How do Indians decide who to trust with their money?
FintechResearchIndia
Context
India’s financial stack moved faster than its trust did. UPI made payments universal, but savings, credit, and insurance still run on relationships: agents, branch managers, and family WhatsApp groups.
Summary
I audited eight trust frameworks and found none built for India. So I synthesised a three-layer model of how trust forms, transfers, and breaks, then designed the study to test it: 15 one-hour interviews across three tiers of India.
Fieldwork is complete. The patterns are below.
An intellectual gap and a business need
Every BFSI project I’ve worked on raised the same question: what makes Indian users trust a product with their money? We kept borrowing Western frameworks. They never quite fit.
01Gap
Every major trust framework was built in Western, individualist contexts. None model India Stack, UPI, or person-based trust.
02Need
We build BFSI products for Indian users across tiers, and every engagement raised the same trust questions. We were guessing.
03Bet
A rigorous qualitative study could surface what surveys and borrowed frameworks miss: the actual mechanics of how trust forms, transfers, and breaks.
I mapped eight frameworks before building anything
I pulled every major trust model cited in BFSI and HCI research, mapped what each framework explains well, and identified what it can’t account for in the Indian context.
Across 30 years of trust research, each model explains one dimension well. None model all three at once, and none account for India’s trust infrastructure.
Covers
Institution-based trust. Structural assurances (encryption, insurance, regulation) and situational normality. Explains why users trust a category before evaluating a specific product.
Misses
Built for Western e-commerce. No concept of government-built infrastructure (UPI, Aadhaar) as a trust proxy. Doesn't model how India Stack participants inherit a trust floor that standalone products must build from scratch.
Covers
Four types of credibility: surface, presumed, reputed, earned. The Stanford study (6,500+ participants) found 46.1% of consumers assess credibility on visual design alone. Directly relevant to interface trust.
Misses
Web-era model. No account of how credibility works when the "interface" is a person (an LIC agent, a branch manager). Doesn't explain how trust earned in one product transfers to another via social networks.
Covers
The trust barrier: for new users, trust must be established before product usefulness even registers. Explains why onboarding anxiety blocks adoption.
Misses
Individualist framing. In India, the trust barrier is often bypassed entirely when a family member mandates a product. The salary account effect (employer-mandated bank accounts) skips the calculative entry phase.
Covers
The ABI model: Ability, Benevolence, Integrity as dimensions of trustworthiness. Clean framework for evaluating what breaks when trust fails.
Misses
Organisational trust model, not consumer. Doesn't distinguish between trust in a product and trust in the person who recommended it. Critical in collectivist cultures.
Covers
Trust transference process. Trust in a person transfers to the product they recommend. Explicitly addresses collectivist cultures where prediction and transference dominate over individual calculation.
Misses
Pre-digital. No account of how social trust operates through digital channels (finfluencers, app store ratings, WhatsApp forwards). The mechanisms have changed even if the principle holds.
Covers
Category-level trust benchmarks: banks at ~66%, crypto at 38%. Income-based trust gaps (high-income 12 points above low-income). Confirms financial services hit the "Trusted" threshold (60%+) for the first time since the Global Financial Crisis.
Misses
Survey data, not behavioural. Tells you what people say about trust, not what they do when trust is tested. No India-specific breakdown for BFSI sub-sectors. No tier-level analysis.
Covers
Reliability rated ~32% weight across service quality studies. E-S-QUAL breaks digital into efficiency, fulfilment, system availability, privacy. Directly maps to interface trust components.
Misses
Service quality is not trust. A product can score high on SERVQUAL and still not be trusted. Doesn't model trust formation, only service satisfaction.
Covers
84.3% trust fintech platforms across four Southeast Asian markets. In Singapore, 35.3% cite family and friend recommendations as the primary driver. Transparent fees at 69.6% as the second most important factor regionally.
Misses
Southeast Asian focus, not India-specific. Doesn't distinguish between tier-level trust patterns. No qualitative depth on how trust transfers within families or how non-digital trust relationships work in practice.
I combined what worked into three layers
01Institutional Trust
The system around the product, not the product itself. RBI regulation, UPI participation, DICGC coverage. In India, government-built infrastructure creates a trust floor that fintechs inherit simply by participating. This is the layer McKnight mapped but never applied to a context where the state builds the rails.
“Is the system this product operates within credible and safe?”
02Interface Trust
Trust formed through direct interaction. The UI, the transaction flow, the transparency of fees. Fogg's credibility research and Gefen's trust barrier live here. But "interface" means something different in Tier 3, where the LIC agent or the post office clerk is the interface, not a screen.
“Does this product itself feel trustworthy when I use it?”
03Social Trust
Trust transferred from people the user already trusts. Family, friends, colleagues, communities. Doney's transference process, amplified in a collectivist culture. 35.3% of users in the UnaFinancial study cited family recommendations as the primary reason they trust a fintech product. This layer often outweighs the other two for initial adoption.
“Do the people I trust, trust this product?”
Trust isn’t a switch. It moves through four phases, and each layer plays a different role in each one.
Some layers dominate early, some fade to background, and all of them re-activate when something breaks. The matrix maps the weight of each layer at each phase.
| 01Calculative Entry | 02Vulnerability Testing | 03Habitual Confidence | 04Rupture & Reckoning | |
|---|---|---|---|---|
| Institutional | DominatesPrimary gate | TestedStructural assurance | FadesUnconscious filter | Re-activatesSystem failure |
| Interface | ActiveSurface credibility | DominatesFirst transaction | EarnedAutopilot trust | ViolatedCompetence failure |
| Social | DominatesDiscovery trigger | Safety netPeer reassurance | ReinforcesPeer validation | Blame shiftsRecommender blamed |
Phase names follow the trust lifecycle: calculative entry, vulnerability testing, habitual confidence, rupture and reckoning.
15 participants across three tiers
Tier matters more than age or income. A salaried 30-year-old in Mumbai and one in a taluka town have the same demographics but completely different trust architectures.
4Tier 1
Metro
6Tier 2
Emerging Cities
5Tier 3
Semi-urban / Rural
| Dimension | Tier 1 (4) | Tier 2 (6) | Tier 3 (5) | Total |
|---|---|---|---|---|
| Women | 2 | 3 | 2 | 7 |
| Men | 2 | 3 | 3 | 8 |
| Age 20-30 | 2 | 2 | 2 | 6 |
| Age 30-40 | 1 | 2 | 1 | 4 |
| Age 40+ | 1 | 2 | 2 | 5 |
| Digital-first | 3 | 2 | · | 5 |
| Mixed | 1 | 3 | 2 | 6 |
| Non-digital | · | 1 | 3 | 4 |
| Total | 4 | 6 | 5 | 15 |
The recruitment matrix, balanced across gender, age, and digital comfort within each tier.
Every question maps to a cell in the trust matrix
10screener questions, gating who qualified
25interview questions across 8 modules
8modules, each mapped to a trust matrix cell
60minutes per session, semi-structured
Want the screener, the discussion guide, or the rest of the research instruments? Email hello@shankar.design.
What the interviews revealed
15 participants, 50 validated research clusters, 200+ evidence points. These are the patterns that held across tiers.
82%refuse to act on AI advice alone
Across every tier, participants would not act on AI financial advice without verifying it with a person they know.
55%trust faster behind a known brand
Trust builds measurably faster when a recognised brand backs the product, even when the product is new.
55%of Tier 3 would go digital with an agent
More than half of Tier 3 users would adopt a digital product if