Amazon moves more freight than most trucking companies.
It operates one of the largest logistics networks in the world, delivering millions of packages daily through a system that includes Amazon Relay, Delivery Service Partners, and Amazon Flex.1
At every level, Amazon sets the standards, controls the technology, monitors performance in real time, and retains the power to deactivate carriers and drivers who fail to meet its metrics. Yet when one of these vehicles causes a crash, Amazon’s first defense is that it is not the employer, not the carrier, and not responsible. The driver works for someone else. The truck belongs to someone else. Amazon merely arranged the transportation.
This defense is under sustained legal attack, and the architecture of Amazon’s control over its delivery and freight networks is the evidence that undermines it. The question in litigation is not what Amazon calls the relationship. It is what Amazon does within the relationship: how much control Amazon exercises over the when, where, how, and by whom of every delivery and every load.2 The more control Amazon exercises, the harder it becomes to maintain that the carriers and drivers are truly independent, and the stronger the argument that Amazon bears responsibility when the system it designed and managed produces a crash.
Amazon Relay: Freight Under Amazon’s Authority
Amazon Relay is the platform through which Amazon tenders freight loads to motor carriers for transportation between Amazon facilities.3 Carriers apply to join the Relay network and must meet specific safety and compliance requirements.
This includes FMCSA BASIC score thresholds for:4
- Unsafe Driving
- HOS Compliance
- Vehicle Maintenance
- Controlled Substances/Alcohol
- Driver Fitness
Carriers must also maintain minimum insurance coverage.5
Once onboarded, carriers book loads through the Relay portal or mobile app. Amazon sets the rates, delivery windows, and performance expectations.6 Amazon monitors carrier performance through a transparent scorecard system that tracks on-time delivery, tender acceptance, and safety compliance.7 Carriers with higher performance grades receive access to more loads and earlier booking opportunities, creating a direct financial incentive to accept every load and meet every deadline.8
Amazon requires all drivers hauling Relay loads to complete identity verification and background checks through the Relay app.9 Amazon tracks drivers’ progress in real time through the app’s GPS functionality. Amazon’s performance monitoring extends to individual driver behavior, not just carrier-level metrics.10
The carriers on Amazon Relay operate under their own FMCSA operating authority, not Amazon’s.11 Amazon does not hold itself out as the motor carrier of record.
The carriers are responsible for compliance with all Federal Motor Carrier Safety Regulations, including:12
- Driver qualification under 49 C.F.R. Part 391
- Hours-of-service compliance under 49 C.F.R. Part 395
- Vehicle maintenance under 49 C.F.R. Part 396
- Drug and alcohol testing under 49 C.F.R. Part 382
However, Amazon’s carrier selection process and ongoing performance monitoring create a factual record of control that intersects with the negligent selection theory discussed in our treatment of broker liability. Amazon accesses and evaluates each carrier’s CSA scores, insurance status, and safety data before approving the carrier for the network.13 If Amazon approves a carrier with deficient safety scores, or if Amazon continues to tender loads to a carrier whose safety performance has deteriorated after onboarding, Amazon’s selection and retention decisions become relevant to a negligent selection claim under the same framework applied in Miller v. C.H. Robinson Worldwide, Inc.14
The Delivery Service Partner Model
Amazon’s Delivery Service Partner program is structurally different from Relay but raises even more acute control questions. Under the DSP model, Amazon contracts with small, independently owned LLCs that hire drivers to deliver Amazon packages using Amazon-branded vans leased through Amazon’s fleet program.15
The DSP model was designed to give Amazon the operational benefits of a company-owned delivery fleet without the legal classification of the drivers as Amazon employees.16
Amazon controls the number of packages assigned to each route, monitors driver progress throughout the day, and contacts drivers directly if they fall behind schedule.18 Amazon limits each DSP to a maximum of approximately 40 vans to prevent the companies from becoming large enough to gain bargaining leverage.19
DSP drivers are employees of the DSP, not Amazon. But the degree of control Amazon exercises over the DSP’s operations raises the question of whether the DSP is a truly independent business or a pass-through entity that exists primarily to shield Amazon from employment liability.20 The Capital Law Review documented that the work of DSP drivers "is typically indistinguishable from that of W-2 employees" and that Amazon "directs drivers’ routes, controls the number of packages that are to be delivered on each route, tracks the drivers’ progress in real time, and even calls drivers" if they fall behind.21
The DSP vans are typically light enough to fall below the 10,001-pound GVWR threshold that triggers FMCSA regulation, meaning that DSP drivers are generally not subject to CDL requirements, hours-of-service regulations, or the Federal Motor Carrier Safety Regulations that govern commercial motor vehicle operations.22 This regulatory gap means that the safety protections that apply to Relay carriers, including driver qualification standards, ELD requirements, and systematic vehicle maintenance obligations, do not apply to DSP operations. The absence of federal oversight makes the state-law liability analysis even more important, because the common law duty of care is the primary safety standard governing DSP driver conduct.
