Redesigning Jira's onboarding to solve the retention gap for 7 million annual users.
The leaky bucket: each year over 7 million new users are added and become monthly active users in Jira. However, only 2.1 million (~30%) of them are retained after 6 months, and ~70% couldn't find sufficient value to stick with Jira.
Pain points &
opportunities areas
We put the burden of discovery on users, expecting them to find their project, work and Jira features they don't yet understand or even know exist.
Our current flows are optimised to invite a lot of new users quickly, but at the cost of clarity and meaningful context.
We ask users to invest time and share details before showing the value of Jira or how it helps them do better work.
The approach
Instead of leaving discovery to chance, we leveraged existing peer connections within the Teamwork Graph to intelligently bridge users into relevant projects, naturally introducing core Jira concepts through the work that already matters to them.
Focus 2: Intentional invite
We transformed the invitation process from a generic entry point into a data-rich handshake, empowering inviters to provide teammate context that allows Jira to dynamically tailor the onboarding journey and surface high-signal first actions.
Focus 3: Productivity boost
By identifying high-intent moments in Confluence, we engineered "Productivity Boosts" that convert static content into actionable Jira objects, providing users with an immediate, tangible win by showing them exactly where their work fits into the broader ecosystem.
Experience principles
Experiences must feel intuitively familiar while remaining intelligently tailored, leveraging existing team data to ensure every user enters a workspace built specifically for their role.
Prioritise learning by doing
Onboarding should favor "learning by doing" over static tutorials, using contextual and timeline-based guidance to help users achieve immediate, high-value outcomes.
To prevent cognitive move, we must prioritize the "bigger picture," orienting users within the team's structure and goals before surfacing granular tasks.
The journey must celebrate incremental progress and surface tangible value early, building a clear, rewarding path toward long-term platform mastery.
What's next
Our design sprints operate on a 3–4 week cadence, focused on iterating and exploring new features for the Jira onboarding experience. We are currently investigating peer-to-peer learning and tailored guidance to help new users master Jira's full potential. Stay tuned.
The approach
I architected the interaction model for treating ai agents as first-class collaborators, allowing users to seamlessly hand off work items so they can transition from manual execution to high-level strategic review.
Contextual agent invocation
by conceptualizing @mention patterns for agents within the commenting flow, i enabled a human-like collaboration model that allows teams to pull agentic intelligence into specific parts of a work item in real-time.
i integrated agentic capabilities directly into jira's workflow engine, designing the logic for agents to automatically trigger and execute complex tasks during status transitions to keep work moving in parallel.
Experience principles
i established a bespoke ai component library that signals agentic capabilities while maintaining a familiar jira design language to ensure a seamless, high-trust user experience.
to minimize cognitive load, i prioritized surfacing immediate agentic outcomes on the work item while offloading deep refinement loops to a dedicated rovo chat interface.
i pioneered future-ready patterns designed to accommodate the non-deterministic nature of ai outputs, ensuring the ui remains stable and intuitive across a diverse range of agent behaviors.
V1: Rovo button agent indicator
V2: Agent question when assigned
V5: Agent as sub assignee
V7: Rovo question animated
V8: Agent in upcoming actions
Results
7 third party MCP agents and all internal agents are available to be assigned either on a work item or as part of the workflow.
For our GA launch, we partnered with Canva, Figma, Amplitude, Box, GitHub Copilot, HubSpot, and Intercom as our first ready-to-use agents.
Research & insights
Blend custom & Jira training
Admins create lots of custom material (SharePoint/Confluence/LMS, videos, etc.) but struggle to distribute it and keep it up to date. They want to plug this content directly into Jira (custom welcome messages, intro screens, in-context help) and mix it with Atlassian/Jira training rather than sending users to disconnected docs and emails.
New users are confused early
Users struggle to understand how Jira is set up (who sees what, which workspace/project to use) and how it's meant to be used ("how we work here"). Admins lean heavily on teams, champions and power users; many new joiners are essentially "thrown in and expected to figure it out."
Healthy adoption → lower churn
Onboarding users should be measured by effective use (active, satisfied users and usage spreading across teams). There's an opportunity to give admins activity/adoption metrics and feedback loops to optimise onboarding flows, and help them understand how better onboarding helps them hit goals (healthy adoption, reduced churn, better ways of working).
Mostly support technical needs
Existing champions/admins favour transmitting technical knowledge and often ignore business projects. Opportunity in creating team-customisable areas for context/guidance, and a more holistic in-product communication layer (beyond one-off onboarding) for announcements, changes to ways of working, and product changes.
End user branded custom modal
![Custom onboarding modal]()
View documentation in Jira
![Practice guide in situ]()
Targeted customers & needs
Need to set up and maintain workflows across projects so teams can get the most value from Jira.
They must learn a new product—and often their team's specific ways of working—before they can even start on their tasks.
Leverage their deep understanding of team operations to deliver on time.
RovoAI powered space welcome
![RovoAI powered space welcome in Jira]()
Adoption metrics
![Custom onboarding metrics dashboard]()
Insights from concept testing with 8 customers
2 Levels of Knowledge
We identified a clear split in ownership: site admins create and maintain the company-wide ways of working, while space admins organically manage and evolve their own team's processes.
On brand documentation
When content matches a company's brand, it feels more relevant and trustworthy—so people are more likely to read, remember, and act on it.
