How Civic Informer Works
Our data sources, processing methods, and editorial standards — documented and open.
Where Our Data Comes From
Official Government Records Only
Civic Informer ingests data exclusively from official police department records. We do not use third-party crime aggregators, social media posts, news reports, or crowd-sourced tips. Every data point published on the platform is traceable to a single, first-party government source.
Every statistic we publish originates directly from the department's own records, with no intermediate sources.
How We Source Data
Civic Informer uses two sourcing models depending on the city:
Public Log Integration
Some departments publish a public activity log or incident feed on a city or department website. Civic Informer ingests directly from that public source. This is the same data available to any resident or researcher.
Data Sharing Agreement
In some cities, Civic Informer has a formal data sharing agreement with the police department, providing structured access to incident data directly.
How We Process Data
Once collected, each city's incident data is normalized into a consistent internal structure. This allows the platform to aggregate and report consistently within that city's dataset.
Incident locations are geocoded to assign each record to the correct neighborhood. We use block-level addresses only and do not publish exact addresses or precise coordinates beyond what the department releases.
Incident narratives are scanned for arrest details, which are extracted for use in statistical reporting. Names and personal identifiers are never published on the platform in any form.
How We Handle Records
Records are published as received from the source. We do not alter, editorialize, or modify source data. Where a data sharing agreement governs our access, we publish only what we are permitted to share under that agreement.
The platform is designed to prevent duplicate records, flag incomplete data, and preserve the integrity of the source record throughout. No incident data is manually modified once it enters the system.
Update Frequency
Civic Informer updates daily. The platform and newsletter reflect the most recent available data from each city's source, published each day as long as new data is accessible.
Source Verification
Every incident record published on Civic Informer retains its original case number from the source department. Any reader can cross-reference a specific incident against the department's own public records. We do not alter or modify source data.
Incident Classification System
Category System
Every incident ingested by Civic Informer is assigned to one of several categories based on its offense type. These categories are used throughout the platform in daily reports, weekly summaries, trend analysis, and the newsletter.
The current standard category set includes:
Administrative
Routine law enforcement activity including assist calls, welfare checks, directed patrol, death investigations, and other non-incident responses
Assault & Violence
Offenses involving physical violence or the threat of it, including assault, robbery, homicide, shots fired, bomb threats, and weapons violations
Behavioral Health
Calls involving mental health, welfare, missing persons, overdose, and suicide-related responses
Disorder
Disruptive behavior and quality-of-life incidents including disorderly conduct, trespass, noise complaints, prowler, and malicious mischief
Domestic & Family
Domestic disputes, child abuse and neglect, custodial interference, and related family incidents
Drugs & Alcohol
Drug and alcohol offenses including narcotics violations, DUI, drunk person, overdose, and liquor violations
Fraud & Financial
Financial crimes including fraud, forgery, identity theft, extortion, and prescription fraud
Harassment & Threats
Threats, stalking, harassment, violation of court orders, and hate crimes
Sex Offenses
Sexual offenses including rape, indecent exposure, prostitution, voyeurism, and internet exploitation
Theft & Burglary
Property crimes including theft, burglary, shoplifting, vehicle prowl, and possession of stolen property
Traffic & Vehicles
Traffic-related incidents including collisions, DUI, hit-and-run, vehicle theft, and parking issues
Vandalism & Destruction
Property destruction including graffiti, malicious mischief, arson, and related offenses
The category set is configured per city deployment and can evolve as offense patterns, department terminology, or reporting needs change.
How Classification Works
Each incident is classified at the time of ingestion based on its offense description. The same offense description will always produce the same category. Classification is consistent and reproducible across the entire dataset.
Violent vs. Non-Violent
Each incident is designated as violent or non-violent according to a defined set of criteria mapped at the individual offense level. The designation is applied consistently — the same offense type will always receive the same designation.
Offenses designated as violent include:
Arson
Assault (felony, misdemeanor, on an officer)
Bomb threat
Child abuse
Explosives
Fight
Homicide
Kidnap / unlawful imprisonment
Rape and rape of a child
Reckless endangerment
Road rage
Robbery
Sex crime (no rape)
Shots fired
Stalking
Threats
Vehicular assault
Weapon violations
All other offense types are classified as non-violent. The violent incident count and percentage are featured in daily reports and trend analysis as a consistent signal of severity, distinct from raw incident volume.
This designation is broader than the FBI's Uniform Crime Reporting (UCR) definition of violent crime, which covers murder, forcible rape, robbery, and aggravated assault. Civic Informer's violent designation reflects a wider set of offenses that involve physical harm, credible threat, or serious risk to persons.