The Control Question in Litigation
The central liability question in Amazon delivery and freight crash litigation is whether Amazon exercises sufficient control over the carrier or DSP to be held vicariously liable for the driver’s negligence, or whether Amazon’s relationship with the carrier or DSP is a genuine arm’s-length arrangement that shields Amazon from respondeat superior liability.
Courts apply control tests that examine the totality of the relationship, not just the contractual label.
The Restatement (Second) of Agency § 220 identifies several factors.23
For Amazon, these factors point in conflicting directions: Amazon supplies the vans, the technology, and the branded uniforms (suggesting employment), but the DSP hires, fires, and pays the drivers (suggesting independent contracting).24
The emerging legal analysis focuses on the functional reality of control rather than the formal structure of the contracts. A platform that sets the delivery windows, dictates the routes, monitors driver location in real time, imposes performance metrics that determine whether the carrier or DSP continues to receive work, and retains the unilateral power to deactivate carriers or DSPs that fail to meet its standards exercises a degree of control that is inconsistent with a true independent contractor relationship.25
Under 49 C.F.R. § 390.5, the FMCSA definition of "employee" includes independent contractors while in the course of operating a commercial motor vehicle.26 For Relay carriers operating trucks above 10,001 pounds GVWR, the statutory employment doctrine may apply regardless of the contractual classification, making Amazon’s Relay carriers statutory employees for purposes of the FMCSRs.27 For DSP operations below the GVWR threshold, the analysis depends entirely on state common law.
Alternative Liability Theories
Even if respondeat superior does not apply because the court finds the carrier or DSP is a genuine independent contractor, Amazon may face liability through alternative theories.
Negligent selection under the Restatement (Second) of Torts § 411 applies when a principal fails to exercise reasonable care in selecting a competent contractor.28 Amazon’s access to carrier safety data through FMCSA databases, combined with its own internal performance monitoring, gives Amazon the information necessary to identify unsafe carriers and DSPs. If Amazon continues to tender loads to a carrier with deteriorating safety scores, or continues to contract with a DSP whose drivers have a pattern of accidents, Amazon’s selection and retention decisions become independently negligent.29
Non-delegable duty theory may also apply. Under the Restatement (Second) of Torts § 428, when a person undertakes an activity under a franchise or privilege granted by public authority, the obligation to perform that activity safely cannot be delegated to an independent contractor.30 Amazon’s delivery network is integral to its business operations, and the argument that Amazon cannot delegate its safety obligations to DSPs or carriers has been advanced in multiple jurisdictions.31
Apparent agency or ostensible agency theory applies when a principal creates the appearance that the contractor is its agent. Amazon-branded vans, Amazon-branded uniforms, Amazon-branded route planning, and Amazon-branded customer communications create a strong impression that the DSP driver is Amazon’s employee.32 A reasonable consumer who sees an Amazon-branded van making deliveries is unlikely to understand that the driver works for a separate LLC with no direct employment relationship to Amazon. This appearance of agency may be sufficient to impose liability on Amazon under the doctrine of apparent authority.33
What Discovery Should Target
Discovery in an Amazon crash case should be designed to expose the degree of control Amazon exercises over the carrier or DSP and to establish the functional reality of the relationship regardless of the contractual label.
Key categories include:
- The complete carrier or DSP agreement with Amazon, including all policies, performance standards, and termination provisions.
