Fleet Management for Enterprise Retail and CPG: Why the Fleet Pays for Decisions it Does Not Make

Most fleet management content treats the vehicle as the unit of analysis. Buy the right telematics, cut idle time, tighten maintenance intervals, and cost per mile falls. That framing works for a fleet whose job is to move itself efficiently. It does not work for enterprise retail and consumer packaged goods, where almost every variable that determines fleet cost is set by someone outside the transport function.

How often a store is visited is a replenishment decision. When a vehicle is allowed to arrive is a store operations decision. Whether a product goes through a customer warehouse or onto a direct store delivery vehicle is a merchandising decision. What comes back on the truck is a packaging and asset decision. The fleet inherits all four as constraints, executes against them, and carries the cost in a transport budget nobody else is measured on.

That is the actual fleet management problem in retail and CPG. It is not primarily an efficiency problem inside the vehicle. It is a problem of measuring, pricing and negotiating decisions made elsewhere, and it needs a different metric set, a different constraint model and a different set of platform capabilities than a generic enterprise fleet deployment.

Why the cost lands in transport and stays there

The commercial context makes this expensive. FMI’s Food Retailing Industry Speaks 2025 puts food retail net profit margins at 1.7%, while food product suppliers reported net income of 7.7%. A retailer with 1.7 points of margin has almost no room to absorb a distribution cost increase, which is why service requirements tend to be pushed onto suppliers rather than paid for.

Suppliers, in turn, mostly fail to recover those costs. McKinsey surveyed 35 senior leaders at 28 North American consumer packaged goods companies and found that only 17% believe they recover more than 75% of the true cost to serve, with most estimating they recoup about half. Three quarters of the same group rated their understanding of service costs as decent or robust. The gap is not primarily ignorance. It is that the cost is understood in aggregate and cannot be attributed to the specific customer, store and service decision that caused it.

The distribution method itself is often chosen on grounds that have nothing to do with transport economics. PepsiCo’s FY2025 annual report states plainly that products delivered to customer warehouses use “a less costly method of distribution than DSD,” and that direct store delivery is retained because it “enables us to merchandise with maximum visibility and appeal” and suits products that are restocked often and respond to in-store promotion. That is one of the world’s largest DSD operators documenting that its most expensive distribution channel is retained for commercial reasons rather than transport ones, with the fleet carrying the premium.

Meanwhile the cost of the asset keeps climbing. ATRI’s operational cost analysis put the industry-average cost to operate a truck at $2.336 per mile in 2025, up 3.4%, with tolls rising 13.2% and repair and maintenance 8.6%. Outsourcing is a limited escape: the same analysis found truckload and refrigerated operating margins below 1.0% and carriers leaving 10% of trucks unseated, which means contracted capacity has little margin to give back.

Also Read: Logistics Fleet Management Software: Enterprise Guide

How fleet management works when the receiver sets the terms

1. Fix the denominator before optimizing anything

A retail or CPG fleet exists to move product, not to make visits. That makes cost per case, per pallet or per unit of volume the governing metric, and it ranks routes differently from cost per stop.

Take two illustrative routes in the same week. Route A serves 30 stores, delivers 900 cases and costs $600. Route B serves 12 stores, delivers 1,400 cases and costs $650. On cost per stop, A is $20.00 against B’s $54.17, so A looks 171% better. On cost per case, A is $0.667 against B’s $0.464, so B is 30% better. Same fleet, same week, opposite conclusion.

This is not a rounding difference. Teams managed on cost per stop will push volume toward high-drop-count routes and add vehicles to keep drop counts up, which raises cost per case while the reported metric improves.

2. Price the frequency grid, because it is the largest single lever

Visit frequency is usually set by replenishment planning against availability and store backroom constraints. Its transport cost is rarely calculated at the point of decision.

Consider a store selling 100 cases a week, with a marginal stop cost of $18 covering drive time and service time, daily demand volatility of five cases, a one-day lead time and a 95% service target. Frequency changes the numbers as follows.

The asymmetry is the point. Moving from three visits to five raises transport cost 67% and cuts safety stock only 15%, because transport scales linearly with frequency while safety stock scales with the square root of the review period. Peak backroom occupancy does fall meaningfully, by 32%, so the store gets a real benefit. The transport budget pays for it.

