this best:

• Interject their brand and personality into seasonal promotions to stand out
• Use activity-based discounts (e.g., offer a discount after the point where 90% of subscribers typically convert)
• Segment based on demographics/psychographics to better align the price of your product with the customer

Value creation is the most important step - you need something people really want:

• Robust: do you have the value promise that is solving a real pain point and has PMF?
• Rapid: it’s hard to retain people’s attention. People need to understand why your product matters to them very quickly (within the first session, the first 30 seconds)
• Repeatable: you need to make sure that the value promise is repeatable and that there are new chapters that bring users back in. Otherwise, there’s a point at which users have gotten most of the value.
• Remarkable: you want a product that is unique and compelling enough that people will want to talk about it

AI has opened a window to create magical first-session experiences that convert users within 30 to 60 seconds. With over 80% of trial starts now happening on day zero (and up to 90% in some categories), your onboarding must deliver immediate value or you will lose that user forever. Invest in AI-powered onboarding that hooks users instantly rather than treating setup as a chore.

Web-to-app funnels provide two critical advantages:

1. Instant iteration without app store review cycles, enabling rapid A/B testing
2. Deterministic attribution data that can be passed back to ad networks to improve campaign optimization and paid ad efficiency.
3. Avoid app store fees

Don't default to web-to-app for every product. If your core 'aha moment' requires native app features or is best experienced in-app (like AI companions), prioritize getting users into the app quickly rather than extending the web funnel. The medium should serve the experience, not vice versa.

People have seen things before they get to your onboarding. It shouldn’t repeat what they’ve seen on the App Store. Once in the onboarding: show, don’t tell: demonstrate how you will deliver the value.

AI-powered apps with chat interfaces have a natural advantage for personalization since the LLM response itself is the personalized experience. For non-AI apps, deep personalization works best when you can ask detailed questions (like fitness goals and experience level) that directly inform product personalization.

It all starts by making sure you deeply understand as a company what your differentiator is. Whatever that is, index on that being the thing that you spend your highly expensive engineering technical resources to do it better than anyone else in the market. Are you a commodity? Do you have a product that no one else has? Do you have a technology? Do you have a particular process or access model? For everything else, try to find the best partner possible to put it on autopilot.

If you’re already at scale, personify the app and celebrate the purchase. Example: Duolingo’s owl and Super Duolingo trial start where the icon changes, fireworks, etc.

Value creation vs. value capture:

• Core product teams focus on creating core value for the user, with the primary metric being retention
• Growth product work on capturing value as well as new user onboarding, with the primary metric being activation

As the app ecosystem becomes more crowded and attention spans shrink, the value of exceptional marketing has increased dramatically. Distribution and attention capture are now as important as product quality in determining success.

The growth model looks different for every company. What you’re looking for is the Minimum Viable Model that really gets at the variables that are most leveraged for driving the business and tied to user actions: acquisition, monetization, and retention variables.

If you’re using a freemium model, you should think about what value you’re capturing from free users and what it is adding to the equation. Example: in AllTrails, free users help improve the data, etc.

Unless you have a very strong network effect (which most consumer subscription apps don’t), you can’t afford NOT to re-invest most of the value you capture into the product. Because if you don’t, someone else will come along and will leapfrog you.

Even if you’re not a product that depends primarily on virality to grow your user base, you want a product that is unique and compelling enough that people will want to talk about it. Extreme examples: Tinder’s “Swipe” and Strava’s “If it’s not on Strava, it didn’t happen.”

The best consumer subscription companies have robust organic acquisition strategies that are their primary growth lever and only use paid ads as a supplement. It can be through word-of-mouth but also content. Example: Quizlet with long-term SEO, Alltrails with hyper-local search.

Consumer subscription businesses only work if they have very high long-term retention rates because that builds the compounding layer cake of net revenue retention.

Ultimately, you want to build a sustainable habit with users. But you’ll never get there if you can’t tell them why they should use your app in their first session. Frontload as much delight and backload as much work as possible so users get to the aha moment as quickly as possible.

Don't neglect brand marketing just because it's hard to measure. Like retention and word of mouth, brand investment compounds over time and becomes a durable growth moat that performance marketing alone can't provide.

AI creative tools can take your paid acquisition from tens of concepts per month to 400-plus, and the benefits go far beyond lower CAC. The faster learning cycle from testing more creatives reveals what messaging resonates with users, which directly informs product roadmap decisions. Treat your creative testing program as a dual-purpose engine for both paid UA optimization and product-market fit discovery.

Use predictive LTV based on quiz responses as a strategic tool for product development to prioritize feature development and product investments toward your most valuable cohorts.

Conjoint analysis is a great supplement to Van Westendorp and helps you get insights into what they value most about your product. It presents users with a set of alternatives: 2-4 packages, each with attributes (price, number of usage-based, range, features you get or don’t get). Then, users go through tasks where they see those packages.