Contest · Experimentación

Para IA real y guardar tests

Infinure

Inteligencia de experimentación

La experiencia completa está en desktop.

infinureCockpit de experimentación

Importar

Importar

Trae backlogs de tests y paquetes de experimentos a Contest — desde nuestro playbook, desde un programa previo o desde tus herramientas de planificación.

Importar backlog

Playbook de Infinure

Experimentos curados de nuestros proyectos con bancos digitales de LATAM. Selecciona los que encajen con este cliente — llegan como borradores, listos para adaptar y auditar.

Single-Screen Onboarding vs Multi-Step WizardA/B Classic~12% esperadoOnboarding completion is the highest-leverage funnel in every digital bank Infinure has worked with, and screen-transition drop-off is the most consistently reproducible finding across clients. This is the recommended first test for any new engagement because the effect is large, fast to detect, and builds immediate credibility for the program.Guided Camera Capture for KYC Document UploadA/B Classic~25% esperadoKYC document rejection is the single largest silent killer of account openings — users who fail once rarely retry, and most banks never instrument the rejection reasons. Guided capture has produced double-digit acceptance lifts at every client where Infinure has run it.Pre-Filled Credit Application from Existing Customer DataA/B Classic~20% esperadoCredit products carry the highest revenue per conversion in retail banking, so completion-rate lifts here dominate the economics of the whole program. Asking existing customers to re-enter known data is the most common and most fixable friction pattern Infinure encounters.Payment Reminder Timing × Copy FramingMultivariate~8% esperadoTiming and framing interact — the best copy at the wrong moment underperforms — so Infinure recommends testing them as a proper 2×2 factorial rather than sequentially. Requires a large recurring-bill segment; confirm sample size covers all four cells before launch.Savings Round-Up — Long-Term Balance ImpactHoldout~15% esperadoRound-up features show strong Day-7 numbers at every bank that ships them, but Infinure has seen the effect halve by Day 60 at more than one client — only a holdout separates durable balance growth from launch novelty. This test doubles as the client team’s introduction to holdout methodology.In-App Card Activation at Delivery NotificationA/B Classic~18% esperadoUnactivated cards are sunk acquisition cost — the bank has paid for issuance and delivery and earns nothing until first transaction. Timing the prompt to the delivery event is the cheapest activation lever Infinure knows, since it requires no incentive spend.Dormant User Reactivation — Concrete Benefit vs Generic Win-BackA/B Classic~25% esperadoReactivating a dormant customer costs a fraction of acquiring a new one, and dormant books at LATAM digital banks routinely exceed 30% of accounts. Across engagements, concrete-value framing has beaten generic win-back copy every time Infinure has tested it.Upfront Fee Breakdown Before Transaction ConfirmationA/B Classic~8% esperadoCounter-intuitively, showing more fee detail usually increases completion — opacity, not price, is what kills trust at the confirmation step. Fee display also carries consumer-protection disclosure requirements in most LATAM jurisdictions, so this test earns compliance goodwill; route final copy through legal review before launch.Biometric Login Enrollment Prompt — Progressive RolloutProgressive Rollout~30% esperadoBiometric users log in more often, abandon fewer sessions, and generate fewer credential-reset support tickets — but anything touching authentication must ship behind staged gates with documented rollback, which is why Infinure mandates the progressive rollout type here. Confirm regulatory posture on authentication changes before Stage 1.Referral Incentive Structure — Two-Sided vs One-Sided RewardA/B Classic~20% esperadoReferral is the cheapest acquisition channel a digital bank has, and the reward structure — not the reward size — is what Infinure consistently finds under-optimized. Holding total budget constant makes this a pure mechanics test with a clean cost-per-acquisition readout.

Desde archivo

Carga un programa histórico (JSON). Los tests con resultados por variante llegan como completados y alimentan los benchmarks de tests relacionados.

Descargar archivo de muestra (programa Banco Andino 2025)

La importación CSV está en el roadmap — JSON cubre fidelidad completa, incluidos los resultados.

Desde planilla

Versión completa

Tests Aprendizajes Impacto Importar Ajustes