At $1,872 per store per year, that decision costs $936,000 annually across 500 stores. If demand drifts and 10% of a 1,000-store estate ends up one band above what its velocity justifies, roughly $187,000 a year of transport spend is buying nothing.

3. Separate what frequency can fix from what it cannot

More frequency is often proposed as the answer to poor on-shelf availability. The evidence says it addresses a minority of the problem. Research by Daniel Corsten and Thomas Gruen, synthesizing more than 50 studies and summarized by ECR Retail Loss, put the global out-of-stock rate at 8.3% and attributed causes across five work processes covering 91% of incidents: store stocking 38%, store forecasting 22%, planning 11%, store ordering 11% and supply 9%.

Supply, the part a delivery fleet controls, accounts for 9%. Shorter review periods do reduce exposure to store forecasting error, so frequency helps beyond that slice. But a fleet asked to raise availability by delivering more often is being asked to compensate for shelf replenishment execution, and it will spend a great deal of money doing so badly.

4. Model the receiving window as a shared resource, not a route constraint

Every planner treats a store’s receiving window as a constraint on their own route. The window is actually a shared queue. A single dock serves one vehicle at a time, and several suppliers and internal routes converge on it.

With a two-hour window and twenty-minute unloads, and arrivals distributed across the window, two vehicles collide about 31% of the time, three about 70%, four about 94% and five about 99.6%. At six arrivals the required dock time equals the entire window. Narrow the window to one hour and three arrivals collide 96% of the time.

Each planner’s route shows on-time arrival. The store sees queueing, the driver waits, and the wait is booked as unexplained service-time variance rather than as dock contention.

Also Read: Cross Fleet Utilization: Key Aspects and Importance in 2026

5. Model capacity in both directions

Retail and CPG vehicles carry returnable transport items back: crates, totes, roll cages, pallets, dollies and unsold or damaged stock. An outbound-only capacity model treats the vehicle as progressively emptying. It does not.

If every unit of outbound volume generates a return that occupies a fraction of the same space, occupancy at the halfway point is that fraction added to the remaining outbound load. Collapsible crates nesting four to one leave the vehicle 62.5% occupied at mid-route against a model that says 50%. Roll cages that do not nest, returning at 90%, leave it 95% occupied, so the model overstates free space by 45 percentage points. Rigid crates returning in full leave the vehicle at 100% occupancy for the entire route, and the model still reports it half empty.

That gap is where dynamic order injection fails. A platform offering a mid-route addition into apparently free space is offering capacity that physically is not there.

6. Decide the owned and contracted split on route economics, not fleet totals

Retail and CPG networks run mixed capacity, and the split is usually set at network level. Route economics vary far more than network averages suggest: dense urban replenishment, long rural DSD runs and multi-drop backhaul-heavy routes have different cost structures and different sensitivity to volume change. The useful question is which route archetypes should sit on owned assets because their volume is predictable and their service requirements are tight, and which should go to contracted capacity because their volume is seasonal.

Also Read: Retail Logistics Management: Strategy and Execution Guide

Where a retail and CPG fleet requirement differs from a generic one

The right-hand column is why generic fleet software underperforms here. It is not less capable. It is modeling a different problem.

Five criteria for evaluating a retail and CPG fleet platform

1. Can it hold receiver constraints as first-class objects? Store receiving windows, dock counts, vehicle class restrictions, appointment requirements, unloading equipment and site access rules need to live as modeled constraints, not as free-text notes a planner reads and applies manually.

2. Does it model the reverse flow on the same trip? Returnable asset collection must be a task with volume, time and sequencing implications on the same route, not a separate reverse logistics workflow. Ask specifically how remaining capacity is calculated at a mid-route point.

3. Can it report cost at store and frequency granularity? Recovering cost to serve requires attributing it to the store and the service level that caused it. A platform that reports cost per route and per vehicle cannot support that conversation.

4. Does it support the master route pattern and controlled deviation from it? Retail and CPG operations need route stability for driver familiarity, store expectations and labor planning, plus a defined mechanism for deviating when volume justifies it. Pure daily re-optimization is disruptive; a frozen template is expensive.

