We spoke to leaders from across the payments ecosystem to explore where revenue is really being lost, where friction still exists, and whether AI can meaningfully close the gap.
Contributors include:
- Tony Craddock, Director General of The Payments Association
- David Birch, author and advisor on digital financial services
- Jon Reynolds, Head of Product at Access PaySuite
- Sandra Mianda, CEO of Paypr.work
- Chris Jones, Managing Director at PSE Consulting
- Julie Taylor, Head of Fundraising Operations and Improvement at King’s College Hospital Charity
Is “hidden revenue” a genuine issue for SMEs?
Hidden revenue refers to income that leaks out through failed transactions, payment abandonment, manual errors and inefficiencies that are not always visible in standard reporting.
Sandra Mianda believes it remains a major blind spot: “Payments are still traditionally treated as a cost centre, so businesses tend to focus on fees and approval rates. But a lot can go wrong between customer intent and settlement. Declines happen for many different reasons, and the financial impact can be far greater than headline data suggests.
As such, there’s a significant opportunity in failed transactions. The ecosystem merchants operate in is complex – multiple payments service providers, cross-border activity, different currencies – which means performance data is fragmented across systems. Understanding that data and pinpointing where improvements can be made is a major opportunity.
Seeing similar patterns in the charity sector, Julie Taylor commented: “It’s common for donors to abandon donations partway through, for people to miss payments, or for their first payment to fail. We’ve worked to map that landscape and identify the opportunities.
“For high-value failed payments, for example, we follow up with personal communication to help supporters complete their donation. Where appropriate, we contact donors to discuss issues such as payment amounts or preferences, including cancellation if that is their choice. This is important as we still incur bank and processing charges on failed payments, even when no income is generated.”
Jon Reynolds says that the issue is often underestimated because it appears incremental: “In many organisations, small failure rates are treated as operational noise. But when you apply those percentages across recurring billing or subscription models, the cumulative financial impact is massive. The challenge is figuring out why they have failed, whether they should be retried, rerouted or escalated, and how that affects overall cashflow performance.”
Where does friction cause the most revenue leakage today?
Chris Jones points to structural complexity across the payments ecosystem: “Payments is a network business, which makes it difficult to attribute issues to any single participant. Across the value chain – issuers, card schemes and acquirers – transactions are often declined for risk or fraud reasons, sometimes unnecessarily because the issuer lacks sufficient context. When better data is shared to provide that context, approval rates can improve.
“But friction doesn’t only sit within the external payment ecosystem. It also exists within internal processes such as reconciliation, refund management and dispute handling. Bringing those processes into a single, unified experience can significantly reduce operational strain.”
Some friction is unavoidable, particularly where regulation is concerned. There is necessary friction, designed to protect consumers and reduce risk, and then there is unnecessary friction that stems from poor design or avoidable complexity. Revenue loss can arise from something as simple as a poorly designed customer journey across devices or interfaces. Consumers make mistakes – entering incorrect details or abandoning a transaction midway through – and some still lack confidence in digital systems. Those behavioural factors, combined with avoidable process friction, contribute directly to revenue leakage.
Jon Reynolds believes that many of these inefficiencies compound quietly: “What we often see is not a single dramatic failure, but a series of small inefficiencies – fragmented reporting, suboptimal retry logic, limited visibility across payment types – that together create meaningful revenue leakage.
“Without consolidated insight, it becomes difficult to distinguish between unavoidable friction and preventable loss.”
Where can AI make the biggest difference in payments and cashflow?
Tony Craddock believes AI’s impact will depend on how creatively the industry develops use cases: “There are opportunities in cross-border payments, payroll optimisation, merchant acceptance and treasury functions, including interbank settlement and international fund distribution.
“Two areas stand out. The first is programmability – adding conditions to digital tokens so that funds can only be used under certain circumstances. This is similar to smart contracts and opens up many possibilities.
“The second is micropayments. AI could enable charging in very small increments – for example, paying for electricity in small usage units rather than monthly billing. While the current system works, there may be circumstances where granular charging makes sense. Combining AI with digital currency could make this viable at scale.”
For David Birch, AI’s promise is that it can identify patterns and trends that humans would not necessarily see or program for.
