Portfolio Presentation · Karthik Srinivasan
Research
that changes things.
Researcher Profile
Karthik Srinivasan
Senior UX Researcher
Computer Science & Human-Computer Interaction
Computer Science & Human-Computer Interaction
"developer turned researcher"
↑ this matters for enterprise UX
"Passionate about transforming enterprise technologies into human-centered experiences that influence product strategy and drive real user impact."
6+
Years Enterprise UX
Cloud · Edge · Data Center
Domain Expertise
Microsoft + Equinix
Where I've Worked
About Me
Where I Started
Software developer, 10 years ago.
Computer Science background. Built systems before I designed them. I think about infrastructure before I think about interfaces.
What I Do Now
UX Research in Cloud, Data Center, and Edge.
Strategic initiatives and high-visibility products. The CS background is a superpower here.
How I Collaborate
Stakeholders across the ladder and disciplines.
Individual Contributors, Managers, Directors
VPs, C-level executives
Branding, Engineering, Product, Legal
What I Built
Research Infrastructure and Research Culture.
Expanded team from 2 to 8 researchers
Research Repository and onboarding programs
FTE researchers, contractors, and interns
Case Study 01
Solution Builder
Designing collaborative experiences for complex network infrastructure solutions
Case File · C-01
I pitched a 3-phase research study to Directors and VPs when the product team had resources but no clarity on what to build. This is the story of how research shaped an entirely new product from the ground up.
The Problem
The Scenario
A network architect is designing a hybrid multi-cloud solution with presence in multiple regions.
But they struggle to work through it.
The complexity GSAs were building
Business Goal
Reimagine the designing and ordering experience for customers to improve revenue
Where They Struggled
Find all information needed in a unified view
Pick the right infrastructure components without expert assistance
Communicate with Global Solutions Architects due to limited technical knowledge
My Involvement
I pitched the research.
The product teams already had prioritized resources to build a tool for the immediate release cycle. But they did not know what to build. They lacked a clear understanding of what to solve.
I leveraged my long-term relationships with the product team and pitched to Directors and VPs, a 3-phase study to understand our problem better. Got their alignment.
Role & Research Arc
Role · Lead UX Researcher
Solution Builder touched multiple products, so the work meant collaborating with stakeholders across products and across levels, from designers to directors.
Core Collaborators
2 Designers + 1 Design Manager
2 Product Directors
1 Researcher (me)
Other Stakeholders
Branding, Marketing
VP of Product
Engineers and Product Managers
Constraint
Short timeline. Tool was prioritized for the next release cycle.
Needed to run research fast and still do it right.
Research Arc
Phase 1 · Research
Understanding the Problem
2 wks
Design
Design Phase
3 wks
Phase 2 · Research
Testing the Solution
1 wk
Dev
Development
2 mo
Phase 3 · Research
Post Launch
2 wks
Research phase (I led)
Design / Dev (I collaborated & gave input)
Research Groundwork · Phase 1
Preparation · Before the First Question
Three moves to set the study up under a tight timeline.
1
Scoped the deadline
2.5 weeks to report insights, so I scoped tightly from the start.
2
Reviewed documentation
Used existing documentation to narrow scope before talking to anyone.
3
Aligned with the team
Conversations with PMs and Designers to understand their needs.
What We Were After
Reimagine the designing and ordering experience to improve revenue.
Research goal: understand how customers design their network solutions, and the challenges.
Research Questions
How do customers know what they want for their use-case?
How do they diagram their network solution?
What are the challenges and workarounds?
What tools and artifacts do they use?
Research Plan · Interviews · Phase 1
Three decisions shaped the interview study: how I would learn, who from, and how I would reach them.
01 · Method
Needed rich qualitative data, so I chose semi-structured interviews.
The research questions demanded a deep understanding of mental models, workflows, and pain points. A conversational format let me follow the thread wherever it led.
SEMI-STRUCTURED
60 MIN SESSIONS
REMOTE · ZOOM
02 · Participants · The Who
Customers and Solution Engineers.