5. Does it decide across owned and contracted capacity per shipment? Mixed-capacity networks need allocation at shipment level against live cost and serviceability, not lane-level assignment against a rate card set months earlier.

Also Read: Top Direct Store Delivery Examples in 2025

What this looks like in enterprise deployments

A leading North American retailer running multi-hundred stores across ocean, rail and road replaced six legacy systems with a single orchestration layer. Store delivery reliability reached over 99% on time with route compliance above 95%, exceptions resolved in under two hours, and manual dispatch effort down more than 80%. The deployment broke even in year one against savings above $1 million. The relevant detail is route compliance: in a master-route operation, a plan the drivers actually follow is what makes cost per case predictable.

A global FMCG manufacturer operating across ten Asian countries with more than 1,000 distributors and 5,000 riders reached 3X ROI while saving over 12,000 trips a month and optimizing more than $4 billion of orders across 1.8 million retail outlets. Trips saved is the CPG version of the frequency argument resolved in the fleet’s favor, through consolidation rather than through cutting service.

Across Locus FMCG and CPG deployments the pattern is consistent: roughly 15% lower freight cost, 20% fewer vehicles required, 84% less planning time and 95% SLA attainment. The vehicle reduction matters most, because it is the only one that removes fixed cost rather than variable cost.

Four mistakes that keep retail and CPG fleet cost high

Managing the fleet on cost per stop. It is the wrong denominator for an operation that moves product, and it rewards adding vehicles.

Treating the frequency grid as fixed input. It is the largest controllable driver of fleet cost, it is set outside transport, and it drifts out of alignment with store velocity continuously. If nobody prices it, nobody revisits it.

Planning capacity outbound only. Returnable assets are not a rounding error. In a full-return operation with non-nesting equipment, they consume the entire theoretical slack a re-optimization engine thinks it has.

Buying telematics and calling it fleet management. Vehicle diagnostics, driver scoring and ELD compliance are necessary and are a different category. Neither one models a store receiving window or allocates a shipment across owned and contracted capacity.

Also Read: Reverse Logistics Explained: Process, Types and Benefits in 2026

Why Locus fits enterprise retail and CPG fleet operations

Locus is an agentic transportation management system built for operations where the constraint set is the hard part. The Fireworks routing engine plans against more than 250 real-world operating rules, which is the layer that matters here: store receiving windows, dock and appointment constraints, vehicle class and site access restrictions, shelf-life and temperature requirements, load sequencing tied to warehouse picking, and returnable asset collection modeled as tasks on the same trip rather than as a separate reverse workflow.

Across the retail and CPG deployments described above, that constraint fidelity translates into roughly 15% lower freight cost, 20% fewer vehicles, 84% less planning time and 95% SLA attainment, with route planning delivering up to 34% fewer miles, 25% higher drop density and 28% fewer trips through order consolidation.

Allocation runs across owned fleet, contracted transporters and a network of more than 1,000 carriers, decided per shipment against live cost and serviceability rather than per lane against a rate card. For mixed retail and CPG networks, that is what makes the owned-versus-contracted split a route-level decision instead of an annual one.

On cost recovery, Locus does not prescribe a cost-to-serve methodology, and whether a business charges by case, by drop or by frequency band remains a commercial decision. What the platform provides is the per-trip, per-store, per-frequency execution record that any such calculation needs as input, through analytics covering more than 250 operational metrics. That is the part that has to exist in the platform rather than in a quarterly spreadsheet reconciliation.

One boundary is worth stating: Locus is not a telematics platform. There is no ELD logging, no dashcams, no vehicle diagnostics and no telematics-based safety scoring. Enterprises running Locus typically keep a telematics provider alongside it. The platform decides what the fleet should do; telematics reports on the vehicle doing it.

Locus supports more than 360 enterprise customers across 30-plus countries, with over 1.5 billion deliveries optimized and more than $320 million in documented client logistics savings, and has been recognized by Gartner for seven consecutive years. Retail and FMCG customers include Nestlé, Unilever, Mars and Makro, the last of which scaled from 500 to 4,000 trucks with 24% higher fleet efficiency and 66% less planning time.