Inês Duvergé
My experience revamping the Inbox at Intercom

Live website

At Intercom, I was responsible for the enhancement of our Inbox – a vital tool for millions of customer support agents.
Over the past two years, I've revamped the Inbox experience and introduced cutting-edge AI functionalities to revolutionise customer support. This page offers a brief summary of my key contributions during this time.
Role
Product designer responsible for research analysis, design and delivery of final product
Timeline
2 years
Type
Research, UX/UI,
Usability testing

A summary of two intense, but gratifying years

While I write a more in-depth case study, I’m sharing here the highlights of what I’ve been up to these past two years. More than just a list, these insights and learnings provide a glimpse into my accomplishments and experiences – offering you a sneak peek into my world ;)

Project 1

Next-gen Inbox 🚀

How can we create a more efficient Inbox that serves the needs of our teammates?
Project highlights:
  • Keyboard-first system creation
  • Dark-mode theme
  • Updated design system library
  • Dealing with migrations of million users  and iterating on feedback

Project 2

The introduction of AI 🤖

Can AI help teammates respond to messages and solve queries quicker?
Project highlights:
  • Working closely with ML engineers
  • Creating new AI patterns
  • Deep collaboration across teams in a fast-paced environment
  • Keeping up with a quickly growing field of tech

Project 3

Search improvements 🔎

Can we improve the core search experience?
Project highlights:
  • Combining data and customer feedback in a deeper level
  • Thinking big to then be able to start small with careful bets
  • Focused prioritization

1. Next-gen Inbox

The Inbox is the go-to tool for customer support agents, vital for responding promptly to customer queries. With agents often judged on response times, it's crucial that our Inbox is both fast and beloved by support reps.

However, over time, it became bloated with features, slow, and outdated. It was clear that a focus on speed and user satisfaction was essential.

My role

When I took over as the main designer, the research phase and project vision were already underway. I stepped in, refined the research findings, iterated and led the successful migration of all users to the new Inbox.  The role of a designer at Intercom is closely intertwined with that of the PM, necessitating close collaboration. Additionally, I've worked alongside other designers and engineers to make informed decisions and maintain a fast-paced delivery momentum. Finally, I guided designers on other teams, ensuring they adopted the new principles and addressed their queries.

Designed for efficiency and delight

In addition to addressing core issues, the project involved implementing several key features: the creation of a command palette, prioritizing keyboard-first interactions, updating the visual design system, and introducing a dark mode.
Creating a dark mode theme
Keyboard-first
New UI and improved layouts

Onboarding by doing

We couldn't overlook the onboarding experience for our everyday Inbox users, which required careful attention. I designed an onboarding experience that embraced a keyboard-first approach, allowing users to familiarize themselves with the new features as they interacted with the updates.



Overall, the new Inbox significantly improved efficiency for customer support, especially for keyboard-first users. This translated to a time saving of up to 2 hours per week, a significant win in a time-sensitive industry.

An updated design system and new principles

After the launch, we didn't stop there. We iterated, refined, and updated our design system. I also shared insights with a large group of designers on the principles of the new Inbox to prevent past mistakes. Additionally, I created a set of guiding principles for other teams to develop features on top of the new Inbox.


2. The introduction of AI

At the onset of the AI industry boom, Intercom made the strategic decision to invest in experimenting with integrating AI heavily into our experiences whenever it addressed a user problem.

My role

My team was tasked with establishing the vision for AI in the Inbox and pioneering the launch of our first experimental features. This involved extensive collaboration with the machine learning team, as well as defining the product vision and delivering experimental features that struck a balance between technical feasibility and user requirements.
Summarise a conversation
Help compose a reply
Auto-filling tickets

Invaluable learnings

The results primarily manifested in increased new customer acquisition and a positive adoption of most features. However, there were beta features that we had to discontinue as they didn't yield consistent results or justify the costs of AI implementation.


This project provided invaluable insights – from collaborating closely with machine learning engineers to navigating rapidly evolving technologies, understanding customer concerns, and managing technology costs and limitations. I look forward to sharing more about this experience soon.
Read full case study

3. Search improvements

Inbox search is one of the most highly used inbox features, with a majority of support agents using it an average of 25 times per day. However, we’ve learned that it’s not fit for purpose for core rep workflows.  Agents struggled to find answers to queries or locate specific conversations quickly, often resorting to using Slack search instead.

The revamped Inbox search was designed to help teammates find conversations faster with fewer clicks. Changes included refining search result relevance logic, highlighting matching keywords, improving conversation previews, and adding extra filters like status, custom date range, and ticket vs conversation.

My role

I led this project from problem definition to delivering the final solution. This included prototyping, testing, and collaborating with engineers on search relevancy. As a product manager wasn't available at the time, I primarily took ownership of the project.

Results

We received positive qualitative and quantitative feedback through a customer survey, with 83% indicating they no longer needed to resort to alternative searches (like Slack channels). The majority expressed greater satisfaction with the new version. However, 24% still reported dissatisfaction, aligning with the issues we chose not to prioritize at this time.
I’m proud of the Inbox's progress over two years and the personal growth it brought. Stay tuned for more detailed case studies on some of these projects soon.

Previous

Arrow left

Next