Subscription pricing for AI apps: plans, compute costs and payback
A simple paywall can sit in front of a much larger plan catalog. Phil Carter explains how to present that choice and account for ongoing AI compute costs. Price, margin and retention determine whether customer value covers service costs and acquisition payback.
By NewForm · Updated
Key takeaways
- A large plan catalog does not mean every option needs to appear on the same paywall.
- Carter proposes matching the displayed offer to user needs; the clip does not verify a specific app’s prediction system.
- Compute is an ongoing cost of delivering the service and belongs in the margin used to assess acquisition payback.
- Longer retention can create more time to recover CAC, but it does not remove ongoing compute costs.
The illusion of simplicity in subscription offers
App store listings often show a long catalog of in-app purchases: weekly passes, monthly renewals and multi-month tiers. Showing every option on the paywall can overwhelm a prospective subscriber and reduce conversion.
Carter calls this an “illusion of simplicity”: a business can maintain multiple plans while presenting a focused choice. He describes predictive selection of a tier or billing cadence as a possible approach. The conversation does not establish that the named apps use a particular model or that personalization necessarily improves conversion.
Account for compute costs in pricing and payback
Carter distinguishes the cost of acquiring a customer from the ongoing cost of delivering an AI service. Compute belongs in the pricing and monetization model, and its effect on contribution margin matters when assessing how much acquisition spend the business can recover.
Longer retention can spread a one-time acquisition expense across more periods of customer value. It does not make compute disappear: the customer may continue generating service costs as well as revenue. With short retention or one-off usage, there may be less opportunity to recover CAC after those costs.
The practical implication is to evaluate price, usage, margin and retention together. A low acquisition cost is not enough to establish a viable payback period if serving the customer consumes most of the revenue.
Worked example: usage changes the payback calculation
These are hypothetical per-customer figures, not an app benchmark or a forecast. Assume $20 collected revenue per paid month, $1 of other variable service costs and $36 acquisition cost. Compute cost changes with the user’s usage. For simplicity, the table assumes constant usage, continued payment and no churn; real cohort payback needs the observed retention and cost history.
Contribution for a paid month is collected revenue minus compute and the other variable service costs. Simple payback in the last column divides CAC by that contribution. It does not predict how long the customer will stay, and fractional months do not change the actual billing schedule.
| Usage scenario | Revenue | Compute | Other variable costs | Contribution | Simple payback |
|---|---|---|---|---|---|
| Light usage | $20 | $3 | $1 | $16 | 2.25 paid months |
| Base usage | $20 | $7 | $1 | $12 | 3 paid months |
| Heavy usage | $20 | $17 | $1 | $2 | 18 paid months |
Connect pricing decisions to the acquisition cohort
At the same headline price and CAC, these scenarios recover acquisition cost at very different speeds. If customers usually leave before the required number of paid periods, the simple table will overstate the ability to recover CAC. Annual cash collected upfront also differs from the obligation to provide a year of service.
Track revenue, refunds, compute and other variable costs by acquisition cohort and usage band. Compare cumulative contribution with the cohort’s acquisition spend. Include the cost of users who activate a trial but never pay when that cost belongs to the acquisition program. Keep the definition consistent before comparing channels or creative concepts.
Then evaluate pricing, usage allowances and creative together. An ad that attracts unusually heavy users can appear efficient on signup cost while bringing a weak contribution profile. A clearer demonstration of the intended use case may attract customers who both retain and receive value within the plan’s economics. Test that possibility with cohort data rather than inferring it from a cheap click.
Our subscription-app creative shortlist explains the questions to ask a partner about installs, trials, subscribers and retention. The brief should identify the event and economics the team is actually trying to improve.
Adapted from NewForm’s original videos on creative strategy and paid social. Worked examples, where included, are NewForm editorial illustrations, not measured client results.
Explore NewForm Framework →