Antom | Knowledge Source

Unlock the Revenue Potential of Every Payment with AI

Written by Antom | Jun 1, 2026 9:32:55 AM

Author: David Guo

Rewriting Payment Success Rates with AI

Payment failures remain one of the largest sources of preventable revenue loss in digital commerce.

According to research published by LexisNexis Accuity, failed payments cost global businesses an estimated $118.5 billion annually. Research from Primer further suggests that nearly 25% of payment failures are caused by recoverable soft declines rather than insufficient funds or confirmed fraud. Many of these failures originate from suboptimal decisions across routing, authentication, retries, and issuer communication.

As payment ecosystems become increasingly interconnected, improving authorization performance requires more than isolated rule-based optimisation.

Traditional payment systems often rely on static rules, similar to navigating with a paper map. While effective in predictable environments, modern payment ecosystems constantly shift due to issuer behavior, regional regulations, network latency, and evolving fraud patterns. In complex cross-border scenarios, payment optimisation increasingly depends on systems that can continuously observe conditions in real time and dynamically determine the most effective transaction path.

Antom’s Card Revenue Booster was designed to address this challenge.

One Engine Coordinating Five Critical Payment Decisions

Antom’s Card Revenue Booster is an AI-powered payment optimisation engine designed to improve authorization performance across the card payment lifecycle.

Within milliseconds of a transaction being initiated, the engine continuously coordinates five optimisation capabilities:

    • Smart Routing
    • Adaptive Messaging
    • Authentication Intelligence
    • Active Retry
    • Credential Lifecycle Management

In modern payment environments, transaction outcomes are rarely determined by a single decision point. Routing, authentication, issuer messaging, retries, and credential management continuously influence one another throughout the payment lifecycle. Optimising authorization performance therefore requires these decisions to operate as part of a coordinated system rather than as isolated workflows.


The Core Differentiator: Antom’s Transformer Model

At the core of Card Revenue Booster is an AI-powered Transformer model purpose-built for dynamic payment environments.

Traditional rule-based systems perform well in deterministic scenarios with clearly defined parameters. However, modern payment environments are highly dynamic, where issuer behavior, fraud controls, and transaction conditions continuously evolve throughout the payment lifecycle.

As a result, payment outcomes are rarely determined by a single decision point.

Issuer responses to earlier retries may influence subsequent authorization behavior, particularly when repeated attempts trigger issuer-side velocity or fraud controls. Authentication outcomes can also affect downstream routing performance, as issuers frequently apply different authorization logic depending on prior authentication context or exemption history. At the same time, fraud mitigation strategies can shift rapidly in response to emerging attack patterns, requiring optimisation systems to continuously adapt decision logic in near real time.

Rather than relying solely on predefined logic, the model analyses relationships across merchants, issuers, devices, customer behavior, and transaction history simultaneously. This allows the platform to identify patterns and optimisation opportunities that are often difficult to detect through static configurations or isolated transaction analysis.

The result is a more adaptive payment optimisation framework capable of improving authorization performance while reducing unnecessary declines and operational inefficiencies.

In practice, this Transformer architecture powers three core characteristics that enable smarter payment decisions across the transaction lifecycle.

Three Capabilities That Drive Smarter Payment Decisions

Holistic Understanding

The model combines signals across issuers, merchants, devices, and cardholders to build a comprehensive view of each transaction. By analysing how multiple variables interact, the system evaluates payments with greater accuracy and contextual awareness.

Unified Decisioning

Built on a shared optimisation architecture, the model coordinates routing, adaptive messaging, authentication, retries, and lifecycle management within a single decisioning layer. Insights generated throughout the transaction lifecycle continuously improve downstream payment handling and recovery strategies.

Real-Time Adaptability

The model continuously learns from transaction sequences, issuer responses, and evolving payment patterns, enabling the system to respond more effectively to shifting payment behaviors and operational conditions over time.

Together, these capabilities enable more connected and effective payment optimisation across the entire payment lifecycle.

Payment Optimisation in Practice

Consider a customer in Washington, US, purchasing a luxury designer item from an independent merchant based in China using a JPMorgan Chase-issued credit card.

The transaction is high-value, cross-border, and involves multi-currency settlement.

From the customer’s perspective, the checkout experience should feel seamless. Behind the scenes, however, the payment requires multiple decisions to be completed within milliseconds, including authentication handling, issuer communication, routing selection, and retry evaluation.

This is where Antom’s Card Revenue Booster comes into play.

The following two examples demonstrate how different optimisation capabilities within Card Revenue Booster work together to improve authorization outcomes throughout the payment lifecycle.


Authentication Intelligence

Effective authentication should strengthen trust without introducing unnecessary friction.

When a transaction is initiated, Antom’s Transformer model constructs a multidimensional view of the payment by analysing behavioral, transactional, and issuer-related attributes together. Rather than evaluating isolated indicators independently, the system assesses the broader transaction profile to distinguish legitimate customer activity from elevated risk.

In this scenario, the platform identified multiple high-confidence trust indicators, including familiar device usage, consistent location behavior, and strong historical payment activity.

Based on this assessment, the engine dynamically applied a 3DS Data-Only authentication flow, enabling richer authentication data to be shared directly with the issuer while minimising unnecessary customer verification steps.

Instead of treating the transaction as suspicious solely because of its high-value or cross-border characteristics, the system recognised it as a legitimate purchase from a trusted customer.

Active Retry

A declined payment does not always mean the transaction should fail. Issuer-side instability, temporary network interruptions, or dynamically changing fraud controls can often trigger recoverable soft declines. In some cases, excessive retry activity within short time windows may itself increase issuer risk sensitivity, causing legitimate transactions to be interpreted as potential fraud or account testing activity.
In this scenario, although the transaction was assessed as legitimate and submitted successfully for authorization, the initial attempt was declined due to temporary issuer-side instability during processing. Rather than immediately treating the payment as failed, the model evaluated the issuer response alongside historical recovery patterns, transaction context, and issuer behavior to determine whether a retry was likely to succeed. Based on these insights, the platform dynamically adjusted retry timing, routing paths, and processing logic before reattempting the transaction. This adaptive retry approach helps improve recovery rates while reducing unnecessary retry activity, issuer friction, and scheme fees. 

Measurable Business Impact

Card Revenue Booster is designed to transform payment optimisation from fragmented, experience-driven operations into a standardised capability powered by data and machine learning models.

This foundation creates value across three dimensions:

Revenue Growth

By increasing payment success rates without additional marketing spend, merchants can achieve 4–8% GMV growth.

Cost Optimisation

Intelligent routing and orchestration help merchants optimise payment processing efficiency and reduce overall acceptance costs.

Operational Efficiency

Card Revenue Booster accelerates cross-border expansion by significantly reducing new market onboarding timelines, compressing deployment cycles from up to four weeks to less than 48 hours.

These improvements extend beyond direct revenue impact, contributing to better customer experience, stronger retention, and more efficient global payment operations.

Built for Global Payment Complexity

Navigating global payment complexity requires more than isolated optimisation tools. It requires a combination of intelligent payment infrastructure, operational expertise, and continuous adaptation across issuers, markets, authentication requirements, and evolving network conditions.

Built on Antom’s global payment infrastructure, Card Revenue Booster is supported by dedicated payment specialists who assist merchants with optimisation strategy, issuer engagement, and operational execution. This combination of intelligent infrastructure and operational expertise helps merchants improve authorization performance, reduce operational complexity, and scale global payment operations with greater efficiency, resilience, and confidence.