TL;DR
This study investigates how Bed & Breakfast (B&B) hosts in the sharing economy respond to negative online reviews. It finds that the way a response is phrased (its "voice") is more impactful than its length, with empathetic and accommodative responses being most effective. Interestingly, response length follows an inverted-U curve, meaning too short or too long responses are less helpful.
What Problem Does This Paper Address?
The sharing economy (SE), particularly platforms like Airbnb, relies heavily on user-generated content, especially online reviews, to build trust and facilitate transactions. Negative reviews can significantly damage a business's reputation. However, many SE service providers, like B&B hosts, are informal and may lack the skills to effectively manage online complaints. This paper addresses the gap in understanding how these informal providers can best respond to negative reviews to positively influence potential customers' perceptions of helpfulness and, by extension, their booking decisions.
Methods
The researchers employed a multi-study approach:
- Study 1: Categorizing Response Voices: They conducted a content analysis of B&B responses to negative reviews. Using existing frameworks and their own coding, they identified three main "voices":
- Defensive: Denying failure, making excuses, or blaming the customer.
- Formalistic: Perfunctory apologies or automatic responses without genuine engagement.
- Accommodative: Sincere apologies, confessions, promises of corrective action, or compensation, often with empathic statements.
- Study 2: Examining Response Impact: They used regression analysis to determine the relationship between response quality (length and voice) and perceived helpfulness votes on reviews. They also explored interaction effects between response characteristics and review characteristics (length, presence of images).
- Study 3: Cognitive Load Theory (CLT) Analysis: To further investigate the role of response length, they applied CLT and tested for an inverted-U curve relationship between response length and perceived helpfulness.
Datasets
The study utilized a dataset of 3696 reviews posted on Ctrip.com (a Chinese travel platform) for 24 privately owned B&Bs. Of these, 2716 reviews had responses. The data was collected before October 2017. The researchers filtered this dataset to focus on negative reviews that received a response within 24 hours, resulting in a final sample of 707 reviews for analysis.
Key Insights and Ideas
- Empathy is King: The tone and intent of a response (its "voice") are far more critical than simply how long it is. Accommodative responses, which show genuine concern and a willingness to resolve issues, are perceived as most helpful.
- The "Just Right" Length: Response length isn't a simple linear relationship with helpfulness. Too short, and it might seem dismissive; too long, and it can overwhelm the reader (cognitive overload). An optimal length exists, suggesting a sweet spot for providing sufficient detail without being excessive.
- Context Matters: The effectiveness of a response can be influenced by the original review. For instance, longer reviews or those with images might benefit more from detailed, empathetic responses.
- Informal Providers Need Support: The study highlights that SE providers, often lacking formal training, need guidance on effective online reputation management.
Key Results
- Response Voice: An accommodative response voice was found to have a significant positive relationship with perceived helpfulness votes. Defensive and formalistic voices did not show this positive effect.
- Response Length: While response length had a positive relationship with helpfulness votes (H1 supported), it was less impactful than response voice. Further analysis revealed an inverted-U curve relationship (H5 supported), meaning moderate-length responses were most helpful.
- Interaction Effects: The relationship between response length and perceived helpfulness was stronger for longer reviews and reviews with images (H3a and H3b supported). However, the empathetic response voice was consistently effective across different review types (H4a and H4b not supported).
Limitations and Future Work
- Measuring Emotion: The study acknowledges the difficulty in precisely quantifying the exact emotions conveyed by response voices, suggesting future research could use sentiment analysis tools.
- Sample Size: The dataset, while substantial, was limited by the availability of negative reviews with responses on Ctrip.com for the chosen period, potentially impacting generalizability.
- Response Complexity: The study categorized responses into distinct voices, but future work could explore responses that might contain multiple voices.
- Dynamic Effects: The research focused on the immediate impact of responses on helpfulness votes. Future studies could investigate the longer-term effects on business performance metrics like sales and future review volume.
- Neuroscience and Behavioral Tools: The authors suggest exploring advanced tools like eye-tracking to gain deeper insights into how consumers process online reviews and responses.
Summary updated