Information in small bites
Steps are a preferred pattern to deliver 1 idea at a time and keep things clear.
Rovo AI should know
Most testers expected their documentation to be easily queryable in Rovo, regardless of where it's stored—such as third-party tools like Google Slides.
Added multilingual and dismiss-once-viewed
![Custom onboarding modal]()
Results
EAP roll out to 6 enterprises
Helped us test and refine our first release and determine the value for admins and end users.
1,200 enterprise adoption post GA
Have created custom onboarding for their internal Jira users to serve them their dedicated documentation.
4 new enterprise deals
Closing the feature gap offering helped our sales department in closing new deals worth millions of dollars of annual contracts value.
The approach
Attracting partners: the b2b lead engine
With a convincing brand website for local operators that highlights Express Wi-Fi is a well-established service, already operating in many regions with strong demand and proprietary technology, and that partnering with us will help them quickly grow their business and customer base.
Onboarding the users: the shared connection flow
I built a lightweight, white-label experience for partners to offer their customers. by simplifying the "find, pay, and connect" journey, i created a friction-less path that brought new users to the partner while simultaneously onboarding them into the meta ecosystem.
Targeted customers & needs
Need to set up and maintain workflows across projects so teams can get the most value from Jira.
They must learn a new product—and often their team's specific ways of working—before they can even start on their tasks.
Leverage their deep understanding of team operations to deliver on time.
Research & insights
Performance-driven accessibility
since the majority of users access the platform via low-to-mid-tier smartphones, i architected a lightweight, responsive experience optimized for limited processing power and slower data speeds, ensuring fluid navigation across all device classes.
The pricing & trust paradox
research indicated a high sensitivity to cost combined with a deep distrust of "cheap" services. i addressed this by designing for radical transparency—using concise, straightforward copywriting and clear value markers to build immediate credibility and user trust.
Scannable visual hierarchy
observing that users scan headlines rather than reading long blocks of text, i engineered a high-speed navigation model. by utilizing bold typography, vibrant localized photography, and strong affordances, i ensured the core message remains clear during rapid scrolling.
Localized mental models
users in this market navigate through strong color cues—associating "red" with vodacom and "blue" with airtel. i leveraged these existing brand associations and integrated local imagery to create an intuitive, familiar environment that drives engagement through cultural relevance.
60 days rollout impact
+20,000 unique visits brand website
Brand website and services were featured in local news, significantly bolstering our presence.
329 leads, 96 partnership inquiries
From Indian ISPs, city councils, and other local retailers looking to expand their network and user base in a cost-effective way.
+230% revenue from white-label site
From local new and returning Express Wi-Fi users, proving a strong business case for partners to operate and expand hotspot locations.
The approach
Porting established digital patterns
The goal was to eliminate the learning curve by integrating the digital elasticity of modern devices into the cabin. We adapted the interactions people are already used to operating—fluid transitions, modularity, and gesture logic—ensuring the interface felt like second nature. By mirroring the speed of everyday devices, we made the transition from phone to cockpit effortless.
Physical input to digital response
We aimed to create a sense of continuity between the screen and the car's analog controls. The focus was to ensure that any manual operation—whether engaging a lever, a switch, or a dial—triggered an immediate, synchronized digital reaction. By grounding the UI in this direct feedback, we ensured the pixels felt as reliable and "real" as the physical hardware, delivering a truly seamless experience.
The evaluation
By cross-referencing DENZA's marketing insights with our existing research, we identified a primary user group: mostly married men living in urban areas, tech-savvy, motivated by saving on fuel and maintenance, and committed to reducing air pollution. They expect the car to actively help them achieve these goals while being intuitive to use.
Pain points and opportunities
The current infotainment system is riddled with issues—from a low-resolution display to an overly complex information architecture that makes features hard to find. It also suffers from key usability problems: touch targets are too small, feedback is delayed or missing, and overall performance feels noticeably laggy.
The direction
In depth refresh
We rethought the instrument cluster from the ground up, using the older layout as a reference and then sharpening each zone's role: the left now surfaces key vehicle features, the middle focuses on driving information, and the right controls and displays the head unit. We also defined the physical control scheme for both the steering wheel and head unit, including a compact set of four haptic-feedback buttons.
Curated customisation
Because the head unit is fully touch-enabled, we prototyped highly interactive UX concepts early so we could quickly share them and test with drivers. The final design uses a flexible card system that users can customize and rearrange, delivering a personalized, modern experience that still adheres to all safety and regulatory requirements.
The insights from concept testing
Behavioral transfer
Participants naturally applied learned digital patterns from their mobile devices. Relying on these existing habits significantly lowered the barrier to entry and reduced the cognitive effort required to perform tasks while driving.
Navigation & hierarchy
Testing confirmed that deep menu structures caused distraction. We refined the interface to eliminate layers of sub-items, moving toward a flatter architecture that surfaced critical functions. This ensured the most used features were never buried under multiple "clicks."
Customisation limits
Excessive personalisation proved counterproductive. Drivers required essential vehicle controls to remain in fixed, predictable locations. Too much flexibility forced users to rely on active memory rather than instinct, which increased stress.
– Sydney, AU
First impressions matter
Simplicity is intentional
Familiarity drive scale
Reward time and effort