Product People Enabled Scalable Driver Enablement Automation for a European Mobility Platform

The Client

The client is a large European mobility platform operating a multi-sided marketplace that connects drivers with end users across several regulated markets. Driver onboarding and compliance are central to supply growth, operational efficiency, and regulatory risk management.

At the centre of this sits the Driver Enablement domain: registration flows, document validation (DocVal), and on-car advertising (OCA) checks. Thousands of drivers onboard every month, which means small quality issues become large operational costs fast, and small compliance mistakes become existential risks.

‍

The Mission: Interim Product Manager for Driver Enablement

During a critical period of automation rollout and market expansion at the client, the Driver Enablement domain lacked dedicated product ownership. At the same time, the team was managing multiple interconnected initiatives: document validation automation, on-car advertising (OCA) validation automation, and changes to driver registration flows.

Manual review still dominated large parts of the client’s onboarding process, creating bottlenecks for drivers and significant operational overhead. Automation existed, but quality issues—false rejections, inconsistent metrics, unclear go-live criteria—limited its impact and made scaling risky.

Product People Trio stepped in to own the full scope across these tightly coupled domains, coordinating Product, Data Science, Operations, Legal, and an external engineering vendor. The mission ran for ~6 months with clear expectations: reduce manual checks, improve turnaround time for drivers, and build scalable automation foundations without compromising compliance.

The mission required navigating complex, evolving priorities across interconnected domains. Requirements shifted as market needs clarified, the data infrastructure needed targeted enhancement, and regulatory risk profiles varied significantly between document validation and on-car advertising use cases. Progress depended not just on shipping features, but on making defensible decisions with incomplete information and high accountability.

‍

Main Challenge: Document Validation Automation at Scale

The Problem

Document Validation sat at the core of driver enablement, directly shaping onboarding speed, operational cost, and activation rates. Across multiple European markets, Operations agents manually reviewed high volumes of driver-submitted documents—licenses, registrations, insurance, permits—each with market-specific formats and rules.

At the same time, automation quality was uneven. False rejection rates (FRR) exceeded 20% on several high-impact document types, meaning legitimate drivers were blocked or delayed. This damaged conversion and eroded trust at the very first interaction with the platform.

Crucially, the trade-offs were uneven: false acceptances carried high regulatory risk, while false rejections harmed conversion and trust. False acceptance risk (FAR) carries severe regulatory and fiscal implications: approving non-compliant or fraudulent documents could expose the business to investigations, fines, or market-level shutdowns. Any automation strategy had to aggressively reduce FRR while keeping FAR within strict thresholds (~2% or below).

This framing shaped every decision. Automation could not be “faster at any cost.” It had to be provably safe.

Our Solution

Rather than treating DocVal as a collection of isolated document issues, we approached it as a system-level automation and quality problem, combining targeted technical fixes with clear governance.To focus effort where it mattered most, we used a structured prioritisation approach: we ranked document flows by performance impact and operational volume, then targeted the highest-leverage issues first (the ones driving the largest share of false rejections and manual workload). This ensured engineering work translated into measurable improvements in automation quality and driver experience.

One of the highest-impact interventions was the introduction of “most recent document” cross-check logic, preventing outdated submissions from triggering false rejections.This change alone reduced false rejections by about 23 percentage points on the affected flows, meaning we eliminated roughly one in four incorrect rejections in those document journeys.

In parallel, we progressed foundational work on automated data pre-population, aligning with the Drivers Quality team to move overwrite logic from concept into implementation planning. This laid the groundwork for reducing redundant data entry once automation confidence thresholds were met.

We also delivered market-specific models into production, each requiring tailored validation logic due to local document formats and regulatory differences. Iterative fixes were coordinated closely with the external engineering vendor, based on live performance data rather than assumptions.

To prevent premature rollouts, we established a formal quality gating protocol. Go-live decisions required evidence across multiple dimensions: field extraction accuracy (~93%), FAR within thresholds (~1.7–1.9%), automation coverage (~26–28%), and demonstrated FRR improvement versus baseline. Where quality failed the gate, rollout paused until fixes were deployed and re-measured.

