Sanctions Screening Benchmarks: Alert Volumes, False Positives, and the Push Toward Machine Learning
Survey data from 36 sanctions and AML leaders reveals the alert volumes, false positive rates, and regulatory pressures shaping compliance programs.
How does your sanctions screening program compare to your peers? For most compliance leaders, that question goes unanswered. Cross-institutional benchmarks are scarce, and without them it is nearly impossible to know whether your alert volumes are typical, your false positive rates are acceptable, or your roadmap is pointed in the right direction.
This article summarizes findings from Enigma’s State of Screening survey, which collected responses from 36 sanctions and AML program leaders at financial institutions ranging from community banks to institutions with tens of millions of customers. The data paints a clear picture: alert volumes are climbing, false positives remain stubbornly high, regulatory scrutiny is intensifying, and machine learning has moved from experiment to mainstream consideration.
Note: The data in this report comes from a fall 2019 survey. The specific percentages reflect conditions at that point in time, but the structural challenges described — high false positive rates, alert volume strain, and regulatory pressure — remain central concerns for compliance teams today.
Who Responded to This Survey
The survey reached 36 compliance leaders across a range of institution sizes and roles.
Institution Size by Customer Base
| Customer Base | Number of Respondents |
|---|---|
| Under 1 million | 18 |
| 1M – 5M | 3 |
| 5M – 10M | 4 |
| 10M – 20M | 1 |
| 20M – 30M | 4 |
| 30M – 40M | 6 |
Half of respondent institutions have fewer than 1 million customers. Roughly 38% have 20 million or more — meaning the findings span both community institutions and large financial firms.
Respondent Roles
| Title | Number of Respondents |
|---|---|
| Chief Compliance Officer | 17 |
| (Global) Head of Financial Crimes Compliance | 8 |
| BSA/AML/OFAC Officer | 3 |
| Regional Head of Financial Crimes Compliance | 2 |
| Chief Risk Officer | 1 |
| Other | 5 |
Chief Compliance Officers and Chief Risk Officers together represent 50% of respondents. The rest come from financial crimes compliance and AML leadership roles. Roughly half of all respondents employ more than 100 full-time employees dedicated to sanctions, PEP, and negative news screening and investigations.
Key Findings at a Glance
- 42% of institutions are experiencing higher sanctions alert volumes than a year ago
- 76%+ false positive rates reported by more than half of respondents across both transaction and customer screening
- 70% of institutions agree they have experienced increased regulatory scrutiny of their sanctions compliance program
- 36% strongly agree their program experienced increased regulatory scrutiny
- 81% of institutions are exploring, ready to deploy, or already using machine learning for sanctions screening
- 17% have already built or deployed machine learning capabilities
Alert Volumes Are Getting Worse, Not Better
The most immediate problem compliance programs face is volume. Alert volumes are primarily one-directional: 42% of institutions report higher sanctions alert volumes — covering both customer and transaction screening — compared to one year prior. Only 14% of institutions definitively reduced their sanctions alerts in the same period.
That means institutions have invested in alert reduction, yet nearly half are still worse off year over year.
Transaction screening generates the most strain. Nearly half of institutions report transaction screening alert rates of 11% or higher. Put that in context: 25% of institutions process at least 1 million transactions annually, and 33% process more than 10 million. At those volumes, an 11% alert rate translates to hundreds of thousands — potentially millions — of sanctions alerts requiring review each year.
Transaction Screening Alert Rates
| Alert Rate | Number of Respondents |
|---|---|
| Under 5% | 10 |
| 5% – 10% | 11 |
| 11% – 15% | 10 |
| 16% – 25% | 3 |
| Over 25% | 2 |
Customer screening alert rates trend lower, with 44% of respondents reporting alert rates of 3–5% or higher — still substantial given the scale of most customer bases.
False Positive Rates Reveal a Deeper Accuracy Problem
High alert volumes become a crisis when combined with high false positive rates. Across both transaction and customer screening, the data shows false positives are not a marginal issue — they are the norm.
For transaction screening:
- 53% of respondents report false positive rates of 76% or higher
- About a third of respondents find more than 91% of alerts to be false positive after an initial review
Transaction Screening False Positive Rates
| False Positive Rate | Number of Respondents |
|---|---|
| Under 50% | 11 |
| 50% – 75% | 6 |
| 76% – 85% | 4 |
| 86% – 90% | 5 |
| 91% – 93% | 2 |
| 94% – 97% | 4 |
| Over 97% | 4 |
For customer screening, the picture is comparable: 50% of respondents report false positive rates of 76% or higher.
