Improving user retention for Microsoft 365
Our subscription management experience allows customers to cancel their Microsoft 365 subscriptions. As part of modernizing commerce, we redesigned the cancellation flow to boost retention by leveraging AI-driven recommendations that highlight the value of the subscription—while keeping the process simple, transparent, and user-friendly.
Company
Microsoft Corporation
Role
Design lead
Year
2024
status
Shipped
EXECUTIVE SUMMARY
Problem
Microsoft 365's cancellation flow felt transactional and generic. The existing flow lacked persuasive messaging, failing to remind users of the platform's full value. This resulted in higher churn rates and missed opportunities to retain users who might benefit from staying subscribed.
solution
We redesigned the experience to integrate Copilot AI based insights to highlight the value of the subscription based on customer's actual usage data, to ultimately improve retention. More importantly, we kept the new experience user-centered, avoiding coercive tactics that could undermind user trust.
RESEARCH AND INSIGHTS
Mixed research
Our research approach wasto combine quantitative usage data, behavioral patterns, market trends and actual user feedback to uncover both what users were doing and why they were doing it.
Research goals
1
Identify the driving reasons behind subscription cancellation.
2
Uncover user pain points within the current cancellation process.
3
Determine how other services handle cancellation and retention.
Research methods
Quick user chat
Chatted with 12 users who recently canceled or considered canceling their subscription to understand what motivated them to make the decision.
USAGE data analysis
Analyzed usage data from canceled and retained accounts to identify patterns. For e.g., users with low OneDrive usage showed a higher likelihood of canceling.
REFERENCES
Evaluated the cancellation flows of popular subscription services like Adobe, Netflix, Slack etc. Noted consistent patterns in persuasive messaging and retention.
INSIGHTS RELATED TO OUR existing EXPERIENCE
Confusion
Users canceled their subscription immediately due to unclear cancellation options that made it seem like it was the only choice.
DIFFICULTY in RECALLING
9 of 12 users didn't remember the key benefits of their subscription, like advanced security tools and OneDrive Storage, while cancelling.
Lack of personalization
Multiple users mentioned how our cancellation experience lacked any form of personalization and did not encourage them to reconsider their decision.
OTHER INSIGHTS
Cost sensitivity
The most cited reason for cancellation was cost, particularly among users who felt they weren’t utilizing enough features to justify the expense.
trust BUILDING FACTORS
Across services like Adobe, Netflix, and Slack transparency, flexibility, and confirmation were a consistent theme in their cancellation flows.
ONE KEY OBSERVATION
KEY FINDING
Users who showed high engagement with core features, like OneDrive storage, were less likely to cancel their subscription.
INTERPRETATION
When users regularly use key features, they better understand the value they receive, which makes the cost feel justified and the idea of canceling less appealing.
IMPACT
This informed our personalization strategy to surface value based on actual usage. By showing users the benefits they stand to lose, we could reinforce subscription value and increase awareness.
DESIGN DECISIONS
Introducing AI-powered personalization
SCENARIO
I initially explored designs highlighting high-level benefits of a Microsoft 365 subscription. But this approach felt too static and generic. Every user saw the same benefits, which may not always capture interest. The challenge was to show value which would be personalized according to every user.
SOLUTION
If AI could help everywhere else, why not here? I floated the idea to integrate Copilot into the cancellation flow, the product team jumped on board, and suddenly—what started as an exciting idea became a real feature.
The AI analyzes individual usage patterns—what features users actively engage with and which they may be underutilizing—and surfaces personalized highlights during the cancellation experience.
Design principles
Transparency
Build trust by clearly explaining each step and potential consequences, ensuring users feel respected and valued.
Flexibility
Keep the cancellation process flexible by providing users multiple ways to cancel, or offer a way to change their plan.
TRUTHFULNESS (AI FOCUSED)
Our AI insights should always be grounded in factual usage data, ensuring users see accurate and trustworthy representations of their subscription value.
HUMAN SUPPORT
Value-based
Highlighting key features reminds users of what they stand to lose, thereby encouraging them to reconsider their decision.
