Computer-aided dispatch is the software that moves a 911 call from the answering console to a patrol unit, and Americans place roughly 240 million 911 calls a year, per Federal Communications Commission estimates. The system classifies each call, recommends the closest available unit and time-stamps every step. The published response time, however, starts only when the call is received and often stops when a unit arrives, leaving most of the human and technical work out of the measurement.
The gap matters because jurisdictions borrow each other's numbers as if they were comparable. They frequently are not. What one city counts as "response time," another may count differently, and neither figure captures the interval a caller actually experienced.
What happens between a 911 call and an officer arriving?
A call passes through several recorded stages before anyone arrives. A call-taker answers and enters the incident into the CAD system. A dispatcher reviews and prioritizes it, selects units and transmits the assignment. Officers acknowledge, drive and arrive. Each transition generates a timestamp, and those timestamps are what every published statistic is built from.
The clock almost never starts when the incident occurs or even when the caller first dials. Queue time before the call is answered, time spent asking questions, and the transfer between agencies when a call crosses jurisdictional lines are often recorded in other systems or not recorded at all. A caller's experience of waiting can be substantially longer than the response time a department later reports.
How does the software decide which units go?
CAD systems do not dispatch on their own; dispatchers make the decisions, using the software's recommendations. The system maintains a live picture of unit status — available, en route, on scene, transporting — drawn from officer inputs and, in some agencies, automatic vehicle location data. When an incident is entered, the system flags the recommended priority based on the call type and suggests the nearest suitable unit.
Priorities encode agency policy: a call involving a weapon outranks a noise complaint, and the software enforces that ranking. But the underlying call types are chosen by humans, and two agencies can classify the same incident differently. Analysts who compare dispatch data across departments routinely find that apparent performance differences track classification practices as much as staffing or geography.
What do response time numbers actually measure?
Most published figures measure the interval between the timestamp the department selects as the start and the timestamp of arrival. Which start point is chosen changes the number considerably, as the table below shows. Each is a legitimate measure; they answer different questions and are not interchangeable.
| Measure | Starts at | Ends at | What it captures | What it omits |
|---|---|---|---|---|
| Call to dispatch | Call received in CAD | Unit assigned | Call-taking and triage speed | Queue time, transfer time |
| Dispatch to arrival | Unit assigned | Unit on scene | Travel time | All pre-dispatch delay |
| Call to arrival | Call received in CAD | Unit on scene | The department's full recorded process | Time before the call was answered |
| Incident to closure | First CAD entry | Incident closed | Full handling, including on-scene work | Varies with closure practice |
Agencies also differ on whether they average across all calls or report the median, and on whether they include the lowest-priority calls that can wait hours. A department that reports median dispatch-to-arrival for priority-one calls will publish a far smaller number than a neighbor reporting mean call-to-arrival for everything.
Related stories: What police records management systems store, and why data quality varies · How gunshot detection systems perform, according to independent studies.
Why do the same minutes look different across cities?
Geography, staffing and call volume shape the raw numbers, but measurement choices shape them too. The Bureau of Justice Statistics' Law Enforcement Management and Administrative Statistics program documents wide variation in how agencies configure dispatch and record performance, and the BJS-led transition to the National Incident-Based Reporting System has highlighted how much local practice varies in what gets recorded at all.
Call classification is the quiet variable. If one agency codes welfare checks as urgent and another files them as routine, identical work produces different statistics. Researchers who audit dispatch data for city councils and oversight bodies have repeatedly advised that cross-jurisdiction comparisons be read as approximations, not rankings.
What happens when a call crosses jurisdictional lines?
Transfers are where time quietly disappears from the record. A cellular 911 call can be answered by a state highway patrol center, transferred to a county sheriff and then relayed to a municipal department, with each agency's CAD recording only its own segment. Reviews of call transfer practice by the emergency communications field have described these handoffs as a persistent weak point, since details can be lost and no single system holds the complete timeline.
The record that survives is a patchwork. When investigators, journalists or oversight boards later reconstruct what happened, they must stitch separate export files together, and small differences in system clocks can shift the apparent sequence by seconds or minutes. This is not a defect of any one product; it is a structural consequence of hundreds of independently run dispatch centers serving one telephone network.
What are the documented limits of the data?
Independent audits of dispatch operations, including reviews conducted by municipal inspectors general and city auditors, have found recurring gaps: timestamps edited or backfilled without notation, clock drift between connected systems, and calls closed in bulk during peak periods. None of this implies misconduct, but it means the data quality depends on disciplined entry under pressure.
The deeper limit is conceptual. Response time is a measure of one system output, not of safety. Decades of police-response research summarized by the National Institute of Justice have found that for most calls, the outcome was determined by circumstances before officers were even dispatched. Fast arrival matters in a shrinking subset of incidents; the data alone does not show which ones.
What should a reader check before trusting a response time claim?
First, find the definition: which timestamps, which call priorities, mean or median. Second, ask whether the figure was audited or merely reported. Third, look at what changed alongside any improvement — staffing, consolidation of dispatch centers, or a change in how calls are coded. An apparent speedup that coincides with a reclassification is a measurement artifact, not a service change.
Computer-aided dispatch software has made the recorded record of emergency response far richer than paper logs ever were. The records are only as honest as the definitions behind them, and the most important minutes — the ones before the call is answered — are usually the ones the system never sees.
For more context, read What police records management systems store, and why data quality varies.
For more context, read case management software.
For more context, read probabilistic genotyping.