The first stage of research is to source, collect and filter relevant data. Based on defined objectives, Isentia crafted an in-depth listening framework with an extensive keyword iteration list. Using our listening tool we collected data in two phases:
Customised crawling of specific websites
We built API crawlers to pull data across thousands of channels across major platforms such as Facebook, Blogs, Forums, Instagram, Twitter, Websites and Online News. This is an essential component of the research process as it helps ensures that the data is collected across all travel-relevant platforms, making the study more relevant and targeted for Airbnb.
Validation of data
Isentia used (a) automated Natural Language Processing and (b) Human Validation of sample data to achieve a 95 per cent confidence of data accuracy. This is crucial in data hygiene as the automated sentiment analysis utilised by most social listening tools cannot, detect or recognise local slangs, sarcasm, various grammatical nuances and culture variations. Hence, the accuracy level at which social listening tools are able to accurately pinpoint or determine the prevailing sentiment of a discussed topic is probably likely to be only at 30% – 50%. With human validation of a sample size, Isentia helps ensure a higher accuracy of data sampling for the study.
For this research, Isentia leveraged social and digital data from 1 April 2015 to 30 June 2015, which are seasonally high travel peaks and hence, when travel-related conversations will likely to be most prominent. From there, key metrics of analysis were derived at a regional scale for the identified target markets.