Merchant Onboarding: Compliance Steps From Application to Go-Live
Merchant onboarding from application to go-live: KYB, underwriting, website review, MATCH, VAMP, transaction laundering, and monitoring after boarding.
TL;DR: ACH fraud is any fraudulent debit or credit moved over the ACH Network, from unauthorized debits to credits induced by business email compromise. Since June 22, 2026, Nacha's fraud monitoring rules have bound every ODFI, every RDFI, and every non-consumer originator and third party in the origination chain, regardless of volume. In the Federal Reserve's 2026 Risk Officer Survey, 97 percent of institutions reported unauthorized debit fraud in the past year, and 41 percent said account holder scams over ACH were increasing, so detection has to cover both directions of the payment.
ACH fraud is the use of the ACH Network to take or redirect money through entries that were unauthorized or obtained by deception. The network has two directions, and fraud behaves differently in each. A debit pulls money out of an account, so the victim's bank holds the strongest evidence and the return system offers a remedy. A credit pushes money to a receiver, so the originating customer often approved the payment and the money has already reached someone else.
Two parties anchor every detection program. The Originating Depository Financial Institution, or ODFI, sends entries on behalf of its originators. The Receiving Depository Financial Institution, or RDFI, posts them to receivers' accounts. The ODFI knows the originator's normal files, and the RDFI knows the account the money lands in.
The Federal Reserve's 2026 Risk Officer Survey shows the shift. Unauthorized debits remain nearly universal but were described as increasing by only 13 percent of institutions. The sharper growth is in account holder scams, where the customer is deceived into sending the money.
Here an attacker gains access to a business's banking portal or a third-party platform and creates ACH credit batches, often payroll or vendor files, directed to accounts the attacker controls. Login signals are covered in Sphinx's guide to account takeover fraud detection. On the ACH side, the evidence sits in the file: new receivers in an otherwise stable batch, a changed effective date, an unusual shift to Same Day, or limits raised just before submission.
Business email compromise produces an ACH credit that a genuine employee authorizes because an impersonator supplied new payment details. Vendor bank-change requests and payroll diversion are the standard forms. Controls for single high-value transfers are covered in wire fraud detection for banks. ACH differs in that the diverted payment can be one line in a large, routine batch.
Every credit-push scheme needs a landing account. On ACH, that typically looks like a newly opened or dormant consumer account receiving business-originated credits, payroll from an employer that does not match the customer's profile, or credits from several unrelated originators followed by rapid withdrawal. Mule account investigation covers the follow-through. The Federal Reserve survey notes that, in many cases, institutions learned of a mule account only after its funds were depleted.
The Nacha fraud monitoring rules require covered parties to establish and implement risk-based processes and procedures reasonably intended to identify entries suspected of being unauthorized or authorized under false pretenses. They took effect in two phases, and both have now passed.
The rule text carries a June 19, 2026 date for Phase 2. Because that day is a federal holiday, Nacha set the practical compliance date as Monday, June 22.
Four details get misread:
Related changes touch the same workflow. Company Entry Descriptions for PAYROLL and PURCHASE took effect March 20, giving RDFIs a cleaner field to monitor. On September 18, 2026, Nacha removed the 5:00 p.m. receipt condition, so non-Same Day credits must be available at 9:00 a.m. local time on Settlement Date.
Every return carries a reason code, and read together those codes reveal originator behavior, account testing, and claim abuse. The windows below come from EPCOR's return reason code matrix, published by a Nacha direct member.
The thresholds are ceilings, and they are poor alarms. Nacha limits unauthorized returns to 0.5 percent of an originator's debits, administrative returns to 3 percent, and overall returns to 15 percent. Nacha reports that the average unauthorized debit return rate for the network is 0.05 percent. An originator running at 0.4 percent passes the rule while operating at eight times the network average.
The windows create a blind spot. A consumer debit can come back as unauthorized for 60 days, so an originator's return rate lags its fraud by up to two months. Grouping returns by origination week shows a problem that a trailing average hides. Business accounts have the opposite exposure: R29 has only two banking days, so a company that reconciles weekly can lose the return before it notices the debit.
R10 is the code that cuts both ways. Nacha advises RDFIs to calculate their own unauthorized return rate and investigate a rate around 0.1 percent, since fraudsters may target an institution that does not research claims. Useful checks include whether the originator has debited this receiver before and whether the entry name matches the account. Nacha also cautions that this work should not deter legitimate claims.
Nacha's October 1, 2024 amendments added the tools that make these signals actionable. An RDFI may use R17 with the descriptor QUESTIONABLE to return an entry it believes is fraudulent. An ODFI may request a return for any reason under R06 while indemnifying the RDFI, and since April 1, 2025 the RDFI must respond within 10 banking days. When returning a consumer debit as unauthorized in the extended window, the RDFI must act by the opening of the sixth banking day after completing its review of the signed statement.
An ODFI's advantage is history. Its monitoring should compare each file to the originator's own baseline, and treat third-party senders as a separate risk because their files mix many underlying clients.
An RDFI's advantage is knowledge of the account. Nacha's own list of risk-based factors includes transactional velocity, SEC code mismatches with account type, account age and average balance.
An RDFI that flags a credit has options. It may return the entry, contact the ODFI, or use the funds availability exemption for credits it suspects were originated under false pretenses. Nacha is explicit that an RDFI cannot delay availability merely because it has not screened a credit, and Regulation CC still applies.
Nacha notes that monitoring aligns with existing suspicious activity obligations, and it encourages communication between compliance, operations, product and relationship teams. Institutions that keep ACH returns in operations and transaction monitoring in compliance will struggle with that. A confirmed case can trigger a SAR on either side of the payment, and the FinCEN SAR lists ACH among its fraud categories. Sphinx's guide on when to file a SAR covers thresholds and timing.
Sphinx investigates alerts inside existing systems. It does not score transactions, block or hold payments, return entries, or connect to bank accounts, and an institution still needs the monitoring processes the Nacha rules require.
When a monitoring tool raises an ACH alert, Sphinx's AI agents gather the account profile, return history and counterparty context, apply the team's SOPs, and produce a traceable recommendation. The team keeps final approval. Sphinx reports a 3.2-minute median case across its deployments.
ACH fraud is the use of the ACH Network to move money through unauthorized or deceptively obtained entries. The main patterns are unauthorized debits, account takeover origination, BEC-driven credits, and mule receipt.
No. Nacha states that the fraud monitoring rules do not require review before processing or posting. Covered parties must have risk-based processes reasonably intended to identify fraudulent entries, and must review those processes at least annually.
Yes. All ODFIs have been covered since March 20, 2026, regardless of volume. All RDFIs have been covered for credit monitoring since June 22, 2026, when the 10 million entry threshold fell away.
A mid-size bank should check for originator-level baselines, new-receiver logic, return analytics by code and origination cohort, inbound credit monitoring against account profiles, and a documented case record for each alert.