5 Customers · Network Engineers and Architects
4 Solution Engineers as technical domain experts to manage the time constraint
03 · Sampling · The Reach
Screener survey for purposive sampling
Outreach via customers enrolled in the Research Panel
Interview Process · Phase 1
1
Research
Plan
Plan
Proposed plan to stakeholders
Collaborated async on iterating the plan
2
Research
Prep
Prep
Identified participants
Prepared outreach plan
Updated stakeholders on Slack
3
Execution &
Analysis
Analysis
Invited stakeholders to observe
Shared interim insights on Slack
Analyzed data
4
Reporting &
Next Steps
Next Steps
Presented insights and recommendations
JIRA tickets to track action items
Collaboration
Throughout
Throughout
Team Syncs
Doc Sharing
Dedicated Slack Channel
Email Updates
watched specific segments of videos for thoroughness
prepared highlight reels for stakeholders
Execution and Analysis · Interviews
1
Interview
Setup
Setup
Semi-structured interviews
60 minutes per session
Done remotely over Zoom
2
Interview
Questions
Questions
Customer Background
Use-cases
Diagramming
Tools and Artifacts
3
Data
Collection
Collection
Research Notes
Video recording
Consent Form
4
Analysis
Thematic analysis
Coded data, identified themes
Pulled relevant quotes from transcripts
Polished report
By the
Numbers
Numbers
9 Interviews
4 Solution Engineers
5 Customers
Insights from Interviews · Phase 1
Key Findings · Phase 1 Interviews
Three pain points confirmed.
Back-and-forth conversations across multiple platforms
Lack of visibility to technical information and pricing
Unable to collaborate on a single place to share information and diagram
Started out research anticipating to build a diagramming tool, but the insights told us we needed to solve for collaboration.
Post-interview synthesis
What This Meant
Needed wider stakeholder involvement and alignment now
Stakeholders were divided
Only 3 days to secure alignment and start designing
this was the pivot moment. expected one thing, found another
Kano Survey · Feature Prioritization
Note to self
31 responses in 2 days. way faster than expected!
Kano Survey · Activity Type: Prioritization
Breaking a 3-day stakeholder deadlock.
Sent to aliases with over 600 employees. Partnered with executives to spread word to their teams.
Must-Haves
Collaboration · Pricing · Visibility into deployments
Nice-To-Have
Latency Metrics · Onboarding tutorials · Templates
Less Important
Other 8 features
31
Responses
2
Days
½
Day to analyze
I ran a Kano survey mid-project. It wasn't in the plan, but securing alignment in 3 days left me no choice.
On the Kano decision
Challenges · Phase 1
What Made This Hard
Three constraints at once.
Tight deadline · only 2.5 weeks to collect data and report insights
Divided stakeholders · needed to align with challenging stakeholders, and come up with a quick means to achieve that (invented the Kano survey)
Executive expectations · manage executive communication without creating noise
How I Navigated It
Leveraged existing relationships with PMs and Directors for rapid alignment
Created the Kano survey in 1 day to generate data-driven stakeholder buy-in
Set up a Slack channel for real-time updates, kept wider audience informed via email
"High visibility project. Lot of pressure. Managing stakeholders was tough, lot of headwinds."
Design Phase · Preparing for Phase 2
While Design Was Happening
I was preparing for concept testing in parallel.
Collaborated with 2 Designers during design phase to ideate and translate features to design
Simultaneously preparing the concept testing protocol and scheduling stakeholders
Recruited 4 customers from panel and 3 Solution Engineers to minimize logistics
Planned efficiently to collect data, analyze, and report insights in 1 week
What came out of the design phase
Research Plan · Concept Testing · Phase 2
Study Setup
7 participants. 1 week.
Participants were provided a short scenario and tasks to set up a hybrid multi-cloud setup on a click-through high fidelity prototype.
4 Customers · Network Engineers, Network Architects
3 Solution Engineers
Questions asked in-between and post-task
Research Questions
How would customers respond to diagramming and configuration features?
How do customers respond to the collaboration features?
Are customers able to design a solution seamlessly?
What information is important to customers after designing a solution?
Building a solution on the canvas
Topology mapped to real metros
Concept Testing Results · Phase 2
Insight 1 · Concept Testing · 4 Customers
The response split down the middle.
Experienced Users
Found the tool empowering and easy to use.
Novice Users
Struggled · needing more guidance and simplified flows.
These insights underscored the need for two distinct flows: guided onboarding for novices, exploratory mode for experts.
Concept Testing Synthesis
Insight 2 · Solution Engineers
Technical feedback sharpened the prototype.
Solution Engineers provided detailed feedback on the technical configuration of assets, surfacing details only domain experts would catch.
This led to a more capable and polished prototype that excited customers in testing.
Solution Engineer Collaboration
Launch Note
Solution Builder was in Beta 2 months after concept testing, and GA 4 months later.
Post Launch · Setting Up Phase 3
Setting the Stage
Good engagement, and a deadline.
The product team was seeing good engagement of the tool. Our team wanted to measure how it actually worked for customers. Then came the deadline: 3 weeks for company all-hands where we wanted to report this data.
It started with a conversation, not a brief.
In conversations, I understood the team was looking to measure metrics and compare NPS with other product areas. That shaped exactly what we designed to measure.
Note to self
the all-hands deadline made this urgent. needed to move fast and still be rigorous
Survey · Phase 3 · Post Launch Measurement
Survey Design · n=251
Measuring real impact after launch.
