Google Cloud for startup

Embedding AI into the Product Discovery Experience for Startup Customers
GenAI
0-1
Web

Duration

4 months

Deliverable

3 tested hi-fi prototypes;
Research report

Role

Lead UX Researcher &
UX designer

Collaborator

1 Senior Designer @Google;
1 Student Researcher @Cornell
3 Student Designers @Cornell
highlights

What's special about this project?

0→1
Innovations

Back in 2023 Q3, Google Cloud website didn’t have a chatbot, so everything in this project was new

Evidence-based
despite limited access

With just 4 months and hard-to-recruit users, I bridged the gap through strategic research planning.

Details

My Role as
a Student Leader

All execution was handled by the team, but I planned “when to do what” and ensured quality assurance

Details
Context

Cloud solution selection is Overwhelming

Google Cloud offers 100+ products, each with different pricing, integrations, and capabilities, making comparisons difficult. Users struggle to find the right solution without clear guidance.

My Responsibility

I led the research from planning to execution, defining methods, sourcing participants, ensuring insights shaped design, and managing the project to keep execution on track.

The Deliverables

0→1 Hi-fi Prototypes

Back then (2023 Q3), Google Cloud website didn’t have a chatbot (though they do now in 2025), so everything we designed in this project was from 0 to 1.

The Chatbot

To enhance discoverability, engagement, and decision-making by providing personalized, real-time recommendations.

Dynamic UI

To adapt to different user needs at various stages of their journey—landing, exploring, and comparing—ensuring efficient discovery and decision-making.

Comparison Tool

To enable side-by-side evaluation with "add to compare" selections

🔒 Take A Closer Look
Impact

Short project, lasting impact

Validated Hypothesis

Final Prototype SUS = 86.3% (high usability). Our also insights confirmed that helping users differentiate between similar cloud solutions is key to improving adoption rates

Comprehensive Hand-offs

Provided Google with raw data, interview protocols, insights, and participant connections to support ongoing improvements and future studies

Driving real implementation

Through 2 rounds of presentations, 6 Google executives expressed strong interest in further exploring the chatbot feature, which is now being implemented on the website

Overview

TLDR

Sprint 1

Identify the project scope

Sprint 2

User Research

Sprint 4

Solution Details

mindset
Initial Problem Statement

Ask myself:
Which stage is causing fiction?
Among them, which stage impacts adoption the most

Narrow Scope
Research Questions

What mental models do startup leaders use when evaluating cloud solutions to purchase?

What are the UX gaps between Google and its competitors in supporting cloud solution discovery?

What are cognitive biases and trust mechanisms that influence digital product purchase?

Choosing the research method

Balancing the feasibility & data significance

Semi-Structured Interviews
Competitive Analysis
Literature Review

N=8 (hard to recruit)

Startup CTOs & Engineer Lead—key decision-makers in cloud adoption

Why this method?

In qualitative research, more interviews don’t alway lead to better insights—they lead to redundancy. Given the limited availability of our target users (Startup CTOs, CEOs, and Founders), I adopted a saturation approach to maximize insights while minimizing redundancy, given hard-to-reach participant pool.  

Competitors

AWS, Azure

Why this method?

Competitive analysis was essential for identifying UX gaps in product discovery and differentiation. AWS and Azure are direct competitors recommended by the client based on previous Google research -- helpful since analysis helped uncover where Google Cloud could enhance AI-powered recommendations

Research Topics

Decision-Making, Online Purchasing, AI Trust

Why this method?

A fast, research-backed foundation to understand cognitive biases shaping decision-making, trust, and risk perception in cloud solution selection.

Why mainly qualitative method, but not quantitative?
  • Depth Over Volume: Startup leaders are hard to reach, making large-scale surveys impractical. Instead, in-depth interviews ensured rich insights from high-value participants.‍
  • Behavior Over Self-Report: Mental models can’t be surveyed. We needed to see how users actually think, filter options, and make trade-offs—not just what they claim to do.
  • Surveys Lack Reliability: Low engagement + varied expertise = statistically weak data. Instead of chasing incomplete numbers, we focused on actionable, behavior-driven insights.
Ownership

Minimize the execution & expectation Gap

Tiered Recruitment
Parallel Planning
Communication

In-Person Recruitment@ Ithaca Social Events

In-Person Recruitment@ Ithaca Social Events

Recruiting the right participants is critical, but our target users were so rare —e.g. Startup CEOs, Technology Officers, and Startup Founders—who are also usually hard to reach due to their packed schedules. Therefore, I implemented a tiered recruitment strategy to balance efficiency and research validity.

  • Direct Users – Current users of Google Cloud, providing firsthand insights
  • Adjacent Users – Users of similar cloud platforms with overlapping needs
  • Potential Users – Individuals exploring cloud solutions but not yet active users

Details on "hierarchical approach" in UX research recruitments >

Visualization I individually created to manage parallel research tasks

Research rarely goes exactly as planned. As the research lead, I strategically planned and assigned tasks to minimize gaps, ensuring efficiency through parallel research.

I assigned research tasks

I reviewed teammate's work to ensure quality

Results

Within 20 days, we successfully recruited 8 startup CEO/ CTOs, ensuring a diverse yet relevant sample.

Insights & Solutions

Please reach out for More Details!

Enter Password

Web-based GenAI