Last-Mile Delivery Optimization — Why It's the Most Expensive and Most Important Part of Your Supply Chain

What are the Key Takeaways from this Executive Summary?

• A failed delivery is two journeys. Both were paid for. Only one could have earned anything. This is usually the largest recoverable cost in the last mile.
• Customer expectations have already moved. Tracking, narrow windows and same-day options are now the baseline, not a selling point.
• The levers are well understood and do not need new infrastructure. Live re-routing, collection points, and reading your own failure history. What is missing in most operations is the measurement, not the tooling.

Why Does Last-Mile Delivery Consume Over Half of Total Shipping Costs?

The maths is plain. Line-haul moves a lot of freight between two points, in near-full trucks, on a route that rarely changes. The middle mile runs hub to hub on a settled pattern.

The last mile breaks one consolidated load into dozens or hundreds of separate drops. Each drop brings traffic, a parking problem, a locked lobby, and an unknown: is anyone home?

The Capgemini Research Institute puts last mile at up to 53% of total shipping costs. Parcel volumes have grown since the pandemic, but drops per route in suburban and rural areas have not kept pace. Vans drive farther and deliver less per mile.

The World Economic Forum expects urban last-mile traffic to rise 36% by 2030 without intervention, adding 6 million tonnes of CO₂ worldwide.

For whoever runs delivery operations, this is not a nuisance. It is where the margin goes.

What Makes Failed Deliveries So Costly — And So Persistent?

Start with the cost of one failure in your own operation. It is two journeys, plus the dispatcher’s time, plus the call. Most operations have never added those up.

The causes are well known. Bad address data. A window that is too wide or absent. No message to the recipient before the van arrives. No check that anyone will be there.

Many operations still treat all of that as friction that comes with the job. It is not. Each of those four causes can be measured, and each has a fix that does not need new vehicles.

The one number worth producing first: of last quarter’s deliveries, how many needed a second attempt, and what did that second journey cost you?

How Are Route Inefficiency and Driver Shortages Compounding the Problem?

A route planned at 6pm yesterday looks efficient on paper. It meets today’s road closure and falls apart.

Most of a driver’s shift is spent in transit rather than delivering, and the exact share is something your telematics already knows. Pull last month: wheels-turning time against time at the stop. That ratio is the gap the planning is fighting, and it is worth having your own figure before anyone sells you an improvement to it.

Driver shortage makes every inefficiency worse. Recruitment is hard across North America and Europe, and turnover is high in dense urban markets. If you cannot add people, the only lever left is using the hours you have better.

City rules narrow the window further. Restricted delivery zones, low-emission zones, timed access to pedestrian streets and scarce kerb space all eat into the shift. A van on a central London or Manhattan route loses a large part of its day to access and parking, not to driving. Measure that share for your own routes before you buy anything to fix it.

How Can Dynamic Route Optimization and PUDO Networks Reduce Last-Mile Costs?

Live re-routing is not faster planning. It is planning again during the day. A road closes, so the remaining stops are re-ordered. A customer says they will be out, so the slot moves.

Improvements are claimed for this, in percentages, by everyone selling routing software. Treat all of them as unmeasured until the measurement is yours: stops per route and fuel per drop, for one depot, for the month before and the month after. Those two numbers are in your own telematics and fuel-card data, and they settle the question for your network rather than for somebody’s average.

PUDO points attack a different cost. A parcel sent to a staffed collection point cannot fail because nobody was home — someone is always there to sign for it. Drops also cluster: one stop serves many parcels. For retailers with high return rates, the same points take returns back.

The two fit together. Live re-routing makes the doorstep drops worth more. PUDO takes the drops most likely to fail out of the van. Whether that trade works for you depends on your own failure rate and your own cost per attempt, which is why those two figures come before any tool choice.

Why Is Predictive Delivery Intelligence the Next Competitive Advantage?

The better operations are not only planning routes. They are reading their own history before dispatch.

An address that has failed repeatedly is not a surprise. It is a record you already hold. The parcel can be routed to a nearby collection point, or the customer called for a confirmed window, before the driver loads the van.

The same history helps with capacity. A volume projection by postcode area, a few days ahead, lets you move vans, adjust shifts and book extra drivers before the spike lands rather than during it.

Two capabilities sit underneath all of this, and they are worth separating. One is demand forecasting — a volume projection by area, with the history it came from attached, so a planner can argue with it. The other is route optimization — the measured distance and travel time for a leg under current conditions, cheap enough to ask again after lunch. Runink FACE does both. Neither prints a saving next to the answer. The routing provider does not return one, and an invented figure beside two measured ones is how an estimate gets quoted back as a fact.

Conclusion

The last mile is not going to get easier. Cities will get more congested, expectations will keep rising, and drivers will stay hard to hire.

But the levers are not exotic. Live re-routing, collection points and volume projection all run on data you already record. None of them needs new depots.

So the question for whoever runs delivery operations is which lever touches their actual cost. That turns on one figure most operations have never worked out: how many deliveries needed a second attempt last quarter, and what each of those second journeys cost. A failed first attempt is two journeys, and it is usually the largest recoverable item in the last mile. Talk to us if it would help to work that number out.

Sources

• Capgemini Research Institute — The Last-Mile Delivery Challenge — Research on last-mile delivery economics, consumer expectations, and the 53% cost share of total shipping expenditure
• McKinsey & Company — How Customer Demands Are Reshaping Last-Mile Delivery — Source for the driver transit-time share and the route planning density and fuel figures quoted above
• World Economic Forum — The Future of the Last-Mile Ecosystem — Source for the urban delivery traffic and CO₂ projections to 2030