Which Last-Mile Delivery Technologies Are Actually Ready to Deploy
Autonomous vans, sidewalk robots and delivery drones absorb most of the attention paid to last-mile innovation. They are also the only three technology categories in the market that are not yet commercially ready.
Net Zero Insights assesses ten last-mile technology categories across two innovation pathways. Seven are at full commercial readiness and can be deployed today. Every one of those seven is either software or a bicycle, which is a useful thing to know for anyone deciding where next year’s capital goes. The first article in this series covered why the last mile became the most expensive stage of shipping. This one works through what each technology actually delivers.
Which last-mile delivery technologies are commercially ready today?
The reason the operations pathway clusters at commercial readiness is structural rather than coincidental. These tools layer onto a fleet that already exists. They do not require an operator to buy vehicles, build charging infrastructure or redesign a depot network, which means they clear the two barriers that gate almost everything on the vehicle side: capital intensity and infrastructure dependence.
How much can route optimisation reduce delivery costs?
A peer-reviewed 2026 study in Scientific Reports compared static routing, where routes are planned before dispatch and cannot adapt, against dynamic routing, where routes update during operations using traffic-aware, real-time dispatch logic. The fleet was unchanged. Only the software differed.
| Measure | Static routing | Dynamic routing | Change |
|---|---|---|---|
| Operational cost | Baseline | 24.3% lower | |
| On-time delivery rate | 68.1% | 92.8% | +24.7 points |
| Congestion exposure | Baseline | 54.4% lower | |
| Maximum delay under traffic disruption | 34.7 min | 8.2 min | 4.2× reduction |
A 24% operating cost reduction with no change to the vehicle fleet is the largest efficiency result available anywhere in the last mile without capital expenditure. The on-time figure matters commercially for a second reason. A delivery that fails has to be attempted again, so reliability and cost are the same variable measured twice.
The caveat is data quality. Route optimisation depends on accurate addresses, reliable location signals and clean integration with order, warehouse and fleet systems. Where address data is poor or systems do not talk to each other, the modelled gains do not materialise. Route optimisation companies on the Net0 Platform.
How much does last-mile delivery cost per parcel by vehicle type?
On dense inner-city routes the binding constraint is stops per hour rather than payload, and that inverts the usual assumption that a bigger vehicle is a more efficient one. A data-driven evaluation of cargo bike performance in central Brussels measured e-cargo bikes against both electric and diesel vans on comparable routes.
| Delivery mode | Parcels per hour | Annual parcels | Vehicle cost per parcel | Lifecycle GHG impact |
|---|---|---|---|---|
| E-cargo bike | 10.1 | 20,200 | €0.10 | 96 to 98% lower vs vans |
| Electric van | 4.9 | 9,800 | €1.05 | Baseline |
| Diesel van | 4.9 | 9,800 | €1.10 | Baseline |
Twice the throughput at roughly a tenth of the vehicle cost per parcel is a large enough gap to change network design, but only within the conditions the study describes. Cargo bikes carry smaller loads over shorter routes, so they generally require nearby depots or microhubs where riders restock during the day. Research using simulated urban freight flows found that cargo bikes reduce van use and CO₂ emissions but can increase total operating time or feeder trips where the delivery network is not designed around them. The vehicle is not the intervention. The network is.
Cargo bike and e-cargo bike companies on the Net0 Platform.
Do electric vans solve the last-mile problem?
Electric light commercial vehicles are the most mature route to decarbonising a conventional fleet, and the most operationally familiar. They preserve the existing delivery model, suit longer routes, larger payloads, refrigerated loads and central-depot operations, and require no redesign of how deliveries are structured.
But they decarbonise the vehicle more than they improve the delivery model. An electric van occupies the same road and kerb space as a diesel one, so it does not address congestion, parking pressure or constraints on urban space. Charging adds a planning dependency, because vehicles must be assigned to routes that account for range, charger locations and charging time, a constraint that bites hardest on heavy loads and tight schedules. And where a city has designated a zero-emission zone, an electric van is what grants access. It is a compliance instrument as much as an efficiency one, which is why fleet electrification timelines are converging on 2030.
Electric commercial vehicle companies on the Net0 Platform.
Does consolidation matter more than vehicle choice?
