Recent technological breakthroughs have given merchants more payment optimization options than ever before. However, the increasing complexity of the landscape makes it challenging to identify value and create opportunities while keeping expenses down.
In a recent Javelin Strategy & Research report, 2025 Merchant Payment Trends, Don Apgar, Director of Merchant Payments, and James Wester, Co-Head of Payments, examined the impact of three emerging trends shaping the merchant experience: artificial intelligence, the fintech bubble, and the shift toward value over cost.
Saving a Dime
Merchants are increasingly focused on payments performance, constantly seeking ways to optimize their payments operations. They closely track their organization’s metrics and often employ dedicated staff or vendors to monitor payments activities.
In pursuit of lower costs, many merchants have turned to payment orchestrations platforms. However, in some cases, the expense of these solutions outweighs the intended savings.
“If you’re running on a payment-orchestrated platform and you’ve got two or three processors, you may have somebody on the finance team whose job title is analyst, but in reality, they spend 100% of their time working on payments-related reconciliation and cost allocation,” Apgar said. “Also, you may have added a vendor who does loyalty or risk reduction or some other ancillary process that only applies when a transaction has certain characteristics.”
When multiple processors and vendors are introduced, the added complexity makes it difficult for merchants to ascertain the true cost of payments.
“It’s the old adage: ‘I saved a dime, but it cost me a dollar to do it,’” Apgar said. “I optimized authorization, and I sold an extra $100 in goods, but it cost me $1,000. I think there’s going to be a focus on drilling down in the merchant organization and asking, ‘What is the real cost?’ The answer is always somewhere in the middle of a completely deconstructed operation with multiple vendors, versus doing everything on your own.”
A Historic Bubble
The vendors that become an integral part of many merchants’ operations are often fintech companies that have sprung up in the past few years. However, these firms face stumbling blocks of their own, and an overreliance on fintech partners could create risks for merchants in the long run.
“I think we’re seeing another fintech bubble,” Apgar said. “History is repeating itself, like the dotcom bubble in the early 2000s. Investors were throwing money at every business plan that didn’t have a typo on it, and it didn’t have to make money, it just had to get market share. We’re seeing a lot of that in fintech, too. We have so many business plans that don’t have a navigable path to positive revenue, but it’s being obscured by all the headlines and the buzz.”
Some of these challenges became evident last year, exemplified by the collapse of Synapse, a fintech whose documentation lapses left banking customers unable to access $85 million of their funds.
The fallout from Synapse’s failure prompted regulators to scrutinize the role of fintechs in the modern financial system—an oversight that is set to intensify this year.
“We’ve seen a blizzard, not even a flurry, a blizzard of compliance fines come down from the Federal Trade Commission, the Office of the Comptroller of the Currency, and the Federal Deposit Insurance Commission,” Apgar said. “Everybody is worried about compliance, fraud, and money laundering. Banks can contract these things out to a fintech, but at the end of the day, the bank owns the responsibility for them.”
As regulators increase their focus on fintechs, pressure will mount on the organizations that don’t have a durable value proposition. Under this heightened scrutiny, some firms may consolidate, others may be acquired, and some may be forced to shut down entirely.
Second Movers
Despite these hurdles, financial technology has irrevocably altered the banking model, and its impact will only intensify as more financial services firms integrate AI. However, this adoption faces obstacles. While artificial intelligence is one of the most powerful and transformative technologies ever developed, it still has many imperfections.
For example, Google faced backlash after its Gemini AI engine provided inaccurate feedback on multiple occasions. There have also been instances where AI has “hallucinated,” generating fictitious information.
As organizations race to capitalize on AI’s advantages, they should be cautious about entrusting critical functions entirely to to artificial intelligence. For example, a financial institution may deploy AI to sift through millions of transactions as part of its compliance efforts—a task for which artificial intelligence is generally well-suited.
However, if AI overlooks something, fails to report an issue, or malfunctions, the bank will still be held accountable for any compliance failures.
“There’s another old adage: ‘You can spot the pioneers—they’re the ones with the arrows in their backs,’” Apgar said. “Our prediction for this year is there is going to be a second-mover advantage. I think that the folks that jump on the bandwagon early and are the first to roll out AI-driven chatbots on their website, they’re going to get a black eye because it’s not going to work.”
While there are still too many gaps to fully rely on AI, companies can’t afford to ignore it. In the coming years, organizations—and the world—will effectively help train the language learning models that power AI. As time goes on, AI will learn from its mistakes, and the technology will improve significantly in the long run.
Therefore, organizations that take a slower, more measured approach to AI implementation will be in the best position to reap the rewards when the technology is fully optimized.
“It’s certainly not too early to start mapping out how a highly functioning model could create efficiencies and potential savings,” Apgar said. “Once the technology becomes more accurate and reliable, then you’ve already got a plan. You will have a framework to evaluate the progress of the technology and say, ‘I think this model is at a point where it’s suitable for this function.’”
“When you look at it in that light, when you do implement AI, you’ve got a much higher probability of success,” he said. “Success being defined as it didn’t blow up and cost me anything.”
