Automating Support Operations at Shell Recharge

Executive summary

Shell Recharge operates one of the world’s largest electric vehicle charging networks, supporting enterprise fleet clients and individual consumers across multiple markets. When I joined as Principal Product Designer the team in Amsterdam, the customer support operation was in its inception and very inefficient. A 30-person team managing B2B and B2C queries across three tiers in Salesforce Lightning Console, with no routing logic and a “double call” problem. Agents frequently had to contact customers a second time to collect information they had missed to capture during the initial call.

The objective was to take care of the entire customer, technical, and financial support services. Eliminate reactive, labour-intensive support through automation, reducing resolution times, cutting cost-per-ticket, and rebuilding the support architecture around a single, streamlined channel.

I held design ownership, operating as the senior UX voice within a cross-functional team spanning sales, support operations, engineering, and Salesforce platform specialists. My scope covered the complete support journey, from customer-facing touchpoints through to agent tooling and internal frameworks, across a two-year contract.

Discovery & strategic insights

Diagnosing the problems

The support operation’s biggest inefficiency was slow and inefficient data capture at first contact. Agents handling queries lacked intake tooling, which meant diagnostic information was often missed during the initial interaction. The result was a cycle of follow-up calls, extended resolution timelines, growing backlogs, and high cost-per-ticket.

Mapping the full support journey from customer contact to technical resolution revealed that the escalation model was not optimised for the actual query patterns. The majority of queries followed predictable, automatable logic flows that could be resolved without human intervention. Only a minority required genuine specialist intervention. However, the three-tier escalation model treated all queries as if they required human handling, creating unnecessary workload and inefficiency.

Translating complexity into a design strategy

The support handled a technically complex cases such as EV charging infrastructure, hardware diagnostics, and installations. Mapping query patterns against resolution paths identified a clear segmentation: the majority of queries followed predictable logic flows. Only a minority required specialist intervention.

This became the foundation for the redesign. Rather than optimising a three-tier escalation model, the strategy shifted to automating the majority and specialising the few, consolidating human support to a single tier reserved for complex cases.

Introducing roadmap prioritisation

To move product decisions from opinion to objective evidence, I introduced a prioritisation framework to the design roadmap. Using RICE scoring, MoSCoW analysis, and Impact vs. Effort mapping, every UX initiative was evaluated against business value and feasibility before entering the backlog. When user needs conflicted with business objectives, the framework made trade-offs visible, giving stakeholders a transparent basis for prioritisation rather than competing intuitions.

The solution

Omni-Channel routing architecture

I re-designed the support journey built on Salesforce Einstein Omni-channel with smart routing integrated at the customer touchpoint. Instead of connecting customers to an agent and then triaging, the system captured intent before connection. Customers typed or spoke their issue; keyword analysis and query classification triggered automated logic flows that either resolved the case through self-service or routed it directly to the appropriate specialist.

The structure was consolidated into a single specialist line, reserved for non-English speaking markets like Spain and France where infrastructure was still maturing and automation was limited.

Automating customer requests with Einstein - detecting query type and routing to the right support group, reduced resolution times and improved user satisfaction.

Smart forms for agent capturing of information

To eliminate the double-call problem at its source, I customised the Salesforce Lightning Console with pre-built, intake forms mapped directly to query type. Front-line agents were guided through capturing every required data point during the first interaction with no exceptions and no gaps. Completed data was immediately available in the technical team’s queue, giving specialists full context. Follow-up calls became rare.

Accessibility remediation

I led a full accessibility programme across all customer-facing touchpoints, spanning the sales, installation, account management, and payment flows. This ensured the platform met WCAG compliance standards across markets and user groups, and embedded accessibility as a design requirement.

Impact & business value

Results were validated across the full two-year contract period.

Quantifiable outcomes

• Support team reduced from 30+ agents to 5 specialists handling complex escalations only; all other query types resolved through automation
• 80% of support tasks fully automated, resolving without human intervention
• 98% automation success rate achieved on automatable query types
• UX initiatives tied directly to KPIs, CLTV, and retention metrics through the prioritisation framework
• Double-call eliminated through structured Salesforce Lightning forms

Qualitative outcomes

• A data-driven prioritisation framework shifted roadmap from subjective stakeholder opinion to evidence-based decision-making, giving the product team a process for evaluating design investment
• Accessibility across all touchpoints established compliance as a design standard embedded into the product lifecycle.
• The single-tier support model created a leaner, more focused team with higher case complexity and clearer criteria