Measuring user satisfaction with discoverability, ease of use, and relevance of the Solution Builder tool. Likert scale (0-10). Also asked the likelihood of recommending the product (NPS question).
Categorized responses into promoters, passives, detractors
NPS Score: 81% (Industry standard 44%)
The NPS Result
NPS Score
81%
0
Industry avg 44%
100
What Drove Satisfaction
Measured satisfaction drivers by correlating recommendation likelihood to individual metrics (Pearson correlation)
Higher ease of use correlated to higher recommendation score (Pearson r=0.68)
I collaborated with a data analyst on this
Ease of use predicts recommendation. Pearson r=0.68. This became the design team's north star metric.
Key analytical finding
Impact · Solution Builder
What It Moved
3K+
Monthly Active Users · Robust tool used by Customers and Solution Associates
2 days
Saved per pre-sales · GSAs reported saving time for each pre-sales engagement
NPS 81%
vs industry avg 44% · strong customer satisfaction
Challenges in Impact
High visibility. Lot of headwinds.
High visibility project with lots of stakeholders and pressure
Managing stakeholders was tough with competing priorities
Short timelines throughout all three phases
"High visibility. Complex stakeholders. Tight timelines. Navigated all three."
Case Study 02
Equinix Portal
Converging all portals through unified Information Architecture
Case File · C-02
Solution Builder was a massive success. This visibility led to our VP enlisting me in conversations for this massive product overhaul from the early stages.
this one was mine to shape ↑
My Involvement & Research Timeline
Role · Lead UX Researcher
More time. More control.
This time, I was given more time and control to plan my research. The VP came to me after Solution Builder.
Core Collaborators
2 Designers + 1 Product Manager
1 Content Writer + 1 Researcher (me)
Other Stakeholders
Branding team
Senior Director of Product
Other Product Managers
Research Timeline · 3.5 of 8 months
Phase 1 · Research
Creating IA Models
2.5 mo
Design
Design Phase
2 mo
Phase 2 · Research
Testing New Navigation
1 mo
Dev
Handoff & Development
3 mo
Research phase (I led)
Design / Dev (I collaborated & gave input)
The Problem · Equinix Portal
The Problem
12 portals. No unified home.
Customers had to manage critical infrastructure across multiple platforms, each with its own navigation system, terminology, and quirks.
Managing each product on separate portals
Switching contexts constantly
Struggling to locate relevant information
Business Goal
Unify products under a single portal.
To manage all information in a single place. One home for all of Equinix's critical infrastructure.
From 12 separate portals to 1 unified experience. Every product, every workflow, every customer. One home.
The north star
12 portals meant customers were basically learning 12 different products
Research Groundwork · Phase 1
Preparation · Before the Card Sort
Two moves laid the groundwork.
1
Met stakeholders and executives
To understand objectives and the impact this work needed to drive.
2
Explored different directions of IA
Mapped possible information architectures before committing to a research approach.
What We Were After
Unify products under a single portal to manage all information in one place.
Research goal: understand how customers group together navigation items.
Research Questions
How do users group navigation items? Any significant variation from how products are structured?
Are there concepts users struggle to understand?
Assumption: customers might want common functions (orders, billing, support) for all products in one place.
Research Plan · Hybrid Card Sort · Phase 1
A hybrid card sort fit the goal: enough structure for many cards, enough freedom for customers to build their own categories.
01 · Methodology Choice
Why a hybrid card sort?
I wanted to understand the customer mental model for grouping navigation. With many cards to sort, a fully open sort was impractical, but I still wanted participants to create their own categories when needed. Hybrid struck the balance.
HYBRID SORT
25 MIN SESSIONS
OPTIMAL SORT
02 · The Cards
Features across all portals.
Product features and functionalities drawn from every portal we were converging, written as sortable cards.
03 · Sampling
50 Internal Users + 30 Customers
Active users in the past 6 months (purposive sampling)
Outreach via email distribution
Card Sort Process · Phase 1
1
Round 1
Internal · n=50
Internal · n=50
72 cards of features and functions
7 predefined categories
Validated that common functions group together
2
Round 2
Customers · n=30
Customers · n=30
Simplified, focused approach
25 cards x 2 activities
Focused on less clear groupings from Round 1
3
Analysis
Similarity scores and dendrogram
80 participants total
Identified strong and weak groupings
4
Reporting &
Handoff
Handoff
Synthesized into IA recommendations
Shared with design and product
Communication
Channels
Channels
Team Syncs
Doc Sharing
Dedicated Slack Channel
Email Updates
two rounds let me pressure-test the groupings round 1 left fuzzy
What I heard
participants thought aloud as they sorted. you could hear them hesitate on our product names, unsure what half of them actually meant.
The Card Sort Interfaces · Phase 1
What participants actually saw: the cards to sort, and the categories to sort them into.
