Senior User Experience Researcher | Digital Optimization, CRO & Customer Insights | MS in Cognitive Science

I turn behavioral data into business decisions.

I am a Senior User Experience Researcher at Clarivate, previously a UX Researcher at Magicbricks, with an M.S. in Cognitive Science from IIIT Hyderabad. My work spans experimentation, conversion optimization, Voice of Customer, behavioral analytics, competitive intelligence, ResearchOps, and AI-enabled research workflows.

WHERE I LEARNED

What shaped my research approach.

Clarivate

Scalable optimization needs repeatable evidence.

I connect A/B testing, CRO, Voice of Customer, Google Analytics, VWO, Microsoft Clarity, and ResearchOps into repeatable frameworks for customer experience, conversion performance, and leadership decision-making.

Methods used

Magicbricks

End-to-end research connects users, journeys, and product decisions.

As a sole UX researcher, I conducted data-driven studies across buyer, owner, PG, and property-service segments, running end-to-end research on new concepts, existing journeys, and competitor comparisons.

Methods used

IIIT Hyderabad

Experimental research builds disciplined interpretation.

My M.S. in Cognitive Science trained me in experimental research, statistical modeling, usability research, 360-degree VR head-movement analysis, and structured root-cause methods like Fishbone diagrams, Pareto charts, performance metrics, and DMAIC.

Methods used

CASE STUDIES

Selected research impact.

Each project is condensed into problem, approach, and outcome so the evidence is easy to scan.

CLARIVATE · A/B TESTING · VWO · CRO · SESSION RECORDINGS

Improved EndNote Contact Sales form completion through A/B testing

I worked on the EndNote Contact Sales form because users were dropping off before submitting, which directly impacted lead generation.

The Problem

Users were engaging less and dropping off before completing the lengthy EndNote Contact Sales form.
This limited lead generation, so we needed evidence on whether a multi-step form could reduce effort.

Control version of the EndNote Contact Sales form

What I Did

Ran a VWO A/B test: single-step control vs. multi-step variation.
Reviewed session recordings to understand completion behavior and effort.

Multi-step variation of the EndNote Contact Sales form

What I Found

Users perceived fewer fields in one view as less cognitively demanding.
Average completion moved from 3m29s to 1m52s, pointing to lower effort.

Session recording evidence comparing completion effort

Outcome

+12.6% submissions from the multi-step experience.
46.41% faster completion, giving the team evidence to scale form optimization.

Back to case study selection

CLARIVATE · VOICE OF CUSTOMER · NPS · KEY DRIVER ANALYSIS · THEMATIC ANALYSIS

Executed Annual Voice of Customer program to improve NPS and identify innovation areas

The goal was to understand customer needs, improve experience, support retention and advocacy, and find areas for data-driven action.

The Problem

The primary objective was to improve customer retention, advocacy, efficiency, and perceived brand value.
Responses and response rates also needed to increase so teams could rely on the VOC findings.

Four-pillar Voice of Customer program framework

What I Did

Measured value to institution, purchase experience, product quality, functionality, innovation, service, education, and communication.
Used QR codes, events, web, social, and email footers to increase responses, then ran key-driver and thematic analysis.

What I Found

Communication, innovation, product functionality, and product quality were important but poorly rated.
I also used Google NotebookLM thematic analysis to identify areas for innovation.

Key driver analysis from slide 17 of the VOC deck

Outcome

NPS score increased by 3 points in 2025.
Responses increased by 24.7% to 19,667, with response rate reaching 2%.

VOC 2025 highlights showing NPS and response gains

Back to case study selection

MAGICBRICKS · RAG LLM · HUGGING FACE · GOOGLE COLAB · DOCUMENT AI

RERA document chatbot POC

I created a working Python-based RAG prototype so users could ask specific project questions instead of reading all RERA documents manually.

The Problem

It was hard for users to get specific information before deciding whether a property was worth investing in.
Legal details and builder reputation signals were buried across handwritten, multilingual, typed, and sketched RERA files.

What I Did

Coded the workflow in Python on Google Colab and connected a RAG LLM using Hugging Face tokens.
Used Google Sheets for inputs and Google Drive for outputs; used ChatGPT only to refine the code for my use case.

What I Found

A chatbot was a more practical experience than asking users to manually inspect every document.
Retrieval-augmented generation could return document-grounded answers from messy project files.

Outcome

Delivered a working Python-based chatbot POC for project-specific RERA queries.
Showed how RAG could reduce effort in legal and builder-reputation checks before confident property decisions.

Back to case study selection

IIIT HYDERABAD · MASTER'S THESIS · VR · QUANTITATIVE RESEARCH

Psycho-motor Correlates of Affective Experience

For my IIIT thesis, I studied whether head-scanning behavior in virtual reality could add an objective signal to emotion assessment.

The Problem

Affective experience is subjective, and self-report alone does not always tell the full story.
We wanted to see whether head movement in VR could work as an objective measure of affective response.

IIIT Hyderabad thesis slide on psycho-motor correlates of affective experience

What I Did

Led a 6-member master's thesis team at IIIT Hyderabad.
Compared yaw-based head-scanning behavior with valence ratings using Spearman's rho correlation.

Participant wearing a VR headset during the affective experience research setup

What I Found

Head movement patterns added a behavioral layer to how people experienced VR content.
The data helped connect emotion assessment with objective measures, not only what users said.

Outcome

Found significant positive correlations with valence: yaw SD r(118)=0.223, p=0.015; yaw scanning speed r(118)=0.27, p=0.03.
This supported head-scanning behavior as a measurable signal for affective experience in VR.

Figure showing significant correlations between valence and yaw-based head-scanning measures

Back to case study selection

RESEARCHOPS & AI WORKFLOWS

Systems I built to make research move faster.

Process documentation and AI-assisted workflows I created to reduce coordination effort, standardize execution, and make interview evidence easier to reuse.

Participant scheduling

Shared Bookings calendar setup

Documented a repeatable Microsoft Bookings process for shared research calendars, interviewer availability, timezone handling, reminders, buffers, and booking rules.

View Microsoft Bookings setup guide

Scaled outreach

Personalized mail merge workflow

Created a no-code Outlook Mail Merge guide to help teams send personalized research emails at scale using Excel fields, Word templates, and Outlook delivery.

View Outlook Mail Merge workflow

AI-assisted synthesis

Interview verbatim extractor agent

Defined an AI agent workflow for extracting verbatims around key research questions from user interviews, helping teams move from raw transcripts to question-level evidence faster.

View verbatim extractor agent brief

AWARDS & RECOGNITIONS

Awards and recognitions.

Four Clarivate recognitions I received for prompt innovation, customer impact, automation, and volunteering.

Prompt-A-Thon Silver Medal certificate for Minaxi Goel

August 2026

Silver Prize in Prompt-A-Thon

Recognized among 328 company-wide entries for a practical Microsoft Copilot prompt that helps find information scattered across emails, meetings, chats, files, and presentations.