- Amazon’s performance scorecard data for the carrier or DSP, showing how Amazon monitored and evaluated safety and delivery metrics.34
- All communications between Amazon and the carrier or DSP regarding the load or delivery assignment that ended in the crash, including any time pressure, schedule changes, or performance warnings.
- Amazon’s deactivation policies and any records showing carriers or DSPs that were deactivated for performance failures, establishing that Amazon retained and exercised the power to terminate the relationship.35
- The carrier’s or DSP’s FMCSA safety data, CSA scores, and inspection history at the time Amazon approved the carrier for its network, establishing what Amazon knew about the carrier’s safety record.
- Amazon’s internal safety monitoring data for the carrier or DSP, including any incident reports, complaints, or safety alerts.36
- The driver’s background check records submitted through the Relay app or DSP onboarding process.
- For DSP cases, photographs and video showing the Amazon branding on the van, the driver’s uniform, and any customer-facing Amazon identification.37
- Amazon’s route planning data and package assignment data for the delivery that preceded the crash, showing the workload and time pressure the driver was under.
- Any prior crashes or safety incidents involving the same carrier, DSP, or driver, and Amazon’s response to those incidents.38
The objective is to demonstrate that Amazon’s control over the carrier or DSP was sufficiently pervasive that the independent contractor label does not reflect the functional reality of the relationship, and that Amazon’s selection, monitoring, and retention decisions contributed to the conditions that produced the crash.
The Performance Algorithm as Evidence of Control
Amazon’s performance monitoring systems are not passive data collection tools. They are active management instruments that shape driver behavior through real-time feedback, scoring, and the threat of deactivation. For Relay carriers, Amazon’s performance scorecard tracks on-time delivery percentage, tender acceptance rate, and safety compliance across every load.39 Carriers whose scores fall below Amazon’s thresholds lose access to premium loads, contracts, and early booking opportunities.40 For DSPs, Amazon monitors individual driver metrics including delivery speed, route adherence, package handling, and customer complaints, and uses these metrics to evaluate the DSP’s overall performance.41
The algorithm-driven performance system is relevant to both the control analysis and the safety analysis. On the control side, a platform that uses performance algorithms to determine which carriers receive work and which are deactivated is exercising a degree of managerial control that is functionally equivalent to an employer’s power to assign, evaluate, and terminate. The carrier or DSP does not negotiate the performance standards. Amazon sets them unilaterally. The carrier or DSP does not choose which metrics matter. Amazon’s algorithm decides. The carrier or DSP does not control whether it continues to receive work. Amazon’s scoring system determines that outcome.42
On the safety side, the performance algorithm creates pressure to prioritize speed and compliance with Amazon’s delivery metrics over compliance with federal safety regulations. A Relay carrier whose tender acceptance rate is declining because drivers are refusing loads that would require hours-of-service violations faces a choice between safety compliance and continued access to Amazon’s freight network. A DSP whose drivers are falling behind on package counts because they are taking legally required breaks faces pressure to push drivers harder or risk losing the Amazon contract. The algorithm does not explicitly instruct the carrier or DSP to violate safety regulations, but it creates an economic environment in which compliance with the algorithm and compliance with federal safety law may be incompatible.43
In discovery, the algorithm’s parameters, thresholds, and consequences should be obtained and analyzed. The specific metrics Amazon uses to evaluate performance, the scores that trigger warnings or deactivation, and the historical performance data for the carrier or DSP involved in the crash all establish the degree to which Amazon’s system influenced the operational decisions that preceded the crash.
The Deactivation Power
Amazon’s unilateral power to deactivate carriers and DSPs is among the strongest indicators of control. In traditional independent contractor relationships, neither party can unilaterally terminate the relationship without consequence. The contractor has the right to complete the work and seek payment. In Amazon’s system, Amazon can deactivate a carrier or DSP from the platform immediately, cutting off the carrier’s or DSP’s access to Amazon’s freight or delivery volume with no obligation to provide advance notice, a hearing, or an opportunity to cure.44
The deactivation power is functionally equivalent to the power to fire. A carrier that depends on Amazon Relay for a significant portion of its revenue cannot afford to be deactivated. A DSP that leases its vans through Amazon’s fleet program and operates exclusively for Amazon cannot survive deactivation. The threat of deactivation creates a coercive dynamic in which the carrier or DSP must comply with Amazon’s performance expectations or face economic destruction.
