The Hidden Cost of Fresh Produce SLAs: A Fall Harvest Capacity-Sourcing Playbook for North American Grocers

The question harvest actually asks

Most harvest logistics analysis lands on speed: how quickly can the operation re-plan when yield arrives differently than forecast. That is a real constraint and it is not the expensive one.

The expensive question is sourcing. When Tuesday delivers six truckloads against a plan built for four, the operation does not primarily have a routing problem. It has two extra loads and a decision about where the capacity comes from. Captive fleet is already committed. The contracted carrier needs notice. Spot capacity is available at a price that moves hourly and moves against you specifically when everyone in the growing region needs it at once.

That decision gets made dozens of times during peak weeks, usually by a dispatcher on the phone, usually without a cost model, and almost always without pricing the one variable that makes produce different from every other freight category: the product is deteriorating while the decision is being made.

The scale of what deterioration costs is documented. USDA Economic Research Service supermarket loss estimates put average shrink at 12.6% across 24 fresh fruits and 11.6% across 31 fresh vegetables, with enormous variance by commodity, from 4.1% for bananas to 43.1% for papayas. Upstream of retail the picture is worse: USDA ERS cites FAO estimates that roughly 30% of fruit and vegetable losses occur at farms and in pre-retail distribution channels, which is precisely the stretch a grocer’s inbound dispatch controls.

Those numbers are not freight costs. They are the reason freight decisions in produce cannot be made on freight economics alone.

Also Read: Agentic TMS for North America’s Cold Chain Logistics: What Food and Grocery Shippers Should Know

Why shelf life inverts normal freight logic

Standard freight practice is a waterfall. Fill owned capacity first because it is already paid for. Then contracted capacity at negotiated rates. Escalate to spot only when the cheaper pools are exhausted. This is correct for durable goods and it is a good default for most of the year.

Perishables break it, because waiting is not free. Every hour a pallet sits waiting for cheaper capacity consumes remaining shelf life, and shelf life converts directly into either markdown at retail or shrink. The waterfall assumes the cost of delay is approximately zero, which is true for canned goods and false for berries.

Restated as a decision rule: the correct capacity choice minimizes freight cost plus expected decay cost, not freight cost alone. Once decay is priced, paying a spot premium immediately is frequently the cheaper decision, and the operation that proudly avoided a premium by waiting four hours has often bought a larger loss at the shelf.

Two consequences follow, and both are counterintuitive enough that they rarely survive contact with a procurement scorecard measured on freight cost per hundredweight.

Premium capacity is correct more often at harvest than off-peak. Not because rates are better, they are worse, but because the decay denominator is larger. Peak volume means more product waiting, and more product waiting means the cost of delay scales while the cost of a premium stays per-load.

The right answer varies by commodity within the same truck. A load of hardy root vegetables and a load of leafy greens have materially different decay curves, so they justify different capacity decisions on the same day out of the same packhouse. USDA’s shrink range, from 4.1% to 43.1% depending on commodity, is the quantitative basis for treating them differently.

Three capacity pools, three different instruments

The most common structural error is treating capacity as one number with a cost attached. It is three instruments with different lead times, ceilings, and cost behavior, and the differences determine which one can actually respond to a surge.

The bottom row is the allocation policy in one line, and it is worth stating why. Captive should carry base load because its cost is sunk and its ceiling is hard, so using it for volatility wastes the one pool that cannot flex. Contracted capacity should carry a forecastable surge, because that is what a committed band is for. Spot should be reserved for genuine volatility and for loads where remaining shelf life makes waiting more expensive than paying.

Two constraints complicate this in practice. Reefer qualification means spot capacity is not freely substitutable, since a truck without verified temperature capability is not capacity for this freight at any price. And driver hours cap captive capacity in a way that is easy to miss during planning: a captive fleet at 90% hours utilization has almost no surge capability regardless of how many trailers are in the yard.

