The biggest crime story in America right now isn't a crime wave. It's a crime collapse. Murder in the United States fell to roughly 14,000 in 2025 — a level not seen in decades — after dropping at the fastest rate ever recorded over the past three years. The country is safer by almost every national metric. And yet most Americans have no idea it's happening.
The Numbers Are Remarkable
To understand how extraordinary this decline is, you need to understand where we came from. In 2020, the United States experienced the largest single-year increase in murder ever recorded — a surge driven by pandemic-era disruption, mass social upheaval, and the near-complete collapse of court systems and social services. The national homicide rate jumped roughly 30% in a single year. Cities that had been trending downward for a decade reversed course overnight.
Then, starting in 2022 and accelerating sharply through 2023, 2024, and 2025, something remarkable happened: murder began falling at a pace that also had no precedent in modern crime data.
According to analysis by Jeff Asher, co-founder of AH Datalytics and one of the country's most rigorous crime data analysts, three major independent tracking organizations all converged on substantial declines in 2025: the Real-Time Crime Index reported murder down 18.5%, the Council on Criminal Justice reported a 21% decline, and the Major Cities Chiefs Association showed a 19.3% drop. These aren't rounding errors — they represent thousands of lives.
The city-level picture is equally dramatic. Detroit, New Orleans, Los Angeles, Philadelphia, and Baltimore all reported their lowest murder totals since the 1960s. New York City recorded the fewest murders through May 2025 in its modern data history. These are not merely good years — they are historically anomalous ones, in the right direction.
Why Most Americans Don't Believe It
Here's the dissonance: even as murder plummeted through 2023, polling found that roughly three in four Americans believed crime had risen that year. Public perception of crime is stubbornly resistant to good news — and for understandable structural reasons.
- 📰Media coverage skews toward crime spikes. A 20% murder drop generates a fraction of the coverage that a 20% surge would. Violent incidents are inherently newsworthy; their absence is not.
- 📍Local experience doesn't always match national trends. A neighborhood that saw a spike in car break-ins doesn't feel the national murder decline. Crime is intensely local, and national averages can mask significant variation at the block and zip code level.
- 📊Data lags are significant. The FBI's official crime reports are often 12–18 months behind. By the time 2023 data was published, the public discourse had already moved on — leaving many without an updated mental model.
- 🧠Availability bias is powerful. High-profile violent incidents are memorable and emotionally resonant. Years of low-crime don't overwrite years of high-crime in people's minds.
The Other Problem: Crime Data Itself Is Often Wrong
Even with historic improvements in crime, a parallel problem has become harder to ignore: the underlying data infrastructure for measuring crime is riddled with errors. Asher has documented this problem extensively — and one case study has become something of a legend in crime data circles.
☕ The Coffee County Problem
In August 2019, Coffee County, Georgia — population roughly 25,000 — reported 7,261 burglary clearances against just 17 actual burglary offenses. That's a 42,711% clearance rate. The following month, they reported 5,000 motor vehicle theft clearances for a single offense. Those two months of data from one small county accounted for approximately 38% of all burglary clearances reported nationally that August. Similar errors have been found in Blairsville, Georgia; Oconee County; and Rochester Hills, Michigan. The takeaway: national crime statistics can be meaningfully distorted by a single agency's reporting mistake.
This isn't an isolated quirk. Law enforcement agencies across the country use dozens of different records management systems, report data at different intervals, and apply inconsistent local definitions of crime categories. The FBI's transition from its Summary Reporting System (SRS) to the National Incident-Based Reporting System (NIBRS) has improved coverage but introduced new comparability challenges during the transition period. As Asher puts it, the need to be careful when using crime data cannot be overstated — because individual reporting errors from small agencies can meaningfully distort national and even regional trends.
This is precisely why the normalization and verification layer in a well-built crime data API is not a commodity feature — it is the entire point. Raw data from police agencies is noisy, inconsistent, and sometimes wildly incorrect. The value of a curated crime data API is the work that happens between the police report and your application.
National Trends Don't Live at the Address Level
Here is the central tension that anyone building with crime data needs to understand: the national trend is remarkable and real, but it is an average — and averages can hide enormous variation.
Consider that even in cities reporting historic murder lows overall, crime is not distributed evenly. In any major city, there are neighborhoods where crime has plummeted and others where it has barely budged. A family deciding where to live, a real estate agent advising a buyer, a security team planning executive travel — none of them can act on a national trend. They need a specific address, a specific block, a specific radius.
This is the enduring value proposition of granular crime data, even in a low-crime environment. The question is never just "is crime down?" — it is "what is the crime picture at this location, over this time period, for these incident types?" That question requires address-level, normalized, verified data — not headlines.
“Murder rose at the fastest rate ever recorded in 2020 and has been falling at the fastest rate ever recorded since 2022.”
— Jeff Asher, Crime & Justice News, AH Datalytics co-founder
What a Historic Crime Decline Means for Crime Data APIs
You might expect that falling crime would reduce demand for crime data. In practice, the opposite is true — and for a few clear reasons.
The Story Isn't Over
America's crime decline is a public safety success story that deserves more attention than it gets. The analysts doing the best work on this — people like Jeff Asher at AH Datalytics, whose meticulous city-by-city tracking has documented both the scale of the drop and the messiness of the underlying data — are providing an enormous service. We should pay attention to their work.
But the lesson for anyone building with crime data is not that the problem is solved. It is that the problem is complex, local, and permanent — and that understanding it at the address level, with verified and normalized data, has never been more important than it is right now, when the gap between national narrative and local reality is so wide.
The data to close that gap exists. The question is whether the tools people use to make safety decisions are tapped into it.
Access Address-Level Crime Data
Real-time incidents · SpotScore™ safety ratings · 36-month trends · 22,000+ US cities. Normalized and verified — because raw data isn't enough.