The card sort interface for Round 1
The card sort interface for Round 2
running it remotely on Optimal Sort let me reach 80 participants across two rounds
Challenges · Card Sort
What Made This Hard
Recruiting 30 customers was challenging. Multiple rounds of outreach needed.
The outreach itself was exhausting. needed a large sample size. I manually compiled a list of over 1,000 customer contacts from multiple data sources.
Lower response rate · recruited some customers through placing product banners in-platform.
Too many navigation items across 12 portals. Challenging to design the card sort, trying to cover navigation items across all 12 portals.
"I manually compiled over 1,000 customer contacts from multiple data sources. That's the unglamorous part of research that makes the difference."
On the outreach effort
compiling 1,000 contacts by hand was the least glamorous part of this study
Insights from Card Sort
Insights · Card Sort · Phase 1
Language mattered more than structure.
Many struggled with "IBX" · Rebranded to "Colocation"
"Packet" grouped under "I don't know" · Rebranded to "Equinix Metal"
Cards with functional names had higher agreement rates (eg: Manage Cables, Install Equipment)
"Customers navigate by what they do, not what we call our products."
Core insight driving the rebrand
80
Participants
50 Internal Users · 30 Customers
this reframed the whole project, from structure to language
IA Models Exploration · Phase 1
I explored three ways to structure the unified portal, each a different bet on how customers think about the platform. Press Space to step through the models.
Aligning on IA Model
The Challenge
Wanted functional naming. PMs were nervous.
Wanted to create a functional naming and task-based navigation model. However, PMs for smaller products were concerned the visibility of their product would be affected.
Decision: Picking product-based navigation to leverage existing branding
But: Use functional naming for many products within it
Worked on final IA · Collaborated with branding and marketing teams
Decision
Model A, but functional naming throughout. Preserved PM visibility. Data made it non-negotiable.
"Research data made the naming changes non-negotiable. PMs worried, data answered."
On stakeholder alignment
Testing New Navigation · Phase 2
Research Plan · Phase 2 · Testing IA
Testing with 12 customers.
Participants were provided several tasks to accomplish on a prototype, right from connecting to a Z-side server, to reviewing reports, or billing data. Asked questions in-between and post task.
Research Questions
How would customers respond to functional naming on left navigation? (eg: Data Center Monitoring, Virtual Services)
How do customers respond to global navigation and unified pages for orders, billing, reports?
Are customers able to navigate across product pages seamlessly?
What information is important to customers on the global landing page?
New navigation under test
Unified pages walkthrough
Usability Testing Results · Phase 2
Usability Testing · 12 Customers
The left and global navigation was a success.
Participants described it as "organized" and "modern" while comparing favorably to AWS and Azure.
Easily Understood: "Resources" → "Inventory" improved clarity. "Colocation" and "Internet Access" immediately understood.
Improved Discovery: Users discovered Virtual Services for the first time. left nav acts as a product catalog.
Unified Branding: "Colocation" consistent across nav, marketing, and content.
The Operational Grind
Manually compiled 1,000+ customer contacts from multiple data sources
Placed product banners in-platform to boost response rate
Many stakeholders worried that changes would affect product visibility, navigated with data
"Research data made the naming changes non-negotiable."
Post-study synthesis
hearing them compare it to AWS and Azure told me we'd landed it
Impact · Equinix Portal
What It Moved
Rebrand
Colocation · Internet Access · Equinix Metal · Name changes that stuck
71%
Revenue Increase · Network Edge improved visibility and gained more revenue
18%
Lower Churn · Improved experience translated to lower cumulative churn rate across all products
Challenges in Impact
Many stakeholders worried that changes would affect their product visibility
Recruiting participants for card sort required tremendous effort and manual outreach
Navigated all concerns with data and evidence from research
"Research data made the naming changes non-negotiable. PMs worried, data answered. That's the power of rigorous research."
the rebrand stuck, and that's the part I'm proudest of
The Connecting Thread
Research that
moves the needle.
Two cases off one wall, the same DNA underneath both: start in ambiguity, build the method, move the stakeholders, land on a decision that moves a number. What connects them isn't a technique. It's a conviction that complex enterprise systems can, and should, be made human.
CASE 03
Azure onboarding
CASE 04
Learnability of AI suggestions
CASE 05
AI trust study
CASE 06
CASE 01 · SOLUTION BUILDER
A tool built from ambiguity
A 3-phase study shaped a product the team had no map for.
3K+
MAU
2 days
Saved / project
CASE 02 · EQUINIX PORTAL
Twelve portals, one structure
Card sorts and IA testing unified a fractured experience.
71%
Revenue ↑
18%
Lower churn
the same thread, both times
Thank you.
two stories tonight. the wall holds the rest.
still developing