Elevating CSAT Survey Experience to Improve Agent Performance
Qiscus Helpdesk is an omnichannel ticketing system that empowers customer support teams to manage, prioritize, and resolve inquiries at scale. As enterprise adoption expanded, measuring agent performance with actionable, real-time feedback became a vital requirement. I redesigned the end-to-end survey builder experience to balance deep customization for admins with frictionless completion for end customers.
The Starting Point & Challenge
Bridging the gap between agent coaching and customer feedback
Qiscus Helpdesk is used by high-velocity customer operations teams across Southeast Asia. As ticket volumes grew, support leaders needed a reliable mechanism to evaluate individual agent resolution quality directly from customers.
The core challenge was two-fold: Enterprise admins required granular configuration (branding, question formats, trigger rules), while end customers required a survey so swift and intuitive that response completion rates remained high without survey fatigue.
Discovery & Competitive Benchmarking
Learning from global standards without overwhelming local teams
To establish baseline requirements, I conducted a deep comparative analysis across leading global helpdesk platforms (Zendesk, Freshdesk) and local regional solutions (Mekari Qontak).
Global tools offer profound flexibility but often introduce steep configuration curves that slow initial onboarding. Conversely, local tools were fast but lacked enterprise brand identity control. Our opportunity was to combine enterprise-level customizability with a guided, linear setup flow.
- Zendesk & Freshdesk: High configuration depth, but required multi-page settings menus that increased admin cognitive load.
- Mekari Qontak: Fast single-screen setup, but lacked live brand previews and custom rating scales.
- Key Takeaway: Build a 4-step focused wizard with live real-time preview to eliminate setup guesswork.
Structuring Intentions: 4-Step Builder Flow
Separating complex tasks into distinct, bite-sized stages
To minimize cognitive load, I structured the creation journey into four focused stages. Each step serves a single purpose and validates against usability heuristics before allowing progress:
Identity & General Info
Define survey name, target language, and brief operational description with character counters and input boundaries.
Survey Content & Questions
Configure core rating logic (5-star, CSAT scale, emojis) and follow-up conditional feedback questions.
Brand Customization & Live Preview
Upload client logo, adjust primary button colors and card styles with an instant side-by-side interactive preview.
Automation Triggers & Dispatch
Set ticket resolution trigger criteria and personalize email notification templates with pre-filled dynamic variables.
Brand Customization & End-User Experience
When a survey feels familiar, response rates rise
Every enterprise client has strict brand guidelines. The live preview editor enables support managers to match company color tokens, custom logos, and greeting copy seamlessly.
For end-users, the submission interface is stripped of all redundant UI elements. It renders instantly on both mobile browsers and in-chat webviews, making submission a 1-tap interaction.
Design Detail: Micro-Interactions
Selected rating states trigger subtle haptic-like spring transitions and instant score confirmations, giving respondents visual clarity that their feedback was registered.
Reflections & Learnings
- Proactive exploration before PRD finalization: Initiating wireframes early allowed our team to identify edge cases in trigger automation before engineering started.
- Clear Figma handoff system: Documented states (loading, empty, error, active) and token annotations reduced frontend iteration cycles significantly.
- Cross-functional Design QA: Participating actively in pre-release staging testing ensured 100% visual fidelity and zero critical UI regressions at launch.
Future Opportunities
- Integrate AI sentiment summarization for open-text CSAT responses.
- Add automated Slack / Discord webhook alerts for negative customer ratings.