A less visible but critical part of the work involved fixing reporting integrity. Dashboards misclassified shadow-mode automation as live, excluded subsets of tickets, or were skewed by pending backlogs. Without reliable metrics, automation decisions became slow and political. Correcting query logic and metric definitions became part of delivery, not a side task.

Evidence

• Global automation reached ~31%, with one core market exceeding 50%, representing a material improvement from the prior baseline
• 23pp FRR reduction achieved on key document flows after targeted logic changes
• Multiple market models delivered to production with ongoing optimization
• A shared compliance severity scale and weekly investigation cadence were established across Product, Data Science, and Operations

‍

Secondary Challenge: OCA Automation Strategy

The Problem

On-car advertising validation (OCA) was operationally expensive, high-volume, and dependent on subjective manual judgment. Drivers would submit photos to prove they had placed required advertising stickers on their vehicles, and Operations agents would manually approve or reject the submissions.

Unlike DocVal, the risk profile was inverted: false rejections carried greater brand and driver trust risk than false acceptances. Drivers doing free advertising but being denied benefits created immediate friction, churn risk, and negative brand perception. Meanwhile, false acceptances (approving invalid stickers) were tolerable within limits: it was a commercial risk, not a regulatory catastrophe.

The team faced strategic ambiguity: should automation start conservatively (acceptances only) or aggressively (first-attempt acceptances + rejections)? The choice would determine operational savings velocity and quality risk exposure.

Our Solution

We aligned stakeholders around a clear strategic direction: maximise first-attempt automation (both acceptances and rejections) with layered safeguards, explicitly prioritising FRR reduction:

• Launched acceptance automation live across all supported markets as the foundation layer.
• Deployed first-attempt rejection automation for several markets, and validated additional markets for rollout. This was the boldest automation path, immediately reducing Ops workload, but with guardrails.
• Established lightweight QA cadence (weekly spot checks, sample logging) appropriate to OCA's bursty volumes, unlike DocVal's daily high-volume review needs.
• Improved driver communications and email copy to reduce resubmission confusion and provide clearer instructions for valid submissions.
• Addressed reporting integrity issues: pending backlog distorted weekly metrics, dashboards pulled wrong subsets, and simulation results required manual validation by Ops leads to establish ground truth.

Evidence

• ~98% global automation, with ~40% automatically approved and ~57% automatically rejected
• Incorrect rejections prioritised as primary quality focus; incorrect acceptances treated as secondary (tolerable within limits)
• Estimated operational workload reduction equivalent to ~80 Ops hours/month through automation at scale
• Clear go-forward strategy established: maximise automation on first attempt + strong safeguards + lightweight QA, with explicit FRR-over-FAR prioritisation due to user experience risk

‍

Side Quests: Registration Flow Migration & Market Expansion

A particularly complex initiative involved a high-risk registration flow migration, driven by regulatory and commercial considerations. Supporting multiple parallel flows introduced significant backend coupling complexity, meaning any solution had to be permanent rather than a temporary workaround.

Decision-making was grounded in impact metrics: fleet size, driver counts, and long-tail effects, all to justify scope. Along the way, evolving requirements surfaced real blockers: age validation logic, bank detail consistency, broken company creation flows due to missing mandatory fields, tooling readiness, translations, and permission gaps.

Despite these constraints, the client’s migration shipped with zero post-launch rollback incidents, supported by close coordination across Legal, Operations, Engineering, and Product.

Additional work included releasing streamlined registration flows across multiple markets, completing discovery for future short-flow expansions, and aligning API-first onboarding plans with legal and operational constraints.

‍

Mission Achievements: Delivered Outcomes

💡 Scaled DocVal automation to ~31% globally, with one core market exceeding 50%, through targeted technical interventions and disciplined quality gating

💡 Achieved ~98% OCA (On-car Adv) automation (40% auto-approved, 57% auto-rejected), saving ~80 Ops hours/month, with explicit strategic prioritization of FRR reduction over FAR due to user experience risk

💡 Delivered a high-risk registration flow migration with zero post-launch rollback incidents despite backend coupling and regulatory complexity

💡 Established a quality gating framework enabling safe, scalable automation across markets

💡 Restored trust in metrics by resolving reporting integrity issues, enabling faster, clearer go-live decisions and reducing stakeholder friction

💡 The mission was extended based on delivery outcomes and stakeholder confidence in navigating high-stakes automation decisions