The game theory of B2B payments and why checks survive

While AI is passing the bar and CPA exams, reading medical scans, writing code, and yes, probably drafting this post, billions of dollars are still moving by good old-fashioned paper checks.

On the consumer side, checks have faded mostly into the background besides the occasional one from a relative for the holidays or that landlord who just refuses to accept anything else. In B2B, they have not. Depending on the industry, 30–50% of B2B payments are still made by check.

The obvious question: why are payments still crawling through the mail instead of flowing through ACH, wires, or shiny new rails like RTP and FedNow?

It is tempting to point to habit or generational inertia. In reality, it is a simple game theory problem where each of the four parties involved — the sender, the recipient, and each of their respective banks — is acting rationally to reduce their own risk and optimize for their own convenience. 

The new breed of digital rails may be technically superior, but uneven adoption, security concerns, and good old-fashioned resistance to change means that the old ways still persist. Paper checks are obsolete, but they’re also a known quantity, considered to be “neutral” in any risk assessment.  

In fact, game theory offers the best explanation of why paper checks endure: they survive not because they are good, but because they are the one payment method that leaves every player “fine enough” given everyone else’s choices. Until someone changes the payoffs, or gets paid to coordinate the switch, the equilibrium holds and the checks keep coming.

Why checks still rule

The first non-obvious insight is that every B2B payment involves four, not two, players. On one side, you have the sender (the business sending money) and their bank. On the other, you have the recipient (the business receiving money) and their bank. Each comes to the table with its own priorities, incentives, and risk limits. 

When a sender wants to settle with a recipient, there are several rails they can choose from. The catch is that none of these players gets to choose in a vacuum. Each sender and recipient have their own motivations. Simultaneously, the banks influence the outcome with pricing, speed, and compliance to manage their own risk and cost.

On paper, digital bank rails should have killed checks by now. ACH, Wires, RTP, and FedNow are faster, more automatable, and usually cheaper to process than paper. But when you run them through each party’s preferences, the picture is less clean.

For the money sender, digital rails mean extra setup work and a sense of vulnerability: they need to collect and store bank details for every supplier, manage portals and formats. In the case of pull payments, give someone the scary ability to reach into their accounts. 

Now for the money recipient, digital rails are exactly what they want. Faster, more certain funds, and cleaner reconciliation (when they receive remittance data electronically) to handle their backoffice workflows. 

For the sender’s bank, push payments over digital rails are comfortable, since they see the balance, have clear authorization, and can screen the transaction before it goes out.

For the recipient’s bank, things are more challenging. Incoming credits are straightforward, but pull payments (ACH debits) are more stressful, because they are effectively vouching for their customer and can be liable when things go wrong.

Digital rails, in other words, are a clear win for one party, acceptable for another party, but challenging for two remaining parties. How that plays out in practice depends a lot on context and the power dynamic between the parties. We often only think of the two parties in the dynamic, but when you add the two parties' banks it becomes multiplicatively complex. 

As counterintuitive as it sounds, checks are still optimal in a surprising number of scenarios because of a structural “loophole” in how they work. A business can write almost anything it wants on a check without its bank pre-approving that specific payment. That lets checks sidestep many of the controls and limits that govern digital rails. Of course banks still run fraud analysis on checks, but the patterns, tools, and playbooks are very well understood and handled at scale.

The opportunity

The status quo won’t last forever, as the nebulous advantages of the paper check are starting to erode. 

Real-time rails like FedNow and RTP are steadily gaining adoption and raising transaction limits. They are push-only today, like wires, and it’s still an open question how pull-style use cases will be handled on top of them. In parallel, stablecoins are emerging as yet another set of rails, offering 24/7 instant transfers but bringing their own regulatory, FX, and treasury-management headaches.

Fraudsters are adopting every AI tool immediately and iterating fast. At the same time, banks and businesses are fighting fire with fire and using AI to manage compliance and fraud with far more sophistication. Account verification, anomaly detection, and behavioral risk scoring are getting better, which makes high-speed digital payments feel less scary than they used to.

Software and AI agents are also chewing through the labor and operational complexity around AR and AP. Instead of humans keying in invoices, chasing remittances, and reconciling bank files line by line, you can increasingly imagine agents that ingest contracts and POs, generate invoices, pick the right rail, attach the right remittance data, and clear exceptions automatically. 

On the other side, an AP agent can sanity-check amounts, run fraud and KYC checks, simulate cash impact, and schedule payment. When you are trying to get paid, it is not hard to picture your AR agent tapping the sender’s AP agent on the shoulder and negotiating toward an outcome that works for both sides: this date, on this rail, at this fee split, with this payment schedule.

The real opportunity is to change the payoff structure that kept checks in place. That payoff structure includes not just risk and workflow, but who gets stuck with the bill. Today, checks often feel “free” to both sender and recipient, while cards and some ACH debit products surface explicit fees to one side, even when they are cheaper for the system as a whole. 

For digital rails to truly replace checks in B2B, they need to:

• Give senders check-like control (no scary pull mandates, no endless bespoke onboarding) while still using fast, electronic rails.
  
• Give recipients speed and certainty by default, with clear, structured remittance that makes their back office easier, not harder.
  
• Make banks comfortable on both sides by wrapping rails with better verification, guarantees, and loss-sharing, so risk does not feel like a black box.
  
• Hide the complexity in software, so AP and AR teams feel like they are pressing “pay” and “get paid,” not designing a payments product or BPO from scratch.

Whoever can package rails, risk, and workflow in a way that makes “all digital” the locally optimal choice for each party, not just the theoretically best option, has a huge opportunity. Replacing checks will not just eliminate inefficiency, it will reshape how businesses pay and get paid and who gets to own the network and data that sit in the middle.