In the context of crash litigation, the deactivation power is relevant in two ways. First, it establishes that Amazon exercised the kind of control over the carrier or DSP that is inconsistent with a genuine independent contractor relationship. An entity that can terminate the relationship at will, without contractual constraint, is exercising employer-level control.45 Second, it establishes that the carrier or DSP was under pressure to meet Amazon’s standards, which may have contributed to the safety violation that caused the crash. A driver who was speeding to meet a delivery window, or a carrier that dispatched a fatigued driver to avoid missing a tender, may have been responding to the implicit threat that failure to perform would result in deactivation from the platform that sustains their business.46
Amazon’s deactivation records should be a priority in discovery. The number of carriers and DSPs Amazon has deactivated, the reasons for deactivation, and the process by which deactivation decisions are made all illuminate the degree of control Amazon exercises and the pressure that control creates.47
Sources
- [1] James Reif, A Prime Opportunity for Tort Law Developments: Liability Issues Related to Amazon's Delivery Service Partner Program, Capital University Law Review.
- [2] Restatement (Second) of Agency § 220 (1958).
- [3] Amazon Relay, Frequently Asked Questions.
- [4] Amazon Relay, How to Become a Carrier for Amazon Relay.
- [5] Amazon Relay, Frequently Asked Questions.
- [6] Amazon Relay.
- [7] Id.
- [8] Id.
- [9] Amazon Relay, Frequently Asked Questions.
- [10] Id.
- [11] 49 C.F.R. § 390.5.
- [12] 49 C.F.R. Parts 391, 395, 396, and 382.
- [13] Amazon Relay, How to Become a Carrier for Amazon Relay.
- [14] Miller v. C.H. Robinson Worldwide, Inc., 976 F.3d 1016 (9th Cir. 2020).
- [15] James Reif, A Prime Opportunity for Tort Law Developments, Capital University Law Review.
- [16] Id.
- [17] Id.
- [18] Id.
- [19] Id.
- [20] Id.
- [21] Id.
- [22] Id.
- [23] Restatement (Second) of Agency § 220 (1958).
- [24] James Reif, A Prime Opportunity for Tort Law Developments, Capital University Law Review.
- [25] Id.
- [26] 49 C.F.R. § 390.5.
- [27] Johnson v. S.O.S. Transport, Inc., 926 F.2d 516 (6th Cir. 1991).
- [28] Restatement (Second) of Torts § 411 (1965).
- [29] Christopher Dolan et al., Sorting Out All the Possible Tortfeasors in Truck Crash Cases, Advocate Magazine (April 2023).
- [30] Restatement (Second) of Torts § 428 (1965).
- [31] Finding Liability and Insurance in Cases Involving Motor Carriers, Plaintiff Magazine.
- [32] James Reif, A Prime Opportunity for Tort Law Developments, Capital University Law Review.
- [33] Id.
- [34] Amazon Relay, Frequently Asked Questions.
- [35] James Reif, A Prime Opportunity for Tort Law Developments, Capital University Law Review.
- [36] 49 C.F.R. § 390.15, Assistance in investigations and special studies.
- [37] James Reif, A Prime Opportunity for Tort Law Developments, Capital University Law Review.
- [38] 49 C.F.R. § 396.3, Inspection, repair, and maintenance.
- [39] Amazon Relay, Frequently Asked Questions.
- [40] Amazon Relay.
- [41] James Reif, A Prime Opportunity for Tort Law Developments, Capital University Law Review.
- [42] Id.
- [43] 49 C.F.R. Part 395, Hours of service of drivers.
- [44] James Reif, A Prime Opportunity for Tort Law Developments, Capital University Law Review.
- [45] Restatement (Second) of Agency § 220 (1958).
- [46] 49 C.F.R. § 392.3, Ill or fatigued operator.
- [47] James Reif, A Prime Opportunity for Tort Law Developments, Capital University Law Review.