September 9, 2026

SAR Volume Benchmarks: How Much Is Normal?

Axle Team

Key takeaways

Financial institutions filed 4.7 million SARs in FY2024, per FinCEN's Year in Review — about 2.6 million of them from depository institutions.
Original SAR filings were essentially flat year over year in calendar 2024 (3,801,691 vs. 3,809,823 in 2023, per Thomson Reuters' analysis of FinCEN data) — the first pause after a decade of growth.
There is no regulator-defined "right" number. The FFIEC manual tells examiners to evaluate your SAR decision-making process, not your count.
Raw peer comparisons mislead; normalize by accounts, transactions, and alert volume before drawing conclusions.
Most volume swings trace to threshold changes, new products, or fraud typologies — document the cause when your trend line moves.

Every BSA officer eventually fields the question behind every search for SAR volume benchmarks: "are we filing too many, or too few?" Usually it comes from the board, an auditor, or a new CCO comparing notes with a prior shop.

There is no regulator-set normal for SAR volume. The practical benchmarks: U.S. institutions filed about 4.7 million SARs in FY2024, roughly 2.6 million from depository institutions. Peer comparison only works normalized — SARs per 1,000 accounts or per million transactions — and examiners judge your decision process, not your count.

SAR volume benchmarks: the national picture

The most current official numbers come from FinCEN's Year in Review for fiscal year 2024 (October 2023 through September 2024), as reported by ABA Banking Journal:

Total SARs filed: 4.7 million
Daily average SAR filings: 12,870
SARs from depository institutions: 2.6 million
Total CTRs filed: 20.5 million

On a calendar-year basis, the Thomson Reuters Institute's analysis of FinCEN's SAR data counted 3,801,691 original SARs in 2024 against 3,809,823 in 2023 — the first year-over-year dip in roughly a decade, and effectively a plateau:

"The total number of SARs for 2024 should land somewhere in the 4.5 million to 4.6 million range, roughly the same as 2023."
— Thomson Reuters Institute, analyzing FinCEN year-end SAR statistics

Fraud remains the dominant driver: the same analysis found check fraud alone generated 682,276 SARs in 2024, and fraud categories (card fraud, ACH fraud, identity theft, account takeover) fill out the top of the list.

Why raw counts mislead

Comparing your absolute SAR count to a peer's tells you almost nothing. Filing volume is a function of business model before it is a function of program quality:

A fintech sponsor bank inherits the fraud exposure of every program it sponsors.
An MSB or crypto exchange files at a completely different rate per customer than a community bank with a stable deposit base.
Two banks with identical risk can show 3x different volume purely because one tuned thresholds last year and the other hasn't touched them since implementation.

That's why SAR volume benchmarks only become useful after normalization — and why the regulatory framework deliberately avoids setting quotas.

What examiners actually benchmark

The FFIEC BSA/AML Examination Manual is explicit that the number itself is not the exam target:

"The decision to file a SAR is an inherently subjective judgment. Examiners should focus on whether the bank has an effective SAR decision-making process, not individual SAR decisions."
— FFIEC BSA/AML Examination Manual, Suspicious Activity Reporting

And on the fear of under-filing:

"In those instances where the bank has an established SAR decision-making process, has followed existing policies, procedures, and processes, and has determined not to file a SAR, the bank should not be criticized for the failure to file a SAR unless the failure is significant or accompanied by evidence of bad faith."
— FFIEC BSA/AML Examination Manual, Suspicious Activity Reporting

What examiners will probe is whether alert volumes, case conversion, and filing decisions hang together: if your alert count doubled and SAR count didn't move, be ready to explain why. The defense is process evidence — documented dispositions, decline-to-file memos, and threshold-change records — not a count that matches some peer average.

The benchmarks worth tracking internally

Instead of chasing a national number, build a small internal panel and watch trend, not level:

Alert-to-case conversion rate. What share of monitoring alerts escalate past L1 review? A falling rate after tuning is expected; a falling rate with no tuning change needs a memo.
Case-to-SAR conversion rate. Stable programs show stable conversion. A sudden jump often means analysts are defensively filing; a sudden drop invites the under-filing question.
SARs per 1,000 active customers (or per $1M in transaction volume). This is the only form in which peer comparison is even directionally useful.
Continuing-activity SAR share. What fraction of filings are 90-day continuing reports on the same subjects?
Late-filing rate. SARs are due within 30 calendar days of initial detection (60 if no subject is identified). This is the one metric where the tolerance is effectively zero.

One observation from practice: the most diagnostic number on that panel is the continuing-activity share, and almost nobody watches it. A rising share means your program keeps re-documenting the same customers every 90 days without anyone making an exit decision — you're paying SAR-writing costs on autopilot for relationships the bank has implicitly decided to keep. Review that population quarterly with whoever owns de-risking decisions; it's usually the fastest SAR-workload reduction available, and it doesn't touch detection at all.

When your volume moves, document why

A volume shift is not a problem; an unexplained one is. Three habits keep the trend line defensible:

Annotate every threshold or rule change in the same system that tracks SAR counts, so the causal story is one query away.
Write a short memo when quarterly volume moves materially — new product, new fraud typology, tuning change, or staffing change.
Sample both sides in QA: filed SARs for narrative quality, and declined cases for whether the decline rationale holds up.

Where Axle fits

Volume questions usually become throughput problems: filings climb and the team writing narratives doesn't. Sam, Axle's SAR narrative agent, drafts audit-ready narratives from case data with consistent structure, so analysts review and approve instead of writing from scratch — and your SAR filing workflow keeps a full audit trail for QA and exams. If you're gauging whether your program is ready for that kind of automation, start with the AI readiness playbook.

FAQ

How many SARs are filed each year in the U.S.?

About 4.7 million in fiscal year 2024, per FinCEN's Year in Review — a daily average of 12,870. Calendar-year analyses show 2024 volume roughly flat against 2023 after a decade of steady growth.

Can a bank be criticized for filing too few SARs?

Yes, but the bar is process failure, not a low count. Under the FFIEC manual, a bank that followed its established SAR decision-making process should not be criticized for a decision not to file unless the failure is significant or made in bad faith. The exposure comes from missing dispositions and undocumented declines, not from filing less than a peer.

What is a normal SAR-per-customer rate?

There's no published regulatory ratio, and business model dominates the math — a crypto exchange, a sponsor bank, and a community bank will sit at very different rates with equally sound programs. Benchmark against your own trend line, normalized per 1,000 active customers, and investigate changes rather than levels.

Filing more SARs with the same team? See how Sam drafts audit-ready SAR narratives your analysts approve instead of write: https://www.axleruns.com/agent/sam

SAR Volume Benchmarks: How Much Is Normal?

More from the blog