Last-Mile Delivery KPIs: The Metrics That Actually Matter (2026)

You cannot improve a delivery operation you are not measuring. Most small fleets run on a feeling — "today felt busy," "that route ran late" — and never see the numbers that would tell them where the money and time actually go. That is expensive, because the last mile is the most costly stretch of the whole journey: industry estimates routinely put it at roughly half of total shipping cost. This guide covers the last-mile delivery KPIs that are genuinely worth tracking, how to calculate each one, what a healthy figure looks like, and how to collect them without hiring a data team.

Why track delivery KPIs at all?

Because the alternative is guessing. A KPI — a key performance indicator — is just a number you watch over time so you can tell whether a change helped or hurt. The point is not the number on any single day; it is the trend. When cost per drop creeps up three weeks running, or first-attempt success quietly slips, the trend tells you before the customer complaints do. Good delivery metrics turn "I think we got faster" into "stops per hour rose from 4.1 to 4.8 after we started optimizing routes," which is the difference between a hunch and a decision.

The core last-mile delivery KPIs

These are the metrics that earn their place for a business running its own drivers. For each one: what it measures, how to work it out, and which way is good.

1. On-time delivery rate

The share of deliveries completed inside the promised window. Calculate it as on-time deliveries ÷ total deliveries over a period. It is the most visible signal of service reliability, because it is the one your customer feels directly. The fastest way to lift it is to plan routes that actually respect the windows you promised — see how to use delivery time windows without fragmenting your routes.

2. First-attempt delivery success rate

The share of orders delivered on the first try, with no redelivery. Calculate it as successful first attempts ÷ total delivery attempts. This is one of the most important cost metrics in the last mile: industry studies regularly find a meaningful slice of e-commerce deliveries fail the first time, and every failure is paid for twice — the wasted attempt and the redelivery. Most misses come from the customer not being home or not knowing the driver was coming, both fixable; our guide to reducing failed deliveries walks through the levers.

3. Cost per drop

Total route cost (labor, fuel, vehicle) divided by completed deliveries. This is the headline efficiency number — the one that tells you whether in-house delivery is actually beating the alternatives. It only means something if your inputs are honest, so we break down exactly how to calculate and lower it in delivery cost per drop.

4. Stops per hour (and per route)

Completed stops divided by hours on the road — a clean proxy for sequencing quality and driver flow. Low stops-per-hour usually points at one of two things: routes with too much backtracking, or service times you have underestimated. Tightening the sequence is the first fix; our guide to optimizing multi-stop routes covers how.

5. Average delivery / service time per stop

How long a stop actually takes from arrival to "delivered." A signature handoff is two minutes; a bulky install is twenty. If your planned day keeps running long, the problem is usually wrong service-time assumptions, not bad sequencing — and this is the number that exposes it.

6. Failed / undelivered rate

The inverse of first-attempt success, tracked on its own because the reasons matter: customer not home, wrong address, refused, access problem. Tagging the reason at the door turns a vague "10% failed" into a fixable list — bad geocoding here, missing notifications there.

7. Plan-versus-actual time

The gap between the route the optimizer built and how the day really ran. A consistent overrun means your constraints (service times, traffic, breaks) are wrong, so the plan is optimistic before the driver even leaves. Close this gap and every other estimate gets more trustworthy.

8. "Where is my order?" contact volume

How many customers contact you to ask where their order is — the WISMO rate. It is a customer-experience KPI that doubles as a support-cost KPI. The cure is proactive communication: an on-the-way message and a live tracking link. See live delivery tracking with an ETA and automated customer delivery notifications for how this falls once customers can see the driver coming.

How many KPIs should you actually track?

Fewer than you think. A new operation that tries to watch fifteen metrics watches none of them. Start with three or four that map to your biggest problem, get them honest and consistent, then add more as the operation matures.

1. If costs are the worry: cost per drop, stops per hour, first-attempt success.
2. If customers are complaining: on-time rate, WISMO contact volume, failed-delivery rate.
3. If you are scaling drivers: stops per hour per driver, plan-versus-actual, first-attempt success by driver — see delivery driver management for coaching on these.

How to collect these numbers without a data team

The reason most fleets fly blind is not that the numbers are hard to understand — it is that the inputs are scattered across paper run sheets, texts, and someone’s memory. The fix is to run deliveries through one system that records the inputs as a by-product of normal work:

• Completed-stop counts — your delivery software should already count completed drops per driver and per route. In Routella those counts are the denominator for cost per drop and stops per hour, captured automatically as drivers mark stops delivered.
• First-attempt and failure data — captured the moment a driver records proof of delivery (a photo or signature) or marks an attempt failed with a reason.
• On-time data — falls out of routing that respects time windows, so the system knows the promise and the actual arrival.
• WISMO signal — when customers have a live tracking link, the questions they would have asked simply stop arriving, which you can see in your support volume.

You do not need a finance team or a BI tool to start — you need consistent inputs and the discipline to look weekly. Routella records the operational inputs as part of the normal dispatch-to-delivery flow, and there is a free plan (1 driver, 50 orders per month, no credit card) so you can start measuring a real operation before paying anything.

Common mistakes when measuring delivery performance

• Tracking vanity metrics. Total deliveries looks good on a slide but tells you nothing about efficiency. Per-stop and per-route numbers do.
• Optimizing distance instead of time. Customers experience time and drivers are paid by time; the shortest-distance route is often not the cheapest one.
• Comparing across different days. A rainy Friday is not a quiet Tuesday. Compare like with like, and trust the trend over weeks rather than any single day.
• Measuring without acting. If a KPI never changes a decision, stop tracking it and free up the attention for one that does.
• Dirty inputs. Half-recorded stops and untagged failures produce confident, wrong numbers — worse than no number at all.

The bottom line

Last-mile delivery KPIs are not an accounting chore — they are the dashboard lights that tell you when your operation is healthy and where to push when it is not. Pick three or four that match your biggest problem, get the inputs honest, and watch the trend, not the day. If you want the operational inputs collected for you as part of running deliveries, see how Routella works or start on the free plan and measure your next week of routes.