What the paper studied
The authors tested how two design qualities of AI travel chatbots — **perceived usefulness** (does the bot help me plan and book?) and **service convenience** (is it fast, accessible, low-effort?) — affect **customer loyalty** to an online travel agency (OTA), and whether **customer experience** mediates that link. They surveyed 200 adult OTA users in DKI Jakarta who had used an AI chatbot in the past six months and analyzed the data with PLS-SEM (SmartPLS 3) using a Stimulus–Organism–Response framework.
Key findings
- **Both stimuli matter directly.** Perceived usefulness (β = 0.202, p = 0.001) and service convenience (β = 0.192, p < 0.001) both have direct positive effects on customer loyalty.
- **Customer experience is the dominant driver of loyalty.** Customer experience has the largest direct effect on loyalty (β = 0.488, t = 8.145, p < 0.001) — bigger than either stimulus on its own.
- **Experience mediates the path from chatbot quality to loyalty.** Indirect effects: PU → CE → CL (β = 0.199, p < 0.001) and SC → CE → CL (β = 0.210, p < 0.001). Functional chatbot quality builds loyalty only after it produces a good experience.
- **Service convenience is the stronger experience-driver.** SC → CE (β = 0.430) outperforms PU → CE (β = 0.408).
- **Model power.** The model explains ~46% of variance in customer experience and ~55% in customer loyalty.
- **Sample skews mobile and active.** Users were 50/50 male/female, mostly age 26–41, mostly with bachelor's degrees, with ~88% having used the chatbot more than once in six months. Main use cases: booking tickets/hotels (31%), tour package info (22.5%), travel recommendations (18%).
Why it matters for hospitality
The Indonesian OTA market is huge and switch-heavy — 71.4% of Indonesians have used an OTA, but customers switch platforms easily. The lesson generalizes: in any commoditized travel marketplace, a chatbot that is technically capable but feels generic will not retain customers. What converts utility into retention is the experience layer — personalization, accessibility, frictionless interaction, and the feeling that the bot "gets" you. Hospitality brands building chatbots should not stop at "it answers the question" — they need to instrument and design for the experience that flows from the answer.
Practical takeaways
- Treat customer experience as the **primary KPI** for travel chatbots, not just task completion or response time.
- Invest in **service convenience** (speed, multilingual support, 24/7 access, simple UI, low effort) as the strongest lever for experience and loyalty.
- Make **perceived usefulness** concrete: relevant recommendations, accurate availability, integrated booking, and clear next steps — not just a wall of text.
- Build a **human escalation** path. The authors flag that chatbots must integrate with human service when they can't solve complex problems; this protects experience quality and trust.
- Continuously measure: customer feedback, complaint history, time-to-resolution, post-interaction satisfaction. Don't assume deployment = success.
- Personalize using known guest context (loyalty tier, prior bookings, preferences) — generic chatbot experiences don't drive loyalty.
- For Indonesia and similar high-switching markets, frictionless cross-channel handoffs (chat → booking → confirmation → in-stay) are a defensive moat against competitor OTAs.