The Payment Screen Is Killing Your Conversions

01 /The Concept

Helium helps app developers optimize their mobile paywall screens – at what the startup calls "the speed of thought."

The team's provocative framing: "mobile app paywalls are stuck in 2008" – despite the fact that in-app purchases generated $150B in 2023, of which $45B came from subscriptions.

Developers already spend roughly a third of their time on features designed to drive paid conversions – in-app nudges, upgrade prompts, promotional flows. But those efforts frequently stall at the paywall itself, where the final act of payment needs its own layer of optimization.

The problem: optimizing paywalls the traditional way means programming and testing, then waiting for the updated app to clear the store review process, then waiting for users to download the update. And repeating that entire cycle for every subsequent experiment. Each iteration can take up to six weeks – and each is just one experiment with an unknown outcome. Running multiple experiments sequentially turns a straightforward optimization project into a months-long slog.

The scale compounds the challenge. One of Helium's clients has 87 distinct paywall screens inside a single app – each one a candidate for ongoing optimization. And optimizing "in general" is often less effective than personalizing to specific user segments: one client saw a 40% lift in purchase conversion after introducing segmentation. But segmentation adds further complexity to an already heavy process.

Helium's solution: paywalls are generated dynamically from the design currently live on the platform. Developers make one integration per paywall location – a single API call specifying which screen to render. After that, all visual changes happen in the platform's editor and go live immediately, with no app store submission required. Marketers and product managers run the experiments themselves, without engineering involvement.

For each paywall location, teams can define multiple design variants. The platform splits traffic across variants, reports conversion rates per design in the dashboard, and lets teams roll out the winner. The Y Combinator company description also mentions a SmartSelect capability – the AI automatically selects the highest-converting design for each individual user based on their history or segment membership.

Helium was founded this year and is currently in Y Combinator, which provided the first $500K in funding. The founding team includes someone who ran conversion experiments for Uber's mobile app, another with a background in ad personalization at Meta, and a third who worked on AI for Alexa at Amazon – a well-suited combination for this problem.

02 /Why It Matters

Helium sits at the intersection of two converging trends.

AI-powered optimization of individual product elements is one converging trend. The most recent comparable example is Bloom ([related review](/review/44-k-konversii-v-pokupku)), which raised $4.3M to automatically identify the best-performing product images for e-commerce listings. The right image, Bloom claims, can lift purchase conversion by 20–44%.

Bloom's AI doesn't just cycle through image variants – it looks for patterns in what converts: composition, color palette, model types, context. The goal is to give merchandising teams guidance on what kinds of images to produce for new products. It also personalizes images at the individual user level, surfacing whichever visual is statistically most likely to convert a given user.

The other is the ongoing expansion of paid in-app features. When an app has a single paid feature, teams eventually find the optimal way to sell it and can settle into a rhythm. But as paid features keep multiplying, each new addition requires a fresh round of discovery – finding the right promotion, the right paywall, the right moment to show it. That creates persistent demand for platforms that reduce the cost and time of that experimentation cycle.

Paywall optimization is part of a broader loop that also includes awareness – getting users to know a feature exists before they can be converted on it. Platforms for that adjacent problem have also started appearing.

Plotline ([related review](/review/nachnut-imi-polzovatsja-prodolzhat-platit)), which raised $2.6M in its first round, works on the same architectural principle as Helium: feature announcement blocks are built by marketers and product teams on the platform and appear immediately in the live app, no resubmission needed.

Appcues ([related review](/review/chto-ja-tut-mogu-poleznogo-dlja-sebja-sdelat)) built a similar platform earlier and has raised $52.8M – with new capital arriving shortly after its initial review.

The implication for Helium: paywall optimization probably can't stay siloed for long. The most effective conversion strategy connects how a feature is announced to how it's presented at the paywall – running them as disconnected experiments will be less powerful than treating them as a unified funnel.

03 /Opportunities

The general direction: platforms for optimizing revenue-critical elements of mobile apps.

The three-layer architecture that Helium and others in this space use is worth adopting as a template:

- A visual editor where marketers and product managers compose the target screens themselves, with changes going live immediately without app store resubmission.

- Automated experimentation that finds the optimal design for maximizing conversion to the target action.

- Personalization that applies the optimal design not just globally, but at the individual user level.

What other standard, high-stakes elements in mobile apps need this treatment beyond feature announcements and paywalls? Onboarding flows and upgrade nudges are the obvious next layer – and both already have documented best practices that an AI could learn to apply and personalize at the user level.