Designing for
Human Complexity

Overview

Bachat Mitra — "Savings Friend" — was a digital savings app designed for 10 million+ rural women in India's National Rural Livelihood Mission self-help groups. The core challenge: how do you build trust in a digital financial product among semi-literate populations with no prior digital finance experience?

Research Approach

1. Conducted 60+ structured interviews and diary-style observations to understand existing savings rituals, social trust networks, and barriers to digital adoption.
2. Designed Qualtrics surveys with adaptive branching logic (N=600+) to measure trust drivers — reliability, social proof, transparency — and segment users by digital readiness.
3. Ran A/B tests on interface prototypes varying onboarding framing, visual metaphors, and voice-guided features to optimize adoption intent.
4. Built a behavioral trust model linking interface attributes to adoption likelihood, validated through multilevel regression across 3 regional cohorts.

Overview

AmTrue India sells biodegradable menstrual cups and intimate hygiene products directly online — with a bold mission to shift Indian women from disposable pads to reusable care. The problem was invisible in the analytics: traffic was healthy, ratings were strong, yet Q-Cup conversion sat below 2% and cart abandonment hit 78%. Standard funnel metrics couldn't explain why. The answer required getting inside the lived experience of the shopper — including the shame and uncertainty embedded in the purchase journey itself.

Research Approach

1. Behavioral analytics audit of 4 months of session data, scroll-depth heatmaps, and click-path logs — locating 63% of abandonment on the product detail page itself, before add-to-cart.
2. 24 moderated remote usability sessions (RITE methodology, two rounds) with women aged 18–38 across Tier 1 and 2 cities — think-aloud protocol paired with retrospective probing on hesitation moments.
3. Qualtrics XM survey (N=320) with adaptive branching logic measuring product comprehension, brand trust, body anxiety, and price sensitivity; multilevel regression (R/lme4) identified unique predictors of purchase intent.
4. K-means cluster analysis synthesized with usability patterns in NVivo produced 3 stable purchase-hesitancy profiles — Body Anxious, Eco-Rational, and First-Timer Overwhelmed — each requiring distinct design interventions.

Overview

EkaScribe is an AI-powered clinical documentation tool built by Eka Care that transcribes doctor-patient conversations in real time, auto-generating structured SOAP notes. Despite strong early adoption signals, retention dropped steeply after the first week in Tier 2–3 city clinics. Eka Care needed to understand whether this was a product problem, a trust problem, or a workflow mismatch — and how to fix it before national rollout.

Research Approach

1. Contextual inquiry across 22 semi-urban clinics (Tier 2–3: Indore, Nagpur, Coimbatore, Bhopal) — live consultation observations mapping interruptions, workarounds, and EHR friction. Followed with a 10-day WhatsApp diary study (n=18 clinicians) capturing daily moments of hesitation and tool avoidance.
2. 28 semi-structured in-depth interviews (30–45 min) with early dropouts, power users, and clinic support staff. Probed cognitive load of real-time AI transcription, regional accent accuracy anxiety, and absence of an inline correction/override flow. Key insight: clinicians left because they couldn’t verify the AI was working, not because it wasn’t.
3. Qualtrics XM post-onboarding survey (N=340, adaptive branching) deployed within 72 hours of each clinician’s onboarding session. Factor analysis isolated 3 dominant drop-off predictors: accent & dialect mismatch anxiety, no real-time correction affordance, and data privacy uncertainty.
4. Usage analytics audit across 6 months of session logs (1,200 clinicians) mapping abandonment events, session-length distributions, and feature engagement sequences. Behavioral data and interview themes converged on the same two failure moments: first AI summary preview and post-session transcript review.

Overview

This three-manuscript dissertation investigated a central paradox in global health: despite significant national progress, stark subnational disparities in maternal and child health persist. Bihar, India served as the research site. The Cultural Ecology of Health framework was developed to explain what data alone cannot — integrating community health worker behavior, traditional knowledge systems, and structural determinants.

Research Approach

1. 12+ months of embedded ethnographic fieldwork across rural Bihar — in-home observations, contextual interviews, and participant observation in health facilities and traditional healing contexts.
2. Paper 1 applied multilevel modeling to examine how ASHAs' own maternal health experiences predicted their clients' perinatal behaviors.
3. Paper 2 triangulated interview data with survey responses (N=820) to identify how cultural beliefs shape antenatal care uptake.
4. Paper 3 synthesized findings into a comparative framework across 3 districts, identifying community-level moderators of health worker effectiveness.

Overview

A two-year Gates Foundation-funded design research initiative with Dalberg Design and Project Concern International India. The project challenged a fundamental assumption: that men in rural India are disengaged from reproductive and maternal health. Through deep ethnographic immersion and co-creation, we discovered rich — but suppressed — male motivation and designed an intervention ecosystem to channel it.

Research Approach

1. 92 contextual inquiries and in-home observations across 5 sites, shadowing couples during antenatal visits, delivery, and postpartum care to map real decision-making dynamics.
2. Led 14 co-design workshops with men, women, ASHAs, and Dais — surfacing unspoken male motivations, social norms barriers, and existing workarounds.
3. Journey mapping across 40+ households revealed 3 distinct male engagement typologies (Protector, Provider, Bystander), each requiring different behavioral nudges.
4. Rapid prototyping of 5 intervention concepts across 3 iterations, validated through concept testing with 200+ community members.

Overview

India's health system invested $3.6 billion deploying 1 million ASHAs while simultaneously marginalizing Dais — traditional birth attendants who hold deep community trust and centuries of ecological knowledge. Rather than treating this as a conflict to resolve, this study revealed a sophisticated ecosystem of complementary expertise.

Research Approach

1. 75 in-depth semi-structured interviews (45 ASHAs, 30 Dais) across rural Bihar, using grounded theory to build an empirical model of how each group conceptualizes their role and expertise boundaries.
2. Developed a comparative framework across 6 dimensions revealing systematic complementarity — not conflict — between the two knowledge systems.
3. Triangulated interview analysis with 3 months of observation across 12 health facilities, validating framework through member-checking with 15 participants.
4. Co-analyzed findings with a multidisciplinary team to produce actionable integration recommendations.