At iFood, the Help Center was often the user’s last resort. Yet most users still couldn’t find what they needed, clarity suffered, and contact center pressure remained high. Internal teams across CX, Loyalty, Groceries, Fintech, Logistics, and more all needed a way to own and optimize support content.

I was brought in to lead a redesign of the Help Center, aiming to reduce friction, improve satisfaction, and empower both users and internal stakeholders.

• 66% of users said they didn’t understand the information presented.
• 50% reported it took “too long” to locate a solution.
• 38% were dissatisfied with the answers—often defaulting to refunds.
• 12% of “order problems” reported were actually unrelated (payments, account).
• In 70% of cases, restaurants didn’t respond during “order preparation”—when help is most critical.
• CSAT dropped sharply as response time increased.

These issues pointed to a failure in clarity, contextual support, and proactive issue detection. Moreover, internal teams lacked visibility over what content was failing, what users were struggling with, and where gaps in coverage were.

I led the end-to-end design process, from research through implementation, coordinating with CX, product, ops, and engineering teams.

Discovery & Research
Because the Help Center was a shared service, I had to work with different teams to gather the data and consider the different perspectives. That combine with data from users, made me able to get a clear picture of the problem.
• Conducted 8 stakeholder interviews across product, support, restaurant operations, and marketing.
• Held 5 qualitative user sessions to understand real help-seeking behavior.
• Benchmarking: analyzed 8 competitor help systems in food delivery / marketplace apps.
• Collected ~300 feedback responses from the existing Help Center.
• Mapped user journeys, pain points, and content gaps.

My strategy and solutions were based on the data and insights gathered during the discovery phase.

1. Contextual Help Experience
• Depending on order status (“in progress,” “delivered,” etc.), we surfaced likely issues ahead of users having to search.
• Dynamically prioritized common issues first, reducing exploratory load.
2. Search + Voice Assistance for Accessibility
• A search bar enhanced with voice input, enabling visually impaired or low-literacy users to navigate support.
• More visual cues, simpler language, and microcopy to reduce cognitive load.
3. Feedback Loop in Content Flow
• “Still searching?” prompt when users couldn’t find relevant articles.
• Users could flag issues as “unclear” or “missing,” which then fed into a pipeline for content teams to iterate.
4. Restaurant-Facing Interface Enhancements
• Pre-built response templates for restaurants to quickly address issues with customers.
• Improved chat queue UI with clearer organization of open conversations and statuses.
5.Proactive & Reactive Notifications
• Automatically detected anomalies (e.g. delivery delays, missing items) and alerted users preemptively.
• After resolution, sent a check-in to confirm satisfaction.
6.Behavioral Design: Credits > Refunds
• When a solution was offered, the interface emphasized iFood credits over refunds—lower cost to the business and greater chance of retained spend.

These are the metrics that will be tracked and the goals we set for the project:
• CSAT increased from 62 → 80.
• Average resolution time drop by 35%.
• Support ticket volume drop by 30%, lowering operational load.
• Restaurant engagement in chat increase by 12%.
• Rate of users choosing credits over refunds increase by 30%.
• Reorder rate from credit users increase by 20%.

The metrics will be watched but even before that these already are the plans in mind to improve the feature:
• Expand Help Center to omnichannel support (WhatsApp, chatbot, voice assistant).
• Increase restaurant dashboard sophistication: analytics, AI-suggested responses, sentiment indicators.
• Add “similar order replacement” workflows to resolve missing-item cases faster.
• Track long-term metrics: retention of credit users, support deflection over quarters, content decay rates.