Navigating different design opinions
#1 ADDING FRICTION
Leyla Dalton (She/her)
We should add a friction point and look at routing them through support, especially for enterprise customers. Thoughts?
Siddharth Naik (You)
I get the intention but forcing support as the only option might feel manipulative. Let's look at other options that would add gentle friction.
Challenge
During early explorations, a stakeholder suggested routing all users to contact support before they could cancel. While this approach could potentially reduce churn, it risked frustrating customers who had already decided to leave and creating unnecessary support cost. The challenge was to balance the business goal of reducing cancellations with the user need for a clear, self-service path that didn’t feel manipulative.
SOLUTION
I recommended a hybrid approach that maintained transparency while introducing purposeful, user-respectful friction. Rather than forcing every customer through support, I proposed a design that would present thoughtful retention options like complimentary subscription period, plan downgrade, or a chat with a product expert, before completing cancellation.
#2 balancing minimal steps and simplicity
Sam Morenza (He/him)
Do you think we can simplify the cancellation flow by limiting it to 1 step?
Siddharth Naik (You)
Yeah, we could, but 1 step might cause more overrload. Let's try a multi-step approach.
Challenge
One of the core challenges in this project was balancing the need for minimal steps with the wide scope of information we needed to present. Our product team wanted a single-page cancellation flow, but early prototypes suggested that we couldn't fit all the required content without overwhelming users.
SOLUTION
We implemented a concise, 2-step flow that guided users through the process with clarity and confidence. By displaying multiple visually clean and digestible sections, we reduced the cognitive load, minimized errors, and made the overall cancellation process less overwhelming.
FINAL DESIGN
Empowering choice, highlighting value
Presenting to you — A seamless cancellation flow that educates users on their subscription’s value, offers helpful alternatives, and respects their decision with clarity and ease.
Flexible cancellation options
We now provide users with clear cancellation options, straightforward language, and visible exit points. This approach built trust and reduced frustration.
AI BASED Dynamic value highlights
Copilot AI generates subtle but personalized highlights of subscription benefits based on usage. For instance, users who have high utilization of OneDrive will see prompts related to data storage.
Support touchpoints
We also added a Talk to a Expert chat option for users hesitant about canceling. Our expert agents have context-aware scripts emphasizing relevant benefits or flexible options.
AI RECOMMENDED Personalized retention options
Copilot AI also recommends alternate options to users rather than canceling their subscription. For e.g. downgrade to a cheaper plan, get training on how to use your subscription etc.
No unnecessary roadblocks
While we presented relevant value propositions, benefits, and alternate actions at every step, we wanted to ensure that users could still cancel their subscription easily.
Clear confirmation
A clear confirmation step ensures that users know their action has been successfully processed. This would reduce unnecessary support tickets or dissatisfaction.
Consistency across different scenarios
When users cancel within the first 14 days, they can choose to cancel their subscription immediately, while after 14 days, they can only stop renewal at the end of their billing cycle. Despite this difference in the cancellation policy, our new design ensures both groups go through the same clear and intuitive flow.
This consistency reduces confusion and provides a seamless experience regardless of cancellation eligibility.
OUTCOME
A/B testing
To validate the design at scale, we launched an A/B test involving a flight of 10,000 users over a two-month period:
Group A
Group A experienced the original cancellation flow.
Group B
Group B used the redesigned flow with improvements.
Increased retention
Group B's interaction with the new flow retained 22% more users compared to Group A.
engagement
~44% of Group B's participants explored the Cancel before next renewal option.
USER SATISFACTION
Group B users reported higher user satisfaction rate by almost 45% as compared to Group A.
Final impact
24%
resulted retention against an estimate of 15-20% retention rate within 3 months of launch.
210,000+
users saved against an estimate of 140,000 users within 3 months of launch.
Key takeaways
POWER OF DAta-driven storytelling to influence stakeholders
Solving all the conflicts was crucial. I learned how to use clear visuals, user insights, and business projections to shift stakeholder opinions related to designs.
Designing for user needs can directly drive business outcomes
By prioritizing clarity and reducing friction in the cancellation flow, we showed that good UX equals measurable business success.