AI can uncover hidden connections in payments data. You do not need dramatic improvements to see impact. For example, understanding why transactions are declined and when to retry them can significantly improve outcomes. Even a small reduction in decline rates can make a meaningful difference to the bottom line. Hidden revenue matters not just because we now have the technology to address it, but because marginal gains can produce substantial financial returns.
Sandra Mianda stresses that we are already seeing productivity gains from better data processing, and the real step forward is interpretation – understanding what the data means in context and turning it into intelligent decisions.
She said: “Take declines as an example. Different players interpret decline signals differently. Assessing whether to retry, reroute or trigger alternative authentication is complex. AI can help design flows that go beyond automation and intelligently manage what happens after a decline, ideally improving conversion without adding friction.”
Jon Reynolds added: “In payments, small percentage improvements – better retry timing, more intelligent routing, earlier identification of at-risk accounts – can materially improve cashflow predictability. For SMEs operating with tight margins, marginal gains compound quickly.”
Thinking about AI, what should businesses prepare for over the next two to three years?
David Birch commented: “If we are moving toward agentic commerce, AI agents will need new payment mechanisms. It is one thing for an AI to pay an airline on my behalf; it is another for one AI to pay another AI for a service. AI agents cannot currently hold bank accounts, so developers often use stablecoins for machine-to-machine payments.
“Whether that is the optimal solution or simply the only practical one today is unclear. But there may be pressure for entirely new forms of payments, including micropayments. Stripe and others are investing in micro-billing models. The intersection between AI and payments could drive new infrastructure.”
AI has evolved rapidly and is increasingly tied to trust and decision-making… payments is a regulated environment, so attention must be paid to how AI operates within trust and compliance frameworks.
Julie Taylor emphasises business readiness: “AI offers opportunities to predict behaviour more proactively and transform processes such as reconciliation between payment systems, banks and finance platforms. For charities, readiness is key. Investment in skills and expertise will determine whether organisations can harness these changes or fall behind.”
And how does this translate into real-world solutions?
With Access PaySuite recently launching a new payment platform with AI built in, we asked Jon Reynolds how this thinking is being applied in practice, and how its new platform is helping SMEs manage hidden revenue and payment friction.
He explains: “Our focus is on bringing fragmented payment data together and surfacing the patterns behind declines and failed transactions. That means helping finance teams understand why payments fail, identify which accounts are genuinely at risk, and optimise retry and recovery strategies without increasing manual workload. It’s about moving from reactive management to proactive optimisation, so businesses can protect predictable income more effectively.”
What emerges from these perspectives is that while AI is unlikely to remove every point of friction from payments, it may give businesses better tools to identify inefficiencies and improve outcomes. For SMEs, the biggest opportunity may simply be turning payment data into clearer insight and acting on it.
FAQs
What is the ‘hidden revenue gap’ and why does it matter for SMEs?
The hidden revenue gap refers to revenue lost through failed transactions, checkout abandonment, silent churn, and payment‑related admin that never appears clearly in financial reporting. For SMEs, this gap can total tens of thousands, or even millions, each year, directly impacting growth.
Why are payment failures so difficult for SME finance teams to detect?
Because payment data is scattered across multiple providers, dashboards, and internal teams, no single function sees the full customer‑to‑settlement journey. This fragmentation makes it hard to diagnose why payments fail and where revenue is being lost.
How much money do UK SMEs typically lose to failed transactions and payment friction?
Research shows that nearly half of SMEs lose between £5,000 and £100,000 annually due to failed payments, customer churn, and manual recovery costs, with around 8% losing £1 million or more each year.
Why are 95% of UK SMEs considering AI to address payment‑related revenue loss?
SMEs increasingly recognise that traditional reporting tools can’t reveal hidden patterns in declines, churn, or checkout friction. AI can analyse complex data signals, predict failure risks, and surface insights that help businesses recover revenue automatically and more efficiently.
How can AI help businesses reduce failed payments and recover hidden revenue?
AI can optimise retry strategies, identify decline reasons, flag at‑risk donors or customers, and highlight friction in checkout experiences. It uncovers patterns humans can’t easily see and enables proactive intervention to recover revenue before it’s lost.