Adnan A Syed · New York, NY
Senior UX researcher, mixed methods.
I also design the AI that makes research defensible. Eight years across Verizon, Charter, Meta, Ford, and Google.
Where I've worked
- Verizon
- Charter
- Meta
- Ford
Start here
4 case studies are open to read. The rest are behind a password, which I share with people I'm talking to. Their cards show a lock.
Request accessFlagship work
Finding the Problems That Repeat Across Nine Verizon Purchase Flows
A seven-reviewer heuristic evaluation of nine purchase flows, sorted with help from AI, that showed which usability problems repeat and gave teams one shared list to work from.
Password required →Solving Intermittent Connectivity for Millions of Spectrum Customers
An end-to-end service audit showed that siloed backend systems were behind intermittent connectivity. The journey maps and business case that followed secured $5M+ for a new AI framework.
Password required →Reading What Verizon Customers Meant, Then Telling Each Team What It Means for Them
Satisfaction scores say how a customer felt, not what they came to do or what got in the way. I built a way to read each session for both, tested it against labels from experts, and shaped what it finds for the teams who act on it.
Password required →Giving Verizon's Menus a First Look With AI Agents, Scored by Code
Before anyone is recruited, AI agents walk a menu, code scores what they did, and a researcher makes the calls. It got sharper every sprint, and the checks caught what the agents got wrong.
Password required →Improving Usability for Google's Query Profiler Launch
Rounds of usability testing on a rebuilt internal query tool, ending in a launch study that scored it 79.7 on the System Usability Scale, ahead of three other tools on the same scorecard.
Password required →How Small Businesses Build Ads, and What They Wanted Most From Meta's Ad Tools
Seven interviews showed how small businesses actually create ads in Meta Business Suite, and a survey of 98 advertisers ranked what they wanted built. Together they gave the Core Ads team a roadmap order.
Password required →Highlights
Writing the Standard for How a Research Organization Adopts AI
AI was arriving in research work fast. I wrote down what it takes to do that without producing confident answers nobody can trace, as a standard with gates.
Password required →Putting a Quick Concept in Front of Customers to See Whether It Fixed What the Heuristic Evaluation Found
The heuristic evaluation said where the plan-selection flow was failing. This study tested a quickly built concept against the live site with five customers, to see how they moved through it and what they made of the fixes.
Password required →Counting How Many People a Research Team Reached in a Year, and Showing the Work
The files held no single count of the people Verizon's research function reached in 2025. I read 582 reports, built one row per study, and produced a number with a written trail behind it.
Password required →Two Small Tools That Put a Research Team's Standards and Vocabulary to Work
A guided survey assistant that applies a team's standards to every question, and a pipeline that mined 1,036 past reports for the acronyms and term groups that search kept missing. A person reviews both.
Password required →Measuring Documentation Quality for Google's Engineering Tools
A recurring survey program that measured documentation quality across ten internal engineering tools, then checked whether the fixes moved the ratings.
Password required →Giving Three Connectivity Problems One Shared Set of Statuses
Several Charter programs each used AI, machine learning or a decision engine to spot connection trouble, and each described it differently. A shared status vocabulary gave them one language, and engineering and product adopted it.
Password required →Structuring Navigation Labels for Four Google Products
Four card sorts on four very different products, from an internal dashboard's chart settings to YouTube's content naming, each used to decide how things should be grouped and what they should be called.
Password required →Why Voters Weren't Reaching the Voter Guide in the Spectrum News App
Most people who tried to open the Voter Guide left at the address request. The team blamed privacy. Eight usability sessions pointed somewhere else, and a live test followed.
Understanding Trust-Driven Conversion for Meta Messaging Advertisers
For small businesses selling through Messenger and Instagram DMs, a sale is a feeling of trust built message by message.
Improving Alert Setup Forms for Google Engineers
A baseline usability study on Plx Alerts, Google's internal alerting tool, that showed which form fields were costing engineers the most time and confidence before general availability.
Benchmarking Cross-Platform Ad Tools for Meta SMB Advertisers
A cross-platform study into how small businesses actually juggle Meta, Google, and TikTok ad tools side by side, and where Meta's own experience fell short by comparison.
Password required →Evaluating Self-Service Activation for Charter Mobile Customers
A recurring monthly usability program that caught why customers were skipping past a self-service mobile activation flow's own guidance, and the team changed the flow before it launched.
Password required →Mapping Billing Surprises for Charter Bundle Customers
A journey-mapping study that traced why customers who bundled streaming or add-on services saw their charges split across bills with no shared logic, and why a promotional discount could end without a clear warning.
Password required →Earlier work
Redesigning Site Navigation for UT Dallas Students and Faculty
A university website has to serve prospective students and current ones through the same navigation. I ran the research and designed a structure that tested well with 27 participants.
Validating Sound as a Data Interface for High-Stakes Monitoring Teams
Could sound convey a dataset as reliably as a chart does? A set of training modules put that question to the test, and the accuracy numbers said yes.
Password required →Identifying Acoustic Collision Points for Accented Speech Recognition
Voice assistants routinely mishear accented English, and not randomly. This study found the specific, predictable acoustic collision points behind that and designed around them.
Password required →Designing a Trustworthy Chatbot for Sensitive Health Topics
A chatbot for sexual health and other taboo topics, where the main design problem was whether anyone would trust it enough to ask the question at all.
Password required →