The work, the stages, the community.

new Chinese subscriptions

Launched Alipay via Stripe; new Chinese subscriptions grew approximately 14× against the pre-launch baseline.

New Chinese subscriptions after the May 2025 Alipay launch, relative to the pre-launch baseline, measured using first paid invoices.

New Indian Subs on UPI

Delivered UPI Autopay via EBANX and adapted subscription flows to mandate constraints; UPI reached ~83% of new Indian subscriptions and total acquisition exceeded pre-decline levels.

UPI's share of new Indian subscriptions within three months of launch in 2026, measured using first paid invoices.

lines of config per supported payment method Previously days of cross-cutting implementation

Designed reactive routing across Stripe, Adyen and Braintree via Recurly, with conditional SDK loading, fallbacks and shared analytics, state and types.

Approximate configuration required to enable a supported payment method through the orchestration layer; not a measured percentage reduction in rollout time.

30-Day Retention 2.14% → 3.49%; +1.35 percentage points

Built and iterated cancellation offers; the second A/B test lifted 30-day retention among cancelling users by 63% relative to control.

Relative uplift in 30-day retention among cancelling users in cancellation-offer v2: 2.14% in control versus 3.49% in treatment.

Release Time GitClear/DORA tracking

Rebuilt release practices and feature-flag management, moving from multi-day release trains to same-day releases.

Reported 88% reduction in time-to-release, tracked through GitClear/DORA work; the original query is no longer recoverable.

Bundle Size

Made payment SDK loading conditional on the selected gateway, cutting the checkout bundle by 60%.

60% reduction in the Sliding Dialog checkout bundle from a 965.72 KiB baseline, based on PR build-output comparisons.

new Chinese subscriptions using Alipay

Alipay became the dominant acquisition method after its May 2025 launch.

Share of new Chinese subscriptions paid through Alipay in calendar year 2025, measured using first paid invoices.

US checkout conversion during tax rollout Observed stability; not a causal A/B-test result

Delivered validated postal codes and failure-tolerant server-calculated price previews; observed weekly US conversion stayed within 1.4–1.6%.

Observed weekly US checkout conversion from March 30 to June 29, 2026, through and after the sales-tax rollout.

Apple Pay Success Rate Australia; checkout success, not gateway authorisation

On Smallpdf (tens of millions of monthly users), one-tap Apple Pay reached ~86% checkout success in Australia versus ~57–67% on cards.

Observed Apple Pay checkout success in Australia during the 2025 measurement window, compared with ~57–67% for cards.

Stripe payment methods shipped

Implemented method-specific flows, setup-intent lifecycle handling and 3DS support.

Stripe methods delivered: Link, PayPal, Alipay, Google Pay, Apple Pay and SEPA.

callsites in pricing data-layer migration

Led TanStack Query adoption across 76 files and trained the team through hands-on workshops.

Callsites covered by the migration from fragmented legacy pricing fetching to a resilient, cached and schema-validated query layer.

frontend error-volume reduction Reported error volume; not a measured incident-rate reduction

Improved error tagging, stack traces and escalation alongside the reduction in recorded error volume.

User-confirmed approximate reduction in recorded frontend error volume after consolidating error handling in TrackJS.

conference platform transactions

Built ticketing, CFP reviews, sponsorship pages and a workshop catalogue with Next.js, Supabase and Stripe.

Cumulative transaction volume processed by the ZurichJS Conf platform by July 2026; not profit or sponsorship revenue. Source