Why These Categories
The category set is built around how municipal offense data clusters in practice, grouping related offense types into readable buckets without requiring readers to parse individual offense strings.
Where a department's offense terminology or a city's incident profile calls for a different structure, the category configuration can be adjusted at the deployment level.
How Statistics Are Calculated
Daily Statistics
Each day, Civic Informer aggregates all incident records into a summary that covers total incidents, violent and non-violent counts, arrests, category breakdowns, neighborhood distribution, and activity by time of day. This daily summary is the foundation for all reporting on the platform.
Trends and Rolling Averages
The platform calculates rolling averages over 7, 14, and 30-day windows to provide context around daily figures. Each day's numbers are compared against prior periods — the previous day, the prior week, and historical highs and lows — to surface whether activity is trending up, down, or holding steady. Category-level trends show which incident types are increasing or decreasing over time.
How Top Values Are Determined
When the platform identifies a top neighborhood, category, or time of day, that value is the one with the highest incident count for the period. These are computed directly from the data with no weighting or scoring applied.
Notable Observations
Each daily report includes a set of notable statistical observations — for example, that incident counts are at a 30-day high, or that a particular category has trended up for several consecutive days. These are generated algorithmically from the underlying data.
How Civic Informer Uses AI
Our Approach
Civic Informer uses AI to assist in drafting the written portions of each daily brief — the lead paragraph, narrative analysis, and local intelligence items. It works from data that has already been computed and verified before any content is written. We are transparent about where and how AI is involved so that city officials, partners, and residents can trust what the platform publishes.
What AI Does Not Do
All incident data comes directly from official department sources without AI involvement. Case numbers, dates, times, locations, offense types, and all statistics are computed directly from source records. AI does not calculate, interpret, or modify any data on the platform.
All charts, maps, and data visualizations are generated from actual data using standard visualization tools.
Human Review
Every edition is reviewed by a human operator before it is sent. The operator checks for accuracy, tone, and completeness. There is no auto-publish mode. Editorial decisions are made by people.
Data Integrity
Source data is never AI-generated or AI-modified. Statistical calculations are computed directly from incident records. What the department reports is what the platform presents.
Daily Reporting & Platform Formats
Daily Reporting, Not Real-Time
Civic Informer is a daily reporting platform. Data is collected and processed once per day, covering the previous day's incident records. A report published on Tuesday morning reflects Monday's data. Incidents are not reported in real time.
Two Formats, Same Data
Each day's data is published in two formats: a web report and an email newsletter. Both draw from the same underlying incidents, arrests, and statistics.
The Web Report
The comprehensive view. Full incident log, interactive charts, trend analysis, neighborhood breakdowns, and arrest details. Designed for anyone who wants to explore the data in depth.
The Daily Brief
The email newsletter — a concise summary. AI-written lead paragraph, key statistics, category and geographic highlights, and three locally-grounded intelligence items. Answers "what happened yesterday?" quickly.
The statistics in both formats are always consistent. Both are generated from the same daily pipeline run. The web report is available each morning after the pipeline completes. The newsletter follows after operator review and approval.
Known Limitations
Source lag
We report what the police department has logged. If a department is slow to enter records, there may be a gap between when an incident occurred and when it appears on the platform.
Classification
Our category system maps offense descriptions to categories using defined rules. Unusual or ambiguous descriptions may not map cleanly and are flagged for our review.
Geocoding accuracy
Addresses are geocoded to block-level precision. Some addresses may not geocode successfully and fall back to the city center. Neighborhood assignments near boundaries may occasionally be incorrect.
Arrest data completeness
Arrest information is extracted from incident narrative text. If a narrative does not follow the expected format, arrest details may not be captured. Arrest counts should be treated as a lower bound.
Single-source dependency
Each city relies on a single data source. If that source is unavailable or changes format, the pipeline may produce incomplete results. We monitor for these issues and address them as they arise.
How We Handle Gaps
When data quality issues are detected, they are logged and reflected in the report. We never interpolate, estimate, or fill in missing data. If a record was not received, it is not reported.
Transparency by default
We believe that public safety data platforms have an obligation to explain how they work. Every data point on Civic Informer is traceable to an official government source. Every statistic is computed by documented methods. Every AI-generated paragraph is produced under human supervision and editorial controls.
If you have questions about our methodology that are not answered here, we want to hear from you. Transparency is not a feature — it is the foundation of trust.
Read our full mission statement →