The NSW FuelCheck app is a helpful tool for viewing daily fuel prices at petrol stations across NSW, with more than 2400 service stations in NSW uploading fuel prices in real time. This free-to-use app allows the public to view petrol prices at service stations on a map view or can be personalised to provide alerts for favourite stations or to help locate the cheapest fuel on a route.
The NSW Government is continually looking for ways to improve its products. To deliver a great customer experience the NSW Government receives feedback directly from customers and uses this feedback to identify priority areas for improvements.
Across the last five years about 11,000 comments have been collected as customer feedback on the FuelCheck app. To help assess this high volume of valuable feedback, it is important to identify and prioritise recurring themes.
The NSW Data Analytics Centre (DAC) was tasked with analysing the comments about the FuelCheck app to identify some of the themes. To make this process easier, the team used Natural Language Processing (NLP) to identify the most frequent and recurring feedback themes and to help the FuelCheck team prioritise improvements to the app for the best user experience.
Natural Language Processing
NLP is where computational techniques are applied to analyse natural speech patterns which turns language into a type of data. There are a few processes involved in NLP, such as identifying key words in sentences and grouping words into themes. Comments can still be viewed individually but grouping helps to speed up insights for large volumes of comments.
Another process used in NLP is to remove the tenses from words and identify the root word. For example “smiling” and “smiled” would both become “smile”, so that ‘smile’ is recorded twice. This shows how frequently themes are mentioned and captures more effective insights than identifying words on their own.
The DAC used NLP to identify issues with NSW FuelCheck and were able to provide a summary list of what themes customers were asking for. Some key findings included:
• Incorporating missing service stations
• Resolving inaccurate prices
• Allowing for multiple fuel types in one search
• Adding more analytics and price monitoring trends
• Adding an option to add vouchers/discounts, which would be reflected in results
This feedback analysis will help the NSW FuelCheck team prioritise new improvements to the application, to provide the best customer experience and continue to ensure customers’ experiences remains central to the design.
FuelCheck data is available on the NSW Open Data Portal here: https://data.nsw.gov.au/search/dataset/ds-nsw-ckan-a97a46fc-2bdd-4b90-ac7f-0cb1e8d7ac3b/details?q=fuel%20check