Half of all respondents find more than 76% of their alerts to be false positive after an initial “level 1” review. These numbers do not indicate edge cases in otherwise functional programs — they reveal deep accuracy challenges baked into current screening processes.
The staffing implications are direct. To manage alert rates, 61% of program leaders plan to hire full-time employees or bring in external contractors in 2020. Alert volume is not just a technology problem; it is driving real cost and headcount decisions.
Regulatory Scrutiny Is Rising Across All Institution Sizes
Alert volumes exist in an environment of sharpening regulatory attention. Nearly 70% of institutions agree they have experienced increased regulatory scrutiny of their sanctions compliance program over the past year. Of those, 36% strongly agree.
The breakdown of responses:
| Agreement Level | Share of Respondents |
|---|---|
| Strongly agree | 36% |
| Agree | 36% |
| Somewhat agree | 19% |
| Neither agree nor disagree | 6% |
| Disagree | 3% |
This pressure is not limited to large institutions. The survey found no correlation between institution size and perceived regulatory scrutiny — all sizes reported feeling this strain. Notably, nearly all of the institutions planning to add full-time employees in the coming year also identified as experiencing greater regulatory pressure.
The regulatory context helps explain the urgency. The Office of Foreign Assets Control issued fines totaling a record $1.3 billion in 2019. As one survey respondent put it: “Compliance functions are certainly undergoing increased scrutiny from regulators, particularly in the light of many scandals at larger financial institutions.”
Growing alert volumes may themselves reflect heightened regulatory pressure. Institutions may be widening screening thresholds as a way to demonstrate the rigor of their programs, accepting higher false positive rates as the price of demonstrating thoroughness to examiners.
Machine Learning Has Moved From Fringe to Mainstream Consideration
Program leaders are looking to technology — specifically machine learning — as the primary lever for addressing false positives and alert volume. 81% of respondents are somewhere on the machine learning adoption spectrum.
Machine Learning Adoption Stages
| Stage | Share of Respondents |
|---|---|
| Have deployed home-grown machine learning | 8.5% |
| Have deployed vendor-provided machine learning | 8.5% |
| Ready to build or buy machine learning | 8.5% |
| Educating self, team, and/or organization on ML as a possible solution | 47% |
| Have not considered machine learning | 19% |
The largest group — 47% of respondents — is actively educating themselves and their organizations about machine learning as a solution. An additional 8.5% say they are ready to build or buy. Combined with the 17% who have already deployed ML capabilities (home-grown or vendor-provided), this represents broad momentum toward adoption.
Multiple program leaders listed machine learning as the single change they would make to their current sanctions compliance program or technology. Several specifically cited the desire to use machine learning in initial screening to reduce false positive rates and lower alert rates.
This appetite has regulatory backing. In December 2018, five regulatory agencies including FinCEN released a joint statement encouraging financial institutions to adopt more advanced technologies: “These innovations and technologies can strengthen BSA/AML compliance approaches, as well as enhance transaction monitoring systems. The Agencies welcome these types of innovative approaches to further efforts to protect the financial system against illicit financial activity.”
Despite the clear interest, only 17% of institutions have actually deployed machine learning — meaning most institutions that reported higher alert volumes over the past year have not yet benefited from ML-powered accuracy improvements. Alert volumes and related strain are likely to ease materially as more institutions successfully deploy these capabilities.
What the Benchmarks Tell You
The combined picture from this survey is one of significant strain. Screening program leaders face increasing alert volumes, persistently high false positive rates, and intensifying regulatory attention — all at the same time.
The shift toward machine learning reflects where the industry believes relief will come from. Reducing false positives at the initial screening stage addresses alert volume, staffing cost, and regulatory demonstrability in a single move.
For program leaders, these benchmarks offer a few practical points of reference:
- If your transaction screening false positive rate is above 76%, you are in the majority — but that does not mean it is acceptable, and peers are actively pursuing ML to close the gap.
- If your alert volumes are rising, you are not alone. Only 14% of institutions reduced alerts year-over-year, even with investment in reduction efforts.
- If you are under increased regulatory scrutiny, so are nearly 70% of your peers, regardless of institution size.
- If you have not yet deployed machine learning, the 81% of institutions actively exploring or implementing it suggest the window for early-mover advantage is narrowing.
Reduce Sanctions Alert Volume Without Increasing Risk
Enigma’s approach to sanctions screening is built around the accuracy problem at the root of these benchmarks. Better data leads to higher-quality matches, which means fewer false positives to review, lower alert volumes, and a more defensible program posture with regulators.
Learn more about how Enigma helps compliance teams improve screening accuracy on our KYB product page, or contact us to discuss your program’s specific challenges.