Pulse AI

Intro

In practice, analytics is an iterative process that involves identifying stakeholders and success metrics, gathering and standardizing data from multiple sources, ensuring data quality, conducting exploratory analysis to uncover patterns and trends, and communicating insights through dashboards and reports.

Amplitude and Mixpanel are powerful product analytics platforms that help teams understand user behavior through dashboards, funnels, cohorts, and retention analysis. However, non-technical users often struggle to explore data independently and derive meaningful conclusions without analytical expertise. More importantly, these tools primarily explain what happened rather than what should happen next. While they provide insights and visualizations, they offer limited built-in recommendations or proactive guidance, leaving users to manually interpret findings and decide on actions. As a result, they function more as analytics dashboards than true copilot-style decision-support systems.

1. Skilled user with knowledge of Analytics: Even with self-serve analytics, users still need to know where to look, which chart to use, and how to interpret patterns. Mixpanel highlights interactive reports and self-serve insights, while Amplitude promotes complex analysis and AI agents, but that still assumes users can guide the process well.

2. Knowledge to transform findings into actionable recommendations: Competitors help identify drop-offs and friction through analytics and session replay, but users often still need to manually decide what to do next. Session replay explains behavior, but does not always convert findings into prioritized recommendations or product actions.

3. Analytics tools are far from being user friendly: Analytics tools are often easiest for analysts, PMs, or growth specialists. Designers, executives, and less technical stakeholders may still struggle to translate data into useful product direction.

GOAL

How might we simplify complex analytics so cross-functional teams can drive faster, more confident product decisions?

I independently researched the product analytics space over an 8-week period to design Pulse AI, an AI-powered copilot engineered to make complex data proactive, understandable, and action-driven for cross-functional B2B teams.

KEY RESEARCH INSIGHT

The First Few Minutes in the Dashboard Shape the Enterprise User Experience

I focused on front-loading my generative research in the first stage of my design process. I distributed a survey amongst 10 people and interviewed 5 potential users to gain a deep understanding of their goals and frustrations.

This research confirmed that enterprise users value intuitive design, especially in AI dashboards where complex insights must be understood quickly. Users often decide within the first few minutes whether to continue using a product, making clarity, trust, and easy navigation essential.

1. 67% of participants who are non- technical struggle to explore / understand data.

2. 45% use Amplitude, the most popular analytics tool, to analyze products.

3. 25% indicating they use more than one analytics tool which highlighted the need for an alternative.

4. 45% of participants wanted features like conversational analytics, auto - generated product insights and predictive user behavior.

SYNTHESIZING OUR INSIGHTS

Defining our Personas

I then synthesized my insights into a person. This would act as reliable and realistic representations of my audience segments that I could refer to while designing thus allowing me to always keep the user in mind.

COMPETITIVE ANALYSIS

Existing Applications lack action-oriented AI experience that helps teams to make product decisions.

In my user research, I learned from users that the existing tool provides just analyzing data which is hard to understand for non-technical users. Keen to not make the same mistakes with my solution, I performed a competitive analysis of Amplitude, Mixpanel and Pendo. This gave me an opportunity to experience the extensive process first hand and made me realize the direction for the MVP - simplicity. This gave me an opportunity to experience the extensive process first hand and made me realize the direction for the MVP - simplicity.

To have my solution perform well as an MVP, simplicity would need to be adopted, and hence I prioritized ensuring there was a quick and easy AI generated product actions.

A Visual example of dashboards with complex data.

INFORMATION ARCHITECTURE

Designing the Familiar

To ensure the MVP was within scope, I crafted four user stories to aid the creation process:

User Story 1: Discover a problem

As a Product Manager, I want the system to proactively surface important product issues, so I don't have to manually search through dashboards.

Scenario: Retention drops unexpectedly.

User Story 2: Investigate why drop-off happens

As a Product Designer, I want to understand where users struggle in a flow, so I can improve the UX.

Scenario: Checkout conversion is low.

User Story 3: Ask questions in Natural Language.

As a Product Manager, I want to ask product questions conversationally instead of building complex reports.

Scenario: Need churn insights.

User Story 4: Turn insight into experiment

As a Growth Analyst, I want recommended experiments based on detected problems, so I can improve metrics faster.

Scenario: Onboarding drop off detected.

SITE MAP

LOW FIDELITY WIREFRAMING

Bringing It All Together

With my user stories mapped out, I moved into the next phase of designing the application, starting with some rough sketches. I incorporated the best practices and existing design patterns to ensure the experience was user friendly and then digitized these wireframes for usability testing.