To design a better messenger from the ground up, we set out to form a deep understanding of conversational interfaces, when and why customers turn to chat for support, and the business challenges teams face when using even the best consumer messengers out there.
To start, we surveyed more than a dozen chat interfaces to identify patterns in modern messaging-based support experiences.
Next, we wanted to understand why messaging-based support works. We surfaced four key reasons:
Customers avoid queues, while support teams get buffer time to form better responses and work together behind the scenes.
The ubiquity of chat interfaces makes messaging feel more familiar to customers than video, voice, or custom GUIs.
Conversational interfaces circumvent the need for verbose introductions, salutations, and signatures. Messaging adds flow, making support interactions feel more human.
Average response times, typing indicators, and read receipts build trust, add transparency, and develop the perception that exchanges unfold in real-time.
With a better understanding of the why, we then wanted to know when customers turn to chat.
Mapping the customer’s journey from moment of awareness to initiation of conversation revealed an insight: customers message in when they aren’t able to find answers on their own.
If customers were given a better way to find answers on their own, teams could scale their support by decreasing conversation volumes and provide more personal support to customers who need individualized help.
Although support products like Intercom and Zendesk offer great help center solutions, they exist as add-on products or separate services. We couldn’t find any messaging-based support solutions that fully integrated help desk suggestions into their messengers.
This idea framed the problem differently and shifted our focus from designing a better messenger towards designing a better way for customers to help themselves.
With a direction to place knowledge at the front of the experience, we started prototyping explorations on paper.
While ideating, we stripped the concept of the help desk down to its very core and oriented the entire experience around its most atomic elements: questions and answers.
We decided to suggest answers contextually – within the “messenger” – to help customers find answers fast and on their own:
Depending on where chat is invoked during browse, we can try to predict why the customer might be looking for help by showing popular answers that helped other customers in similar contexts.
By showing popular topics on focus, we reduce cognitive load, declutter the interface, and provide helpful shortcuts to trending topics that are helping other customers.
Autocomplete makes searching feel faster and helps guide customer searches. Suggestions that customers find helpful can be remembered to improve predictions over time.
In the event a customer continues to browse with the messenger open, an indicator can appear to load new suggestions as they become available.
We also wanted to provide a way to let customers message in when they need personalized help. But we didn't want to badge the hamburger or bury chat too deep. The solution had to be flexible and easy to access.
We came up with a toggle that appears only when the messenger is expanded.
We made the toggle stateful to add transparency and make interactions feel more instant.
We added cards to let teams initiate conversations when the messenger is collapsed.
We added reactions to articles to give businesses a way to measure what's helpful. They also let teams know which answers need work.
Learning what customers search for can give teams ideas on what to write next. Over time, search patterns can reshape existing topics and intelligently form new ones.
We also worked out a lot of details behind the scenes to make messaging as useful for teams.
With four months to public launch, our work on messaging validated our ability to create value and framed Kustomer's vision in a way that helped Kustomer raise a $10M Series A to bring its messaging platform to life.