Detention makes the ceiling tighter still. ATRI research found drivers were detained at 39.3% of all stops in 2023, losing between 117 and 209 hours per year. At harvest, when packhouse and DC queues lengthen, detention consumes capacity you have already committed and paid for, which means effective capacity falls at exactly the moment nominal demand rises.

Also Read: The Three-Workforce Fleet Reality: How Owned, 3PL, and Gig Drivers Actually Operate at Most Enterprises

The commitment timing problem

Contracted capacity has to be committed before yield is known, which is the structural bind of harvest sourcing. Commit high and you pay for volume you did not move. Commit low and you enter peak week buying spot from a position of visible desperation.

Most operations resolve this by committing to last year’s peak and absorbing the variance in spot. That is a defensible default and it systematically overpays, because last year’s peak is a single number applied to a distribution.

A better approach commits in tiers against yield confidence rather than a single volume. Establish a base tier at the volume you are confident about across the season, a second tier against the range you expect in a normal peak, and leave the tail deliberately uncommitted for spot. The tiers are negotiated together, before the season, which is when you have leverage.

This is also where the USDA data earns its place in an operating process rather than a market report. The USDA Agricultural Marketing Service Specialty Crops National Truck Rate Report publishes weekly rates from major shipping areas to ten destination markets, alongside a truck availability rating for each origin scored 1 for Surplus, 2 for Slight Surplus, 3 for Adequate, 4 for Slight Shortage, and 5 for Shortage. The Agricultural Refrigerated Truck Quarterly adds regional volume, rate, and availability trends for refrigerated truckload movements.

An origin moving from Adequate to Slight Shortage is a leading indicator that spot rates on those lanes are about to move, and it is published free, weekly, by a government agency. Very few grocers wire it into a dispatch decision, which means the signal exists and the response does not. The practical use is a trigger: when an origin you draw from crosses into shortage territory, pull committed capacity forward and stop treating spot as the overflow valve, because the overflow valve is about to get expensive.

Also Read: The Real Cost of Manual Dispatch in North American 3PLs

Allocating by shelf life and certainty

Combining the two variables gives a workable allocation matrix. Volume certainty determines which pool can be committed; remaining shelf life determines how much delay the load can tolerate before decay dominates.

The bottom-right quadrant is where most money is lost, because it is where the waterfall instinct is most wrong and most tempting. Low certainty plus short shelf life is exactly the situation where a dispatcher waits for a cheaper option, and exactly the situation where waiting is the most expensive available choice.

Operationalizing the matrix requires three data items to reach the dispatch layer: harvest date and cooling history so remaining life is known rather than assumed, a per-commodity decay expectation so greens and roots are treated differently, and live capacity state across all three pools in one view. The first is usually the missing one, because it lives in a packhouse or grading system nobody has integrated.

Also Read: Beyond In-House Fleet: When Should Enterprise Shippers Move to Multi-Carrier Orchestration?

How Locus handles capacity sourcing at harvest

Locus, the world’s first Decision-Intelligent, Agentic TMS, provides dispatch management software built to evaluate mixed capacity in a single allocation decision rather than in a sequence of escalations. Within DiSCO, the Capacity agent forecasts demand and matches available capacity across owned fleet, contracted carriers, and on-demand pools while holding remaining driver hours as live state, so surge capability is assessed on hours rather than on trailer count. The Carrier agent holds contract terms and rate structures as the live source of truth, which is what allows a committed band and a spot premium to be compared inside one decision. The Dispatch agent plans and re-sequences against more than 250 real-world constraints per computation, including time windows and shelf-life limits, so a plan that would deliver product outside its usable window is infeasible rather than merely expensive. The Hub agent models facility readiness and dwell, which is where detention is either absorbed or surfaced.

Governance is what makes this deployable in a food business. Six mechanisms bound autonomous action, including autonomy levels and human-in-the-loop override, so an operation can run automatic re-sequencing and capacity substitution while holding product disposition and quality decisions with a person.

