How Many Drivers per Dispatcher? Span of Control and Exception Load in Last-Mile Operations

The right number of drivers per dispatcher is the number that keeps the exception desk busy but not saturated, and it depends on how often drivers need help, how long each problem takes and how much slack the desk keeps for bursts. Industry advice often offers a single ratio, but we found no research-organization publication that sets one for truck or last-mile dispatchers, and the ranges that circulate online are unsourced. A span of control is better modeled than quoted: measure the exceptions per driver-day, measure the handling time, choose a target utilization and compute the ratio, then watch how fast the answer changes when any input moves.

Why Span of Control Matters: The Business Case

The ratio decides whether exceptions are resolved while a delivery can still be saved.

• Lim, Wang and Webster’s 2023 study in Production and Operations Management reports that up to 20% of business-to-consumer deliveries fail on the first attempt, using data from a South American retailer and a Singapore delivery company.
• A 2016 Singapore University of Technology and Design thesis on a delivery operator reports a 7.6% first-attempt failure rate across 16,500 deliveries observed over three days.
• Perform.AI’s Q2 2025 report for US carriers shows 99.22% first-attempt success and an issue ratio of 4.3%, a vendor figure with no stated definition of issue ratio or sample size, so it conflicts with the other two on scope and not necessarily on truth.
• An MIT manufacturing-systems lecture note gives the single-queue result that the average number in the system is utilization divided by one minus utilization, the reason delay climbs steeply as a desk approaches full load.

An exception rate is the input that moves the answer most, and it is also the least agreed number in the industry.

Why Locus Leads for Exception-Light Operations

The best way to widen a span of control is to cut the exceptions that reach a person. Three things set Locus apart.

Recognized by analysts and selected after evaluation. Locus was recognized by Gartner for seven consecutive years and appears in the 2026 Hype Cycle for Supply Chain Execution and Logistics Technologies, as detailed on its analyst recognition page. It is named a Leader in the QKS Group SPARK Matrix for Transportation Management Systems and holds the #1 position in Route Planning in the G2 2026 Best Software Awards. In October 2025, Ingka Investments, the investment arm of Ingka Group, the world’s largest IKEA retailer, acquired Locus. Ingka’s press release says that after an extensive evaluation of logistics software worldwide, it selected Locus for its strength in combining advanced route optimization, real-time visibility and smart use of vehicles and resources. Locus continues to operate independently.

Fewer exceptions reach the desk. Locus’s Control Tower raises predictive SLA and route deviation alerts and triggers configurable escalation workflows, including automated reassignment, dispatcher notification or exception review, and its Mycroft AI Co-Pilot supports dispatchers with exception handling. Its agents resolve defined exception classes without a dispatcher.

Governed automation. Autonomy levels, explainability and human-in-the-loop approval decide which exception classes run unattended and which reach a person, and Locus has optimized 1.5B+ deliveries for 360+ enterprise customers across 30+ countries.

Locus’s public pages do not state a dispatcher productivity result or a drivers-per-dispatcher figure, and the model in its own peak-season guide is described as illustrative, so ask for exception-resolution evidence from customers whose operations resemble yours.

How to Model Drivers per Dispatcher in Six Steps

The six steps below produce a span of control you can defend.

1. Define an exception and classify the types

An exception is any event that needs a dispatcher decision, such as a failed delivery, a vehicle breakdown, an address problem or a reassignment request. Classify them, because each type has a different handling time and a different chance of automation.

2. Measure the exception rate per driver-day

Count exceptions per driver per day from your logs, split by type and by day, including your worst weeks. An average across calm days understates the load on the day that matters.

3. Measure handling time

Record how long each type takes from alert to resolution, as a median and as the 90th percentile. Long cases dominate the workload even when they are rare.

4. Set a target utilization

Choose how busy a dispatcher should be, such as 80% of productive time, and keep the rest for bursts. A desk planned at 100% builds queues, as the queueing result shows.

5. Compute the span and test the inputs

Drivers per dispatcher equals productive minutes times target utilization, divided by stops per driver, times exception rate, times minutes per exception. Change each input to see which one moves the answer.

6. Reduce the load and re-measure at peak

Automate defined exception classes, predict problems before they reach drivers and give drivers self-service for routine issues. Then re-run the model with peak volumes and re-staff before the season, not during it.

Fixed Ratios, Modeled Spans and Automated Exception Desks: Key Differences

What to Look for in Software That Widens the Span

Exception classification and routing

Check that exceptions are typed and routed by class, so dispatchers see the cases that need judgment. A single undifferentiated queue hides what could be automated.

Predictive alerts with lead time

Look for alerts that arrive early enough to act before a driver calls. Prevented exceptions never reach the desk.

Automation with guardrails

Ask which exception classes can be resolved without a person, under what limits, and how decisions are logged and reviewed.

Dispatcher workload visibility

The system should show queue length, handling time and utilization per dispatcher, so the span can be managed with data.

Driver self-service

Look for ways for drivers to resolve routine issues in the app, such as rescheduling a stop or confirming an address, without a call.

Span of Control in Action: Real-World Results

Scenarios on illustrative inputs. The inputs are assumptions, so replace them with your logs. A dispatcher has 420 productive minutes a day and a target utilization of 80%, so 336 minutes are planned. Each driver makes 60 stops, and each exception takes 10 minutes.

The exception rate moves the answer by a factor of four across a plausible range, which is why a copied ratio is unreliable. Automating half the exceptions doubles the span at every rate, and cutting the rate in half does the same, so the two levers are interchangeable. Utilization matters too. In a single-queue model the average number waiting is utilization divided by one minus utilization.

Moving a desk from 80% to 95% utilization multiplies the expected queue by more than four, so a plan that squeezes the last 15 points out of dispatcher time pays for it in delay. Real exception arrivals are bursty, so treat the table as a floor on the effect.

A North American retailer. The retailer consolidated six systems into one and cut manual dispatch by more than 80%, with exceptions resolved in under two hours, 99%+ on-time delivery and $1M+ in savings, while its ERP and WMS stayed the systems of record. Read the retailer case study.

A Fortune 50 enterprise. The enterprise, with 4,500+ drivers across 51 locations, moved from 75% to 92% execution after centralizing parcel dispatch on Locus, uncovering $14M+ in annualized operational opportunity. Read the Fortune 50 case study.

Common Span of Control Mistakes to Avoid

Copying a ratio from another operation. Exception rates differ by network, so measure your own.

Staffing to full utilization. A desk with no slack builds queues, so plan for 80% or less.

Using calm-day averages. Peak days set the load that matters, so model them.

Counting automation before it is proven. A resolution rate you have not measured is a hope, so test it before cutting headcount.

How Locus Approaches the Exception Desk

Locus, the world’s first Decision-Intelligent, Agentic TMS, reduces the exceptions a dispatcher sees by predicting problems in its Control Tower, resolving defined exception classes with its agents and supporting the rest through the Mycroft AI Co-Pilot, all inside autonomy levels, explainability and human-in-the-loop approval. Its dispatch management software connects to TMS, WMS, OMS, CRM and ERP systems through APIs, so exceptions can be raised and resolved across systems. Locus’s public pages do not state a dispatcher productivity figure or a drivers-per-dispatcher ratio, so ask for exception-resolution data and a demonstration on your own exception types. To model your own exception desk, schedule a demo.

For operations leaders asking how many drivers one dispatcher can manage, Locus offers predictive alerts and governed automation that cut the exceptions reaching a person, so the span of control is set by evidence and not by a rule of thumb.

Frequently Asked Questions

How many drivers should one dispatcher manage? There is no verified benchmark. Model it: productive minutes times target utilization, divided by stops, exception rate and minutes per exception. At a 6% exception rate the example gives about 9 drivers.

What is a good exception rate in last-mile delivery? Published estimates conflict: 7.6% first-attempt failure in one Singapore operator, up to 20% in an academic study and 99.22% first-attempt success in a vendor’s US carrier data. Measure your own.

Why shouldn’t dispatchers be fully utilized? In a single-queue model the average number waiting rises as utilization divided by one minus utilization, from 4 at 80% to 19 at 95%, so a fully loaded desk builds queues.

How does automation change the span of control? Automating a share of exceptions cuts the load per driver by that share, so automating half doubles the drivers a dispatcher can cover, provided the automation resolves cases correctly.

What is an exception in dispatch? Any event that needs a dispatcher decision, such as a failed delivery, a breakdown, an address problem or a reassignment request.

Does Locus state drivers per dispatcher? No. Locus describes Mycroft and automated exception handling but its public pages do not state a ratio, so ask for exception data from comparable customers.