Why Southern European Delivery Operators Are Redesigning Workforce Architecture

A Director of Operations at a Madrid-based food delivery platform sits through a vendor pitch on AI-powered driver retention. Predict churn 30 days out, intervene with targeted offers, lift retention. It’s a competent product. It also doesn’t address the actual problem on her desk.

The problem on her desk is that Spain’s Riders’ Law fundamentally changed what “retention” means four years ago. The riders her platform works with sit inside an evolving employment-status conversation. Multi-app driving is the norm, so retention against any single platform is structurally limited. The EU Platform Work Directive is being implemented across member states, including Italy and France where her operation also runs. In a workforce environment shaped this dramatically by regulatory and structural change, churn prediction is solving downstream of the actual problem.

European delivery operators in Spain, Italy, and France are increasingly recognising the same thing: retention isn’t a prediction problem. It’s a workforce-architecture problem. Predicting which gig driver is about to leave is useful only if there’s a viable intervention — and in markets where employment classification, algorithmic transparency rules, and multi-app driver economics are all in flux, the meaningful interventions happen at the level of workforce design, not at the level of retention algorithms.

According to the European Commission’s impact assessment for the EU Platform Work Directive, more than 28 million people work through digital labour platforms in the EU, with European delivery and ride-hailing among the largest categories. The Directive, formally adopted in 2024, is being implemented by member states through 2026. It is the most consequential change to European gig work this decade.

The Regulatory Reality Reshaping Southern European Operations

Three regulatory contexts shape how delivery operators in Spain, Italy, and France need to think about workforce architecture in 2026.

Spain. The Riders’ Law (Ley Rider), passed in May 2021, established a presumption of employment for delivery riders working through digital platforms and introduced algorithmic transparency requirements. The law has materially reshaped how delivery platforms operate across Madrid, Barcelona, and the Spanish market — including triggering the high-profile exit of one major platform shortly after passage. Operators continuing in Spain have re-architected workforce models around its requirements.

Italy. Italy has not enacted a single national reclassification law equivalent to Spain’s, but a series of court rulings — including notable actions in Milan and Rome — have reclassified specific platform riders as employees in particular cases. The legal landscape varies court by court, platform by platform, producing a fragmented but consequential operational reality.

France. France has a long-running platform-worker legal framework, evolving through multiple legislative cycles. French delivery operations in Paris, Lyon, and Marseille operate under specific worker-protection requirements that differ in structure from both Spanish and Italian frameworks. French unionisation in the delivery sector — particularly through coursiers organisations — adds an additional dimension.

The EU Platform Work Directive. Layered over all three national contexts, the EU Platform Work Directive introduces algorithmic management transparency requirements and presumption-of-employment provisions that member states are translating into national law through 2026. Operators planning multi-year workforce strategies are doing so against a regulatory baseline that will shift substantively across the next 18–24 months.

The operational implication for Directors of Operations is straightforward: workforce architecture decisions made in 2026 will be evaluated against a regulatory regime materially different from the one that existed when current platforms were designed.

Also Read: How AI Improves Driver Experience: Route Fatigue to Retention

Why Churn Prediction Misses the Point in This Market

Churn prediction is a technically sound capability. In a stable workforce environment with clearly defined employment relationships, predicting which workers are at risk and targeting interventions can deliver meaningful retention gains.

That is not the environment Southern European delivery operators are in.

Three structural realities limit what churn prediction can deliver in this market.

Multi-app driving is the norm. Most active delivery riders in Madrid, Milan, and Paris work simultaneously across multiple platforms. “Retention” against a single platform is structurally bounded by how attractive that platform’s earnings are versus the alternative apps the rider already has open on the same phone. A churn-prediction model that flags a rider as “at risk of leaving” often catches a rider who already left months ago, in the sense of allocating most of their hours elsewhere.

The interventions a model can trigger are constrained by employment status. A platform operating under reclassification pressure cannot freely vary individual rider compensation, schedule guarantees, or work allocation in the way a traditional gig platform could. The interventions that retention algorithms typically suggest — bonus offers, route preferences, schedule guarantees — interact with employment classification in ways that legal and compliance teams need to review.