This is the finding that reframes the whole comparison. Analysis of unit delivery cost across delivery modes shows that how much a vehicle carries per trip, and how much human supervision it needs, dominates which vehicle it is.
| Mode | Operating case | Unit delivery cost | Energy cost | Emissions |
|---|---|---|---|---|
| Drone | 1 package, 1 drone per observer | ~$13.50 | $0.06 | 1.07 kg CO₂e |
| Drone | 1 package, 20 drones per observer | ~$1.80 | $0.06 | 1.07 kg CO₂e |
| Electric car | 1 package, 1 driver | ~$9.40 | $0.45 | 1.40 kg CO₂e |
| Electric car | 5 packages, 1 driver | ~$3.00 | $0.12 | 0.36 kg CO₂e |
| Electric van | 1 package, 1 driver | ~$11.00 | $0.65 | 2.08 kg CO₂e |
| Electric van | 100 packages, 1 driver | ~$1.80 | $0.07 | 0.20 kg CO₂e |
| ICE van | 1 package, 1 driver | ~$11.60 | $2.50 | 6.40 kg CO₂e |
| ICE van | 100 packages, 1 driver | ~$1.90 | $0.27 | 0.64 kg CO₂e |
Two things follow. First, a fully loaded electric van at $1.80 per package matches the best-case drone and beats a lightly loaded electric car by a wide margin. The boring option, well utilised, wins.
Second, the drone number falls by a factor of seven with no change to the aircraft. What changes is how many drones one observer is permitted to supervise. At one-to-one supervision, labour can represent up to 95% of drone delivery cost. Drone economics are therefore gated by a regulatory parameter rather than an engineering one. That is a different kind of risk for an investor or an operator to underwrite, and it should be assessed differently.
Why are autonomous delivery, robots and drones still in validation?
The prize is real. Semi-autonomous and fully autonomous delivery vehicles could reduce urban delivery costs by approximately 10 to 40%, primarily by reducing the labour required to drive and park. But each of the three categories is held back by a distinct constraint, and none of them is principally about whether the technology works.
On-road autonomous vehicles must navigate unpredictable outdoor environments, including other vehicles, pedestrians, cyclists and roadworks. Current systems typically operate only within defined geographic areas, road types and speed limits. Wider commercial deployment depends on local policy, and restrictive or uncertain rules leave operators facing high uncertainty before they scale.
Delivery robots work best where routes are predictable: campuses, business parks and controlled urban environments. For short trips they improve delivery reliability and automate tasks that would otherwise need a courier. But they are slower and more limited than vans, bikes or couriers, so they are a targeted solution rather than a fleet replacement. Their deployment also raises questions about pressure on pedestrian space and accessibility, which makes adoption partly a matter of public acceptance and local rules.
Drones can fly over traffic, which suits urgent, lightweight or hard-to-reach deliveries. They cannot carry heavy packages, their batteries limit trip length, they must follow strict rules about where they may fly, bad weather grounds them, and noise and privacy concerns persist in many neighbourhoods. As the cost table above shows, their commercial viability turns on supervision ratios.
These three are also where the largest cheques are being written, which is worth holding alongside the readiness picture. See who is building and backing them.
What should an operator do with this?
The evidence supports a fairly clear sequencing, and it is close to the reverse of the attention each category receives.
- Start with utilisation and routing. Dynamic routing delivered a 24.3% operating cost reduction and a 25-point on-time improvement on an unchanged fleet, and the consolidation data shows packages per trip dominates mode choice. Both are addressable with software against vehicles you already own.
- Match the mode to the route, not to the fleet. E-cargo bikes doubled throughput at a tenth of the vehicle cost per parcel on dense inner-city routes. Electric vans remain the right answer for longer distances, larger payloads and refrigerated loads. This is a segmentation decision, not a replacement decision.
- Treat zero-emission zones as a deadline, not a preference. Where a city has designated one, fleet composition determines market access on a fixed date.
- Design the network before buying the vehicle. Cargo bikes need microhubs. Electric vans need charging assigned into route plans. Both can underperform, or add feeder trips, if deployed into a network built around diesel vans.
- Underwrite autonomy as regulatory exposure. The 10 to 40% cost prize is real, but the gating variables are permits, supervision ratios and public acceptance rather than engineering milestones.
Frequently asked questions
Next in this series: the last-mile delivery startup and funding landscape.
The post Which Last-Mile Delivery Technologies Are Actually Ready to Deploy appeared first on Net Zero Insights.