Locus has been recognized by Gartner for seven consecutive years, featured in the 2026 Hype Cycle for Supply Chain Execution and Logistics Technologies, named a Leader in TMS by QKS Group (SPARK Matrix), and ranked #1 in Route Planning on G2’s 2026 Best Software Awards. In October 2025, Ingka Investments, the investment arm of Ingka Group, the world’s largest IKEA retailer, acquired Locus. Locus continues to operate independently.

Two North American deployments show the two halves of the problem.

A leading Canadian grocery brand delivers fresh and perishable food, from weekly meal kits to grocery essentials, into homes across more than 30 cities, moving through a national fulfillment network and into the hands of multiple contracted 3PL carriers. Every order raced a freshness clock, and the sourcing decision was entirely manual: warehouse associates logged into each carrier’s website to create orders and generate labels one at a time, while carrier choice was a judgement call made by checking each order against serviceability sheets line by line, validating addresses by hand, then comparing rates and ETAs order by order. The allocation logic lived in planners’ heads rather than in a system, and once a shipment left the dock there was no visibility at all. For perishable freight, every hour of that data entry was freshness lost in transit. After moving order creation and carrier selection into autonomous orchestration, the brand reported 33% faster deliveries, 15% lower fulfillment costs, 25% less time spent on manual shipping tasks, and customer support resolution 10 to 20 times faster.

A Fortune 50 parcel and logistics leader demonstrates mixed-pool governance at the scale harvest demands. Its 4,500-strong driver pool spanned captive shifts running zone-based routing and third-party carriers requiring tendering and on-demand assignment, with no single tool unifying them, which meant capacity in one pool could not absorb pressure in another. Locus deployed Capacity and Carrier agents to govern the entire pool under one policy, running zone-based, tendering, dynamic, and on-demand logic inside a single decision engine. Weekly execution across 51 service-center locations moved from 75% to 92%, and a single-site analysis surfaced $565,000 in unused capacity, including premium-tier service given away on cheaper classes, which scaled to more than $14 million annualized across 25 sites.

That last detail is the harvest lesson in miniature. The capacity was not absent. It was mispriced and unallocated, which is what happens when three pools are managed as three separate conversations.

Also Read: The Back-to-School Capacity Trap: Why Static Fleet Planning Breaks Under Predictable Surges

What to do before peak week

Five actions, and none of them requires new software to begin.

1. Price your decay curve by commodity. Establish, per commodity group, what one hour and one day of delay cost in expected markdown and shrink. Without this number the waterfall will win every argument, because freight rates are visible and decay is not.
2. Measure captive surge capability in driver hours, not trailers. A fleet at high hours utilization has no surge capacity regardless of equipment count, and this is routinely discovered during peak rather than before it.
3. Negotiate tiered commitments before the season. Base, expected-peak, and a deliberately uncommitted tail, agreed together while you still have leverage rather than during the week you need capacity.
4. Wire the USDA availability rating into your weekly operating review. Track it for every origin you draw from and define what happens when a lane crosses into Slight Shortage. A free government signal that triggers no action is not an early warning.
5. Get harvest date and cooling history into the dispatch layer. If remaining shelf life lives only in a grading or packhouse system, the allocation matrix cannot run, and every decision defaults to the freight-rate waterfall.

Item one is the unlock. Every other decision in this playbook depends on being able to compare a freight premium against a decay cost, and most produce operations have never calculated the second number.

The sourcing question, not the routing question

Harvest exposes an operation’s capacity strategy more sharply than any other period, because it is the one time when volume, volatility, and perishability peak together. The routing question, how to sequence what you have, is well understood and reasonably well served. The sourcing question, where the next two loads come from and what waiting costs, is usually answered by a dispatcher under pressure with no decay model and no unified view of three capacity pools.

Fix that and the routing improvements compound. Leave it and the best route optimization available is optimizing a plan built on the wrong capacity.

Book a Locus demo to model your harvest capacity mix across captive, contracted, and spot pools with shelf life priced into the allocation.