During my last 6 years at Synthetix, was one of my main responsibilities to research, design and prototype solutions to improve the UX of automated digital communication platforms and CX products such as Intelligent BOTs, Live Chats and Web Self-Service FAQs tools.
As a team we had the opportunity to deliver a number of bespoke solutions to help enhance engagement and improve the online experience for a wide range of high-profile clients: well-known brands such as Boots, Virgin Trains, Toyota, YBS, Esure, Fortnum & Mason, National Express and more.
For this particular project the company goal was to find new ways to offer a scalable, highly engaging and more effective CX solutions. A platform capable of guiding user choices just enough to achieve better conversion/customer satisfaction and by possibly reducing those margins for mistakes and doubts that generally make users think!
For this project I was the only product designer and researcher. I decided for a lean approach and collaborated with stakeholders, product managers and developers during focus groups and planning sessions to ensure outcomes were agreed and all departments involved understood our aims and directions.
Our experience within the CX sector, extended field researches and competitor analysis have enabled us to identify issues and frictions within our line of products and investigate new solutions to create more efficient and scalable experiences with the implementation of AI technologies.
By formulating hypothesis and looking into validating them throughout interviews and prototype testing, we came to the conclusion that our set of tools could be better integrated and unified under one single intelligent product. A solution capable of offering a better experience by guiding the users all the way during their journey while also enabling escalation to different communication platforms whenever needed.
In order to be able to test our hypothesis we decided to adopt a specific industry for the product journey. We choose the insurance industry for this specific case study. In fact, having had the opportunity in the past to work closely with Esure one of our clients, our knowledge about their industry products and clients' needs became very handy for this project.
The main Persona we decided to focus on and empathise with for this project was in fact a direct result of these researches.
Based on our insights, I've designed a default user journey which we thought it could be then customised and enriched by our clients with more conversational interactions in order to make the experience more personal and unique. Our main hypothesis, in order to achieve our outcomes, were the following:
We decided to apply the IA of the insurance industry as for the content of this MVP prototype. By focusing at first on one specific journey, we figured out the content we needed for our prototype and extended the architecture further as we needed for testing iteration after iteration.
This is where I started sketching and wireframing our product - mobile first. The idea was to make sure the interface didn't look like a chat, but still maintained the conversational feeling of a chat. This was necessary in order for the BOT not to be confuse with the actual Live Chat tool integrated within the same platform amongst all the other available escalation options.
Also, because users needed to be guided throughout the journey, but at the same time have the ability of typing questions, or starting a chat whenever needed, I opted for a burger menu with options at the bottom right hand side of the screen (close to the user thumb) which could be easily accessed at all time during the journey.
After sketching wireframes and having a clearer idea of the user flow, I started creating a digital prototype using Adobe XD. My priority at this stage was to make sure that the user was able to easily follow the flow of the BOT and potentially also find his way back to previous steps of the journey in order to change his path whenever needed.
That would have allowed in our opinion the user to be 100% in control of the journey, without though falling outside the BOT "hand-holding" process.
I was at this point ready to test my first prototype. 5 users were observed while completing specific tasks, and all sessions were recorded. I tested this early prototype in house with staff members from outside the dev and design teams.
Results were encouraging, but not perfect of course:
The main issue reported was that users would not understand how to "go back" in the journey. The action wasn't clear enough.
Based on the results of my first test, I have amended the design where needed. I also felt confident that the High-fidelity prototype would have helped tackle some of the previously recorded issues. The use of brighter colours, different iconography and branding styles gave the product a visually clearer and more appealing look and feel.
In order to make the "Go back" action even clearer, I also decided to add an arrow down icon next to the previous choice in order to help the user understand that by clicking, or even dragging down from the top, they could go back in the journey.
The prototype went through a couple rounds more of testing iterations. The results showed massive improvements each time. The product was very well received by the test users in terms of ease of use, clarity and look and feel. The issues recorded also from previous tests were at this point successfully overcame and the overall experience was validated - including integrated escalation options to FAQs Search and Live Chat tools.
I tested this last stage of the prototype on 10 more people, in-house and outside the company - guerrilla testing type.
Because within the team we couldn't 100% agree on it, we decided to run an A/B testing around some of the Live Chat functionalities. This was extremely helpful for us in order to clarify small doubts we had regarding the chat interface on matters like where to place the "End chat" button and how to best allow the user to move in and out of the chat room during a live chat session.
Based on test results, option B was far more successful than A:
This Intelligent BOT prototype was very well received by everyone in the team and also from some of the clients we have tested it with. The MVP prototype, which was tested ultimately on 30 people in total and designed within the 4 weeks deadline, is now destinated to become a product and it's currently being developed and while further tested as we speak.