From Tokyo to Fukuoka, Finding Answers in the Field: Our CS InquiryChat Implementation Journey (opens in new tab)

MessagingHub replaced Demaecan’s third-party customer-service chat with its own InquiryChat platform. The transition eliminated annual licensing costs, improved operational flexibility, and reduced repeat customer inquiries by about 20%. The project’s success depended not only on rebuilding features, but also on observing call-center work directly, aligning distributed stakeholders, and balancing agent convenience with user privacy.

Why Demaecan Chose an In-House Platform

• Demaecan’s existing chat provider was being discontinued, prompting a choice between upgrading to another vendor solution or adopting InquiryChat.
• InquiryChat was selected because it offered:
  • Zero licensing costs
  • Flexible customization for Demaecan’s processes
  • Direct internal technical and operational support
  • Secure access to customer information
  • Real-time analytics and reporting
• The migration had to preserve service continuity while changing agents’ workflows and tools.

Problems with the Existing Service

• Agents could preview messages before users sent them, creating a transparency and privacy concern.
• Chat sessions were not preserved when users left, forcing customers to repeat their inquiries.
• Vendor customization and integration options were limited.
• Some existing features appeared useful on paper but were not actually used in call-center operations.
• Because agents were accustomed to the existing system, they needed evidence that the replacement would improve their daily work—not merely solve problems identified by headquarters.

Understanding Real Call-Center Needs

• Requirements initially arrived indirectly through multiple departments, making their operational context difficult to understand.
• The team reorganized requirements into:
  • Standard platform features
  • Features requiring customization or review
  • Features requiring new development
• PM Kim Seri visited two call centers in Fukuoka, observed peak-hour operations, and interviewed agents and managers.
• Field research revealed:
  • A manager-support chat feature was unnecessary because agents used hand signals instead.
  • Audio alerts were ineffective because office environments kept sound muted.
  • Integration with the existing CS system was essential because agents had to summarize and record every interaction after a chat.
• These observations helped the team remove unnecessary features while raising the priority of workflows that were genuinely critical.

Managing a Distributed Migration

• The project involved teams in Korea and Japan, two Fukuoka call centers, external operators, and multiple internal departments.
• To create a shared operating model, the PM:
  • Built a Jira dashboard to visualize progress and support data-driven decisions.
  • Created a master specification as a single source of truth.
  • Led internal product and development QA to ensure the original intent was implemented correctly.
  • Produced detailed operational guides for launch and adoption.
• These processes reduced ambiguity across the project and supported a smoother transition from development to live operations.

Focus Group Testing and Interviews

• A focus group test (FGT) brought all stakeholders together to act as customers and CS agents.
• Participants followed complete scenarios, from opening an inquiry through resolution.
• The FGT exposed user-experience issues before formal QA, including:
  • Unclear role and status indicators
  • Android push-notification instability
  • Keyboard and input-field overlap
  • Missing timestamps on links
  • Push-title wording issues
  • Chat logs not consistently reaching the CS system
• Several issues were fixed immediately, while clearer role and status displays were scheduled for further improvement.
• Follow-up focus group interviews found the system generally stable and easy to operate, though some participants initially needed time to understand operator and manager roles.

Balancing Agent Convenience with Privacy

• Agents strongly valued the old “message being typed” preview because it helped them anticipate replies and reduce average handling time (AHT).
• From a platform perspective, previewing unsent or deleted text was a serious information-ownership and privacy risk.
• Rather than simply removing the feature, the team introduced a typing-status indicator.
• This preserved agents’ awareness of the customer’s response rhythm without exposing the customer’s actual unfinished text.

The project demonstrates that successful system internalization requires more than feature parity. Direct observation, structured stakeholder coordination, realistic user testing, and privacy-conscious design enabled Demaecan to reduce costs while improving both operational efficiency and customer experience.