Ecommpay has expanded its use of artificial intelligence to improve payment transaction success rates for merchants, saying the system lifted portfolio-wide transaction success rates by 2.6% in the first quarter of 2026.
Individual clients recorded improvements of between 5% and 15% after the introduction of its proprietary AI-driven performance analytics engine, which targets three common causes of payment declines: 3DS authentication failures, fraud and suspicious transaction flags, and client-behaviour errors.
Ecommpay made what it described as a substantial investment in artificial intelligence during 2025 as part of a wider push to improve checkout performance for merchants. Founded in 2012 and headquartered in London, the company operates a global payments platform offering acquiring, payment methods and processing through a single API.
The new system moves beyond the manual reviews previously carried out by analysts on relatively small samples of transactions. It analyses entire transaction data sets containing tens of thousands of payments in one pass, aiming to identify where merchants can adjust configurations to better match issuer requirements and processing logic.
The engine uses separate AI models for each category of decline. Developed in collaboration with card issuers and based on established technical standards, the models allow engineers and analysts to guide merchants on where payment settings and flows may need adjustment.
How it works
The models are also used predictively for prospective customers, allowing Ecommpay to estimate possible approval-rate improvements before a merchant goes live on its platform.
The approach reflects growing competition in the payments sector around transaction optimisation, as providers try to reduce failed payments that can translate directly into lost sales. Approval rates are a critical commercial measure for merchants, particularly in online retail and digital services, where customers may abandon purchases after a failed attempt.
Until now, much of that work relied on teams reviewing a limited number of declined transactions to identify patterns. Ecommpay said its AI tools now allow it to assess far larger volumes of data and direct technical resources to the areas likely to have the biggest effect on approval rates.
"Previously a team of analysts would manually review a sample of perhaps 50 to 100 transactions to identify pain points and potential for improvement," said Daniella Sjarki, Business Growth Specialist at Ecommpay.
"Now our AI models process entire transaction data sets, encompassing tens of thousands of individual payments in a single pass. The system instantly identifies where the highest-impact problems lie, enabling engineers and analysts to direct their efforts in partnership with our merchant customers to where they will have the greatest effect on approval rates," Sjarki said.
Ecommpay expects the gains to continue across its client base and said it is on course for a 10% improvement across the portfolio by the end of 2026. It did not disclose the size of its AI investment.
Merchant focus
The company is making the technology available as a standard part of client relationships rather than as an additional paid feature. That may appeal to merchants seeking to improve payment acceptance without building specialist in-house data science teams.
Payments providers have increasingly turned to machine-learning models to address issues such as fraud screening, authentication and transaction routing. Ecommpay's emphasis, however, is on diagnosing the source of failed transactions and recommending operational changes merchants can make.
"In a market where we believe only a small number of global payment processors are investing at this level, at Ecommpay we are making our most powerful capability a standard feature of every client relationship, not an optional extra. Our AI-powered platform enables merchants to act immediately without requiring deep technical expertise. This means merchants can see measurable uplifts in payment performance and a better customer experience," Sjarki said.