Dynamex ABC Test Application in California Delivery Litigation

California's test for who counts as an employee no longer asks what feels fair. It asks three fixed questions, in order, and a delivery company has to answer yes to all three or lose. This piece walks through how that test actually gets applied against delivery operators in court, what a 2021 retroactivity ruling and a 2026 Supreme Court arbitration decision changed about the exposure, and what the operational record shows about which business practices keep appearing in the losing column. Knowing which prong is in play, and why, is the first step toward managing the risk before it surfaces in a complaint.

What the Dynamex ABC test is and how it displaced the Borello standard for delivery companies

Until 2018, California used a multi-factor test known as Borello, named for the case that set it up. Dynamex Operations v. Superior Court, decided by the California Supreme Court in 2018, threw that balancing act out for a large category of claims and replaced it with something far less forgiving.

The ABC test does not balance anything. Fail any one of the three, and the worker is an employee for purposes of the wage orders at issue. The California legislature then took this a step further with AB 5, which made the ABC test the default standard not just for wage-order claims but for the Labor Code and Unemployment Insurance Code as well.

For an operator running a courier business, Prong B is where the test turns into a trap door. A company whose entire purpose is moving packages or food from one place to another cannot plausibly argue that the people doing the moving are performing work outside its usual course of business. That single prong does most of the damage in delivery misclassification cases, and the rest of this piece explains why.

Retroactivity and Present-Day Litigation Exposure for Old Contractor Arrangements

A test only matters if it reaches backward as well as forward, and on January 14, 2021, the California Supreme Court made sure this one does. Jan-Pro Franchising International, the court held that Dynamex applies retroactively, meaning contractor arrangements set up years before the 2018 decision can be evaluated under the ABC test today, not under whatever standard was in effect when the arrangement was built.

The court's logic was that Dynamex did not invent a new rule so much as interpret language, "suffer or permit to work," that had already existed in California's wage orders for decades. The issue was open the whole time, and the court also found the ABC test was foreseeable anyway, since its three elements were already sitting inside Borello's factor list.

Wage-order claims involving minimum wage or overtime, pending as of April 2018 or still within the statute of limitations for pre-Dynamex conduct, now get decided under the ABC test. Claims brought purely under the Labor Code, such as business expense reimbursement, are not subject to this retroactive reach for conduct before January 2020, when AB 5 took effect. That distinction matters for lawyers structuring a defense, but for operators the headline point is simpler. A contractor fleet put together in 2014, 2015, or 2016 gets no credit for being built before Dynamex existed. The statute of limitations, not the calendar, decides what is still exposed.

Prong B in Court Against Delivery Companies

Prong B asks whether the worker's services fall outside the hiring company's usual line of business. For a last-mile or courier operation, that question answers itself before the lawyers even get involved, since delivering things is the business. That single prong can be enough to sink a contractor classification regardless of how carefully a company structures the other two.

A 2026 California state court case against a company called Dlivrd shows how plaintiffs turn that structural weakness into a filed complaint. The suit alleges that Dlivrd drivers pick up, transport, deliver, and set up catering and other high-value orders, which is precisely what Dlivrd sells to its customers, and that drivers operate under company control to hit customer satisfaction standards. That is a Prong B argument built directly from the company's own sales pitch: if the service drivers perform is the service the company advertises, it cannot also be outside the company's usual business.

The same complaint piles on evidence that reads as control under Prong A: Dlivrd allegedly pays a flat fee that is not open to negotiation, tracks drivers in real time through its app, ranks and prioritizes drivers using customer ratings and other performance metrics, maintains a policy for terminating or deactivating drivers whose metrics fall short, and directs how drivers handle catering set-up at the delivery site. None of these are exotic tactics. They are standard features of a well-run delivery operation, which is exactly the problem: plaintiffs in these cases consistently point to app-based tracking, non-negotiable pay, performance scoring, and deactivation policies as their factual basis.

What GrubHub's California class action shows is not a verdict on the merits but a timeline. The case covered a large class of food delivery drivers and settled only after more than ten years of litigation, including class certification, summary judgment motions, and five separate mediations. The settlement required a payment, not a reclassification of anyone's employment status. But a decade of litigation costs money and management attention regardless of who wins, and that cost is itself a business risk independent of the eventual outcome.

How Prongs A and C compound the exposure when operational control is embedded in delivery technology

The tools that make a delivery network run well, like app-based dispatch, location tracking, customer ratings, deactivation rules, and fixed pay rates, are the same tools courts point to as evidence of control under Prong A. The two goals pull in opposite directions.

Prong A asks whether the worker is free from the hiring company's control in how the work gets done, and courts look at who sets the routes, who tracks location, who dictates how a driver talks to the customer, and who controls timing. The Dlivrd complaint checks every one of those boxes: live app tracking, ratings-based dispatch, flat pay with no negotiation, and specific direction on how catering gets set up at the door. These common industry practices are also the first things a plaintiff's attorney looks for.

A Minnesota attorney general case against Shipt, while not a California precedent, shows the same argument pattern playing out in a different jurisdiction. The case shows regulators building a Prong-A-style argument around scheduling control and outside-work restrictions, a pattern that appears in California complaints as well.

Because the three prongs are conjunctive, a company has to win all three to classify a worker as a contractor, and losing any single one is enough to make that worker an employee for wage-order purposes. A delivery company is exposed on Prong A, Prong B, and Prong C simultaneously, and a weakness in one does not get offset by strength in another.

What Proposition 22 Covers for Delivery Operators

Proposition 22, passed by California voters, carves out an exception for app-based rideshare and delivery platforms, but the exception comes with conditions. Drivers on these platforms keep contractor status so long as the company does not require them to work specific days or hours, does not require them to accept a minimum number of rides or deliveries, does not require them to accept any particular service request, does not stop them from working for competing delivery or transportation companies, and does not stop them from working any other lawful job.

Prop 22's protection only reaches consumer-facing, app-based platforms. B2B delivery operations, last-mile logistics providers, courier services, and freight distribution companies that do not run through that kind of app are left out entirely and remain fully subject to the ABC test with no exception available. Even for companies that do qualify, Prop 22 does not end the litigation. A platform fully compliant with Prop 22 can still be fighting a parallel government lawsuit over conduct from before the exemption existed.

How the U.S. Supreme Court's May 2026 Arbitration Ruling Closes a Procedural Escape Route for Delivery Companies

On May 28, 2026, the U.S. Supreme Court issued a unanimous ruling that changes how far delivery companies can push workers into arbitration instead of court. The decision held that workers, including certain last-mile delivery drivers, who handle an intrastate leg of a continuous interstate shipment may qualify for the transportation worker exemption under Section 1 of the Federal Arbitration Act, even if the driver personally never crosses a state line and even if every mile the distributor runs stays inside one state.

Forced arbitration clauses had been the main procedural tool delivery companies used to keep misclassification disputes out of class action courtrooms, routing them instead into individual arbitration proceedings where the stakes and the plaintiff's leverage are both much smaller. The May 2026 ruling makes that scale easier for plaintiffs' counsel to reach.

The pattern already played out once in the Uber case. The California Attorney General's lawsuit against Uber was stayed while Uber tried to compel arbitration, and only after the California Supreme Court denied review did the stay lift and the case return to Superior Court. That sequence, arbitration motion, stay, eventual return to open court, is the exact playbook the May 2026 ruling now makes harder for delivery companies to run again.

What the litigation record reveals about which operational facts consistently drive misclassification findings

Lay the GrubHub settlement, the Dlivrd complaint, the Shipt settlement, and the California Attorney General's posture toward Uber next to each other, and the same five operational facts show up as evidence every time: flat pay rates that drivers cannot negotiate, real-time tracking through a company app, dispatch systems that prioritize drivers by customer rating, deactivation triggered by performance metrics, and specific direction on how to interact with the customer.

Flat, non-negotiable pay does double duty against a company's defense. It undercuts Prong C, since a worker who cannot set or negotiate a price is hard to describe as running an independent business, and it reads as control under Prong A at the same time. Real-time app tracking and ratings-based dispatch function as a substitute for a human supervisor standing over a worker's shoulder, and courts treat them that way under Prong A. Deactivation tied to response rate, customer ratings, or on-time performance functions as a disciplinary system, the kind an employer runs, not the kind one independent business runs against another. The Shipt settlement made this concrete: the Minnesota Attorney General required written explanations before deactivation and a formal appeals process, treating the deactivation power itself as something that needed guardrails. Specific direction over how a driver handles the final steps of a delivery, like the catering set-up detail in the Dlivrd complaint, pushes control into the last moment of the job and makes the work look indistinguishable from what an employee on staff would do.

None of these facts come with an expiration date. If a company built real-time tracking, flat pay, and a ratings-based deactivation system into its platform back in 2015, the retroactivity rule from Vazquez means that platform design is live evidence in a 2026 lawsuit, not a historical footnote protected by when it was built.

Structuring Contractor Relationships and Compliance Infrastructure to Reduce ABC Test Exposure

Nothing in the operational record above suggests the ABC test can be out-argued. It has to be out-designed. The companies with the least exposure are the ones that treat Prong A, B, and C as design constraints on the business itself, not as a legal brief written after the fact to describe practices that were never built with the test in mind.

On Prong A, that means pulling back from real-time, continuous tracking and replacing it with pickup and delivery confirmation only, letting drivers set their own routes and schedules, and giving them real pricing input on their rates. Any contract language or app design that restricts a driver from working for a competing service should be reconsidered entirely, since that restriction is one of the clearest control signals a plaintiff's attorney can point to.

Prong B is structurally harder to fix, because it asks whether the work itself sits outside the company's usual business, and a delivery company cannot easily redefine what business it is in. Some operators manage this exposure by narrowing what the contractor role actually covers, separating the transportation leg from service work like setup or installation that genuinely sits outside the core delivery function, and documenting that separation in the contract and in practice. That will not convert a courier company into something exempt from its own purpose, but it can shrink the scope of what gets challenged.

For platforms that qualify, Prop 22 compliance has to be monitored on an ongoing basis, not treated as a one-time classification made at signup. And because the May 2026 arbitration ruling narrows the path to individual arbitration, companies need to assume that misclassification disputes are more likely to reach class certification than they were even two years ago, and budget legal and compliance resources accordingly. The GrubHub timeline, a decade from filing to settlement, is a reasonable planning horizon for what a serious class action actually costs, in time and attention, long before anyone finds out who wins.