Feedback in the U.S.er S.uR.vey
How we have used modern technology to get user feedback on one of our oldest flood services.
The Floodline interactive voice response (IVR) service (accessed via 0345 988 1188) has been around for over 15 years. It enables users to listen to flood warning advice issued by the Environment Agency (EA) and Natural Resources Wales (NRW) during flood events.
The Floodine IVR received more than 13,000 calls over a one month period from the middle of December 2020 to January 2021 which for a more traditional channel, is a well used service. Until now however, we simply haven't been able to know what our users really think of the service.
Digital Services in the Environment Agency have been addressing this challenge by exploring the options for adding a user survey feedback component into our current service. After assessing our options we decided to design and develop this feedback component in-house using Twilio. This approach had many benefits:
- We can keep costs down for the EA, NRW and the taxpayer
- We can bring accessible and available tools into our team domain, increasing skills and knowledge with our colleagues
- We can be agile. Changes to the module can be made easily and quickly with very little overhead in cost and effort. This is doubly important as we intend to tailor our feedback questions to hone in on areas for improvement
- We can provide a bespoke service to our colleagues in Natural Resources Wales by offering simultaneous feedback flows to users in Wales, both in the English and Welsh languages
- Using the Amazon Polly text to speech engine, the audio prompts provide a seamless experience whether listening to flood warning information or answering service feedback questions
Hello (and) Goodbye
So how do we get feedback without negatively impacting the user journey from start to finish?
Designing the feedback module to fit in with our existing flood warning service was a challenge. Firstly, we had the complexity of having two existing and distinct IVR’s; one that plays our flood warning messages from our flood warning system, and one that serves the Floodline service and partner organisations around the country.
Users still need to access our service in all the ways they currently do; hear the flood information that is right for them; be able to leave feedback, and be routed quickly between these components logically with the minimum amount of fuss.
The main obstacle to our proposed solution was passing calls between the (now 3) IVR systems so that the caller is sent to the intended destination with a minimum number of user interactions. This was done by using an API that allowed for specific and seamless routing and thereby removing the often frustrating menu selections.
Feedback questions were mocked up and a content designer was brought on board to refine the statements to make the user journey as clear and consistent as possible. We designed the new service to ensure there are no “dead-ends” for the users, and made sure that they could repeat call prompts in case they missed them the first time. This increased the complexity of the design but ultimately made it clearer for our users.
Users can now leave a recorded message with their feedback on our service. This feedback is automatically transcribed using IBM’s Watson Speech to Text - a AI-powered transcription service. We tested a number of transcription services, but Watson gave us the best results, and also has the ability to learn and improve over time.
Finally, to allow all of the data to easily be accessed and shared we created a number of Microsoft Power Automate flows to automatically collect and collate the feedback so we can report the findings with colleagues.
Finding ways to meet the evolving needs of users using traditional channels can all too easily be overlooked. Using new technology, we can now better understand what our users need from our IVR service.
The coming months will see us sift through the responses and work on those areas most in need of improvement. But that’s a good thing, and now we know what needs fixing, we can work it out.
