FAQ CHATBOT WITH
A LOCAL MESSAGING APP
PROJECT SUMMARY
A customized chatbot designed to assist small, less tech-savvy lodging owners with common support questions, reduce manual calls, and improve partner retention
As a Business Development Manager in Expedia Korea, I was responsible for adding new lodging facilities to the platform. By 2019, all major hotels were already listed, and the local growth opportunity came from smaller accommodations such as motels, guesthouses, and family-run vacation rentals usually operated by less tech savvy owners in their 60s-70s.
​
While competitor OTAs supported these partners with high-touch phone-based onboarding, Expedia required partners to self-serve via a global portal and email-based support—an unfamiliar and difficult process for many operators in Korea. As a result, although the team signed more than 100 new facilities per BDM each quarter, churn was high as partners quickly dropped off when they struggled with operations.
​
To make the business sustainable, improve retention, and free BDMs from time-consuming calls, I led a project to create a chatbot on Kakaotalk (the Korean equivalent of WhatsApp). The chatbot automated responses to the most common partner questions, reducing support burden and enabling BDMs to focus on signing new properties.
TIMELINE
Jun - Aug 2019
TOOLS
Google Workspace, Kakaotalk Chatbot Builder
TEAM
Korea Lodging Business Development Team (myself + 3 BDM colleagues)
MY ROLE
Project Lead – user research, concept development, validation, internal communication
WHAT THIS PROJECT DEMONSTRATES
-
Local market adaptation – recognizing differences between Korean small-business owners and U.S. partners and designing a culturally relevant solution.
-
Operational efficiency – automating routine inquiries to free up limited BDM resources.
-
Innovation under constraints – creating a scalable tool without product/tech team involvement.
-
Influence and leadership – initiating a new project in a newly created team, gaining director approval, and coordinating peers.
PROBLEM CONTEXT
Expedia Korea’s partner base was limited compared to local competitors because only hotels were listed. Competitors offered broader coverage across motels, guesthouses, and vacation rentals, which appealed to domestic travelers.
When these smaller facilities were added to Expedia:
-
Barriers emerged: Owners were unfamiliar with Expedia’s self-service system, reliant on email with global reps, and unused to managing bookings online.
-
Behavioral mismatch: Most owners were 60+ and accustomed to local OTAs that provided live phone support.
-
Outcome: BDMs received constant operational calls from partners who had been signed but could not manage independently. When support was unavailable, many partners shut down their listings, driving high churn.
The result was high signing rates but unsustainable retention and efficiency. Each BDM managed fewer than 40 new partner contacts per week, far below targets.
TIMELINE
A phased rollout starting with logging and categorizing partner questions, designing and testing chatbot flows, launching to new partners with escalation support, and achieving full team adoption with impact monitoring.
WEEKS 5 - 6: Designed and tested chatbot conversation flows in Kakaotalk
WEEK 9-12: Full adoption across the Korea team, monitored impact on support calls and BDM workload
WEEKS 1 - 4: Logged and categorized inbound questions, drafted automated scripts
WEEKS 7 - 8: Rolled out chatbot to new partners; directed unresolved queries to Expedia’s partner support call center
CHALLENGES
-
Limited resources: No engineering support—solution built entirely with local tools (Kakaotalk chatbot builder, Google Workspace)
-
Change management: Partners accustomed to direct calls resisted chatbot use at first; we positioned it as faster and available 24/7
-
Team buy-in: Momentum secured after director approval; all BDMs contributed to data collection and rollout.
RESEARCH & VALIDATION
To identify patterns in partner challenges, I led the team in tracking every inbound support call over four weeks.
-
Data collected: 200 calls → 98 unique questions.
-
Analysis: Classified into 5 categories (e.g., password resets, booking cancellations, room type setup, rate changes) with subcategories of questions
-
Validation: The same themes were repeated across all four BDMs, confirming consistency of issues.
This gave us the foundation to design automated FAQ flows that could resolve ~70% of routine inquiries.
RESEARCH & VALIDATION
Through daily partner calls, we identified that many newly signed small accommodations were struggling with Expedia’s self-service tools.
To validate and structure these pain points, the Korea BDM team systematically collected and analyzed real inbound questions.
​
-
4-week data capture: Logged every call from new partners into a shared Google Sheet, creating a list of questions from 200+ calls with multiple questions entailed.
-
Pattern finding: Grouped questions by recurring topics such as listing setup, pricing, inventory, reservations, payments, and support.
-
Expansion & refinement: Continued collecting until we had 300+ entries, then consolidated them into 12 parent categories (with 4 subtopics).
​
Why This Validation Was Critical
-
Provided data-driven evidence that operational inefficiency—not just partner inexperience—was causing inefficiency.
-
Identified top recurring questions (~80% of calls) that could be automated, proving the business case for a chatbot.
-
Created a structured knowledge base that guided chatbot conversation flow design and enabled scalable self-service.
-
Helped leadership approve the project by showing time saved for BDMs and the potential to reduce repetitive support calls.
*A list of questions organized into 12 categories with 4 subcategory questions for each.
* chatbot images recreated
DELIVERABLES
-
Chatbot Q&A library – a list of questions classified into 12 categories.
-
Automated Kakaotalk flow – provided self-service responses to FAQs.
-
Escalation system – routed unresolved issues to Expedia partner support.
KEY OUTCOMES
-
Before the project: Each BDM averaged 100 partner calls per week, with about 30% (≈30 calls) spent answering basic “how to use Expedia” questions that should have been self-served or handled by the support center.
-
Week 9 (chatbot launch): Support calls stayed similar at 32 per week.
-
Week 10: Dropped to 18 calls.
-
Weeks 11–12: Fell to fewer than 10 calls per week.
​​
This sharp decline verified that the chatbot successfully reduced repetitive support inquiries, freed BDM time for signing new partners, and made onboarding more efficient.
70% Reduction
LESSONS LEARNED
Cross-Regional & Local Adaptation
This project highlighted the importance of tailoring solutions to local partner realities. Unlike the U.S., where small businesses could self-serve online, Korean partners required more guidance — making the chatbot a culturally relevant bridge.
​
Balancing Growth with Sustainability
While signing 100+ facilities per quarter met growth goals, the team was spending a significant share of time on support calls that BDMs were not meant to handle. Ignoring these questions risked partner dissatisfaction and contract churn, yet because this area was not our primary focus, answers were often inconsistent or lacked depth. The FAQ classification and chatbot made responses structured, accurate, and repeatable, reducing churn risk while freeing BDMs to focus on growth.
​
Monitoring & Knowledge Sharing
Although the impact was clear from the call reduction trend, more detailed tracking of retention and support usage would have strengthened the business case. Sharing these results and insights earlier with other APAC teams could have invited feedback, accelerated refinement, and supported broader regional adoption.
​
Future Opportunities
-
Track churn and long-term retention impact to quantify success more precisely.
-
Supplement the chatbot with step-by-step onboarding guides or video tutorials.
-
Explore adoption in other APAC markets facing similar challenges.
Overall, the chatbot project demonstrated the ability to identify operational inefficiencies, design localized solutions, and improve partner sustainability, while showing the value of structured answers and deeper monitoring for regional scalability.