DDS Wireless · Technical note

Route Optimization Benchmark Results

DDS Wireless’s product team conducted internal tests to examine how different optimization objectives affect a proposed operating plan. This summary covers three separate comparisons involving delivery deadlines, delivery throughput, and vehicle use. The organizations involved are not identified.

Comparison approach

Each test compared a proposed engine-generated plan with results from an existing operation. The engine used a custom benchmark configuration directed toward the objective described below. These were planning comparisons, rather than measurements of improvements achieved after a live deployment.

Delivery deadline comparison

The first test examined a courier operation’s customer delivery deadlines. The reported dataset contained 642 deliveries.

Missed customer deadlines in an existing operation and a proposed engine plan, for a dataset of 642 deliveries.
Measure	Existing operation	Proposed engine plan
Missed customer deadlines	87	0

The proposed plan met the deadlines evaluated in this test. This is a result from a planned schedule under the test assumptions, rather than a claim that the software eliminated late deliveries during live operations.

The example illustrates why service commitments can be a meaningful optimization objective. Assessing a plan requires considering when deliveries are scheduled to arrive, alongside the resources and travel required to serve them.

Delivery throughput comparison

The second test examined a commercial-linen operator’s busiest day. The workload contained 564 deliveries, with five vans available. The benchmark configuration prioritized the number of deliveries covered.

Fleet size and deliveries covered in an existing operation and a proposed engine plan, from a workload of 564 deliveries.
Measure	Existing operation	Proposed engine plan
Vans used	5	5
Deliveries covered	198	268

The proposed plan covered 70 additional deliveries. Relative to the existing operation’s 198 deliveries, that represents an increase of approximately 35%.

The proposed plan covered 268 of the 564 deliveries. The result therefore demonstrates additional planned throughput within this test, rather than coverage of the entire workload.

Parcel-delivery vehicle comparison

The third test used four full days of historical parcel-delivery routes covering 17,735 stops. It examined whether the engine could plan the same stops with fewer vehicle assignments. The reference routes and proposed plans were scored on the same OpenStreetMap road network using OSRM.

Aggregate results across four delivery days comparing reference routes and proposed engine plans.
Measure	Reference routes	Proposed engine plans
Stops covered	17,735	17,735
Vehicle assignments across four days	119	100
Total road distance relative to reference	1.00×	1.08×

Across the four days, the proposed plans used 19 fewer vehicle assignments, approximately 16% fewer, with approximately 8% more total road distance. Vehicle assignments are summed across the four days. The totals do not represent a single day’s fleet or a count of distinct vehicles.

Individual delivery-day results with anonymized day labels. Road distance is expressed as a multiple of the corresponding reference routes.
Delivery day	Stops	Reference vehicles	Proposed vehicles	Distance multiple
Day 1	1,503	10	9	1.07×
Day 2	2,132	14	11	0.99×
Day 3	5,506	39	31	1.01×
Day 4	8,594	56	49	1.14×

A distance multiple of 1.00 means the same road distance as the reference routes. The daily figures range from approximately 1% less distance to 14% more. The aggregate 1.08 multiple is the total reported in the internal benchmark report, rather than an unweighted average of the daily multiples.

The result illustrates a tradeoff between vehicle use and travel. Whether fewer vehicle assignments would outweigh the additional distance depends on the operation’s vehicle, driver, and travel costs.

Testing closed on 14 September 2026. The September report states that the benchmarked setup had not yet been incorporated into the shipping product at that time. These are proposed plans from an internal test, rather than savings achieved in a live deployment.

Interpreting the results

The tests evaluated different objectives using different operational datasets. The courier comparison examined deadline performance. The commercial-linen comparison examined delivery throughput with five vans. The parcel-delivery comparison examined vehicle use and the accompanying change in road distance.

The results are specific to the data, objectives, assumptions, and custom configurations used. They are not guaranteed outcomes from default product settings and should not be interpreted as a forecast for another operation. None of these comparisons establishes a financial saving without the relevant operating-cost information.