Supply Chain Analytics for 3PLs: The Four KPIs That Actually Matter

We’ve walked into a lot of 3PL operations rooms. The wall of dashboards usually looks impressive: revenue per customer, shipments per day, OTIF by region, on-time delivery, total volume, fleet utilization, top customers by volume, top routes by margin, monthly trend lines for each. A dozen tiles, refreshing every fifteen minutes, watched by exactly nobody.

The problem isn’t the dashboards. It’s that the dashboards measure what the BI tool could measure easily, not what actually drives the business. The four metrics below are what we wire up first in any logistics analytics engagement. The rest is decoration.

1. On-Time In-Full (OTIF) — measured against the commit, not the plan

Most 3PL OTIF dashboards report against the internal plan. The truck was scheduled for 8 AM departure; it left at 8:15; that’s a minor variance. From the customer’s perspective, that’s not what they bought. They bought a delivery window — and OTIF should measure delivery against the customer commit, not your internal schedule.

The right OTIF measurement:

In Full: the shipment delivered the quantity ordered, no shortages, no substitutions.
On Time: delivered within the window the customer was committed to — not the window your dispatcher set.
Per shipment, not aggregated. Aggregate OTIF hides the customers who are silently churning.

The interesting cut isn’t the overall OTIF number; it’s OTIF by customer. A 96% overall OTIF can hide one customer running at 82% who is six weeks from leaving you.

2. Lead Time Variance — not lead time average

Reporting average lead time across a corridor is mostly useless. Customers care about predictability. A shipment that always takes 4 days is operationally easy. A shipment that takes 2 to 7 days unpredictably is a planning nightmare for the receiving end.

The metric that matters:

Lead Time p95 minus p50. How much wider than median is the 95th percentile? Tight distribution = predictable corridor. Wide distribution = something is broken upstream.
Per lane, per service level. Aggregating across the network masks lane-level issues.

When this metric tightens, OTIF tends to follow without you having to chase it directly. When it widens, you’re seeing the cause of an OTIF problem you haven’t received complaints about yet.

3. Cost-to-Serve, by customer

The customers spending the most aren’t always the most profitable. The mix of expedited orders, partial truckload demands, weekend dispatches, and special handling can quietly turn a top-line favorite into a margin sink.

Cost-to-Serve breaks down per customer per period:

Transportation cost (line haul + last mile).
Warehouse cost attributable to the customer’s volume.
Special handling, returns, claims.
Customer service cost (call volume × handling time × loaded labor).

Subtract this from revenue per customer and you get true contribution margin per customer. Most 3PLs we’ve worked with discover that 2-3 of their top-10 customers by revenue are in the bottom-10 by margin. That’s actionable. The aggregate margin number isn’t.

4. Warehouse Productivity — units per labor hour, not warehouse uptime

Most warehouse dashboards report uptime, dock occupancy, and total throughput. None of those tell you whether you’re getting more efficient over time.

The metric to track:

Units per labor hour (where “units” matches your operational reality — cases, pallets, picks, depending on the operation).
Trended week-over-week, not just snapshot.
Cut by shift and by activity type (receiving, putaway, picking, packing, shipping).

When you can see that picks-per-labor-hour is up 12% quarter-over-quarter on day shift but flat on night shift, you have a concrete operational lead — and probably a training, layout, or staffing issue to investigate. When you only have aggregate throughput, you’re reading tea leaves.

The metrics most dashboards over-index on

In order of “looks important but isn’t actionable”:

Total shipments per day. Volume isn’t a quality signal. A bad day with more shipments is still a bad day.
Revenue by customer. Without cost-to-serve, this is misleading. See KPI 3.
Fleet utilization percentage. High utilization is good — until you discover it correlates with overtime and driver burnout. Pair with cost-per-shipment if you must.
Customer-facing on-time delivery (against your plan). Always reports well. Always misses the customer experience. See KPI 1.

We’re not saying don’t track these. We’re saying don’t make decisions on them.

What you need underneath to measure this honestly

Most 3PLs we work with can’t measure the four KPIs above accurately because the data lives in five disconnected systems — the TMS, the WMS, the ERP, the telematics provider, the customer portal. OTIF lives in the TMS but the customer commit is on a PDF in a shared drive. Cost-to-Serve lives across the ERP and the WMS and nobody has stitched them together.

The pattern that works:

A canonical event log for every operational event (pickup, scan, dispatch, transit hop, delivery, settlement). Tied to a shipment ID, a customer ID, a lane.
A small data warehouse on top (Postgres + ClickHouse or BigQuery is plenty for most 3PLs) with dbt-modeled metrics.
Dashboards that query the warehouse, not the operational systems directly. Operational systems are tuned for transactions; analytical queries belong somewhere else.

We’ve built this exact pattern for several logistics operators. The standup cost is real — typically 8-12 weeks for the core data layer plus the four KPI dashboards — but every subsequent analytics question becomes a SQL query, not a six-week project.

Where to start

If you’re a 3PL operations or commercial leader and you only have budget to instrument one metric properly this quarter, pick OTIF by customer. It’s the metric most likely to surface a hidden problem in the next 90 days — usually a customer relationship that’s silently deteriorating. The other three are leading indicators of OTIF anyway; if OTIF by customer is solid, the rest will follow.

pdpspectra builds data and analytics platforms for logistics operators worldwide. If you want to scope what an honest analytics layer would look like for your operation — or just want a second opinion on what your existing dashboards are missing — tell us what you’re trying to measure. Or read more on our logistics solutions.