Structural drivers of churn dominate individual ones. Earnings volatility, app algorithm changes, weather, fuel costs, and broader economic conditions drive the largest share of European gig delivery turnover. According to Eurofound research on platform work, structural earnings and working-condition factors consistently outweigh platform-specific retention efforts in shaping rider tenure decisions. A prediction model can identify these patterns but cannot fix them.

The result: in Southern European markets, the platforms achieving the strongest retention outcomes are typically the ones that redesigned the workforce architecture upstream of any prediction layer.

The Four Workforce-Architecture Decisions That Actually Move Retention

Operators building durable workforce strategies in Spain, Italy, and France are making four architectural decisions that shape retention more than any prediction model.

1. Employment-tier mix. Most large operators now run multi-tier workforces — some combination of employed riders, contracted riders on fixed-term arrangements, true gig riders where legally and operationally viable, and 3PL or agency-supplied riders. The architecture decision is not “should we have gig riders?” — it is “what work goes to which tier, and in what proportion.” Peak-hour flex can rest on gig and agency tiers; baseline urban density can rest on employed riders; specialty deliveries (high-value, regulated, premium SLA) typically benefit from employed-tier handling.

Also Read: AI Route Optimization to Deal with Europe’s Driver Shortage

2. Earnings design across tiers. Compensation transparency is a Riders’ Law requirement in Spain and an emerging norm under the Platform Work Directive. Operators designing earnings systems for predictability — guaranteed minimums, transparent bonus structures, comprehensible tip flows — consistently outperform operators relying on opaque algorithmic compensation. Multi-app driver economics matter here: an earnings system that reliably produces 10% better hourly than the competing app produces retention without prediction.

3. Algorithmic transparency. Spain’s Riders’ Law and the EU Platform Work Directive both require that algorithmic management decisions affecting workers — assignment, ranking, deactivation — be transparent and contestable. This shapes how dispatch and routing systems present decisions to riders. The same algorithm running with audit-grade explainability is operationally identical to one running without it; the difference is regulatory compliance and rider trust. Last-mile execution platforms that produce auditable decision logs for every routing and dispatch choice make this architectural requirement meaningfully easier to meet at scale.

4. Infrastructure and tooling investment. Rest areas, weather gear, training, communication tools, and grievance procedures are not soft benefits — they are operational infrastructure. According to research, operational investments in workforce conditions consistently rank among the highest-leverage retention drivers across gig and contractor workforces, particularly in markets where regulatory pressure is increasing the cost of treating riders as fully disposable.

The Director of Operations Evaluation Framework

Five questions for Southern European Directors of Operations evaluating workforce strategy in 2026.

1. What is our employment-tier mix today, and is it deliberate or inherited?
   Many operators are running tier mixes that emerged organically from rapid scaling, not from intentional design. Inherited mixes rarely survive regulatory scrutiny.
2. Are our earnings structures transparent and predictable enough to compete on a multi-app driver’s screen?
   If a rider running our app alongside two competitors cannot quickly tell which produces the better hour, the retention question is already lost.
3. Do our dispatch and routing systems produce auditable decision logs for assignment, ranking, and deactivation choices?
   Both Spain’s Riders’ Law and the EU Platform Work Directive require this. Adding it after the fact is materially more expensive than designing it in.
4. What is our investment in rider infrastructure — rest, training, communication, grievance procedures — relative to operators we benchmark against?
   Operational conditions show up directly in tenure data.
5. Is our workforce strategy designed for the regulatory environment of 2024, or the one expected by 2027?
   The Platform Work Directive’s national implementations will substantively change the baseline within most operators’ current planning horizon.

The Real Question for Southern European Directors of Operations

Predicting which rider is about to leave is interesting. Designing a workforce architecture that fewer riders need to leave is the actual lever. In Spain, Italy, and France — markets where regulatory direction is moving toward employment presumption, algorithmic transparency, and stronger worker protections — the question Directors of Operations should be asking isn’t “how accurate is our churn model?” It is: is our workforce architecture deliberate, regulatory-resilient, and competitive on a multi-app driver’s screen — or are we trying to predict our way out of design choices we never made intentionally?