Why Products Stall
Most products don’t fail because the technology is weak.
They stall because the human layer goes misunderstood.
Features ship before teams understand how users interpret them.
AI outputs are technically accurate but emotionally destabilizing.
Collaboration tools assume workflows instead of observing them.
Activation feels unclear, so value takes too long to surface.
Metrics measure activity, but not trust, confidence, retention, or internal advocacy.
Community spaces exist, but advocacy never forms or no internal champion emerges.
Growth compounds when people feel confident, understood, and willing to bring others in.
I work at that layer: where interpretation becomes adoption, and adoption becomes advocacy.
What I Do
I’ve spent 15 years leading research in high-stakes domains like higher education, healthcare systems, financial platforms, and civic tech, where decisions carry weighty consequences. Today I apply that experience to teams building AI products, B2B SaaS platforms, and complex systems.
I'm the person you call when:
You’re building conversational AI and need to define how it should behave
You’re launching AI features and can’t tell whether they create clarity or confusion, or whether users will trust them enough to rely on them
You’re making a significant product bet and need to understand what users need before you commit
Your product spans multiple stakeholders, workflows, or roles, and no one sees the whole system
You have research, but it’s not translating into clear product decisions
Standard UX methods don’t map cleanly to your constraints
“Doesn’t get any better than having one of our key internal stakeholders propose that we replicate the work. Clearly, the value was found. Game changing, actionable insights to drive growth!”
Selected Work
Designing a Trustworthy AI Dialogue Coach at Scale
Shaped product requirements from multi-source discovery research, then designed the quality evaluation approach for an AI dialogue coach: golden dataset methodology, synthetic test generation, adversarial stress testing, moderated usability studies, and diagnostic analysis of 286 AI feedback instances. Identified six distinct learner profiles that informed how tasks were framed and feedback was delivered. Led the work from pre-development through pilot launch.
Outcome: 7 activities shipped meeting 95%+ accuracy; error taxonomy designed for direct use in system prompts, redirecting engineering investment toward highest-impact fixes; 80% user satisfaction at pilot
AI Products · EdTech · Quality Evaluation · Trust & Safety
Connecting Three Service Siloes That Shared the Same Customer
Synthesized a year of prior research across three business units, interviewed home sellers, and facilitated service blueprinting workshops to map the full seller journey for a digital real estate platform. Delivered current-state and future-state blueprints with a validated five-phase journey framework.
Outcome: 10 service failure points identified; platform repositioned as a trusted resource earlier in the seller journey; cross-sell moments mapped to specific pain points between pathways
Real Estate · Service Design · Enterprise Systems · Journey Mapping
Improving Child Care Access for Eligible Families
Conducted discovery research for a digital platform connecting Detroit families with child care assistance. Led caregiver journey mapping, multilingual listening labs, and provider surveys to understand both sides of the marketplace and identify where the platform could deliver immediate value.
Outcome: Narrowed platform scope from comprehensive tool to focused provider discovery and eligibility screening — the two pain points where families needed the most help
Civic Tech · Discovery Research · Marketplace Design · Accessibility
Building a Coalition with Research & Shared Evidence
Mapped Detroit's collaborative workspace landscape, fielded a 770-person community survey, and developed neighborhood feasibility scoring to help a coalition of seven organizations align on a shared services strategy. Research revealed the client's role needed to shift from brand owner to operational partner.
Outcome: Seven organizations aligned around an evidence-based plan; research package supported a joint $12M funding ask; concept reframed from "coworking" to neighborhood workspace rooted in community needs
Community Development · Multi-Stakeholder Alignment · Strategic Research · Place-Based Design
“
Mary is a legit badass UX Researcher. Rarely do you have the privilege to work alongside someone who brings the heavy research and results that she does.
”
—Senior Designer
When to Call Me
I work at the layer where human interpretation determines whether products succeed. Teams bring me in when:
You’re building AI features and can’t tell whether they’re actually helping.
The model performs well, but users hesitate, override, or disengage. I surface the gap between technical accuracy and human confidence and translate that into design and product direction.
Adoption is slower than expected.
Usage metrics exist, but trust has not formed. I identify where clarity breaks down, where expectations misalign, and what is preventing users from relying on the system.
Your product has multiple stakeholders and competing realities.
Workflow-heavy systems often assume how people work instead of observing how they navigate constraints. I map lived behavior and surface friction points that stall momentum.
You have research, but no clear direction.
I synthesize fragmented insight into coherent strategic choices that product, design, and engineering can act on.
My work connects qualitative depth to roadmap decisions, especially in AI-enabled, regulated, and high-stakes environments where trust drives adoption.
How I Work
I typically work as a fractional research and product advisor, joining teams at the point where interpretation, trust, and adoption shape roadmap choices.
-
Ongoing engagement
When experience decisions influence roadmap, risk, activation, retention, or expansion, I work alongside product leadership as a senior partner.
This often includes:
Clarifying which questions actually require research
Translating research into product tradeoffs and prioritization
Advising on AI evaluation standards and quality thresholds
Identifying adoption risks before they surface as churn
Facilitating cross-functional alignment around user insight
You get senior research leadership without adding permanent headcount.
You also get continuity. I stay close enough to understand context and shape decisions over time. -
4–6 weeks
When adoption stalls, AI behavior feels uncertain, or teams lack clarity on what users are experiencing, we run a targeted diagnostic.
This includes stakeholder interviews, user conversations, and workflow analysis designed to surface the interpretation gaps driving friction.
You leave with:
A clear articulation of what is happening and why
Prioritized recommendations tied to roadmap
A decision framework your team can continue using
-
Project or embedded support
When AI features influence real decisions, evaluation must reflect lived use AND technical benchmarks.
I help teams design human-centered evaluation criteria, stress-test failure modes, and define quality standards that product and engineering can operationalize.
“Mary excels at drawing deep insight from any audience. She never loses sight of the broader business challenge. And, she does all this with lots of laughter.”
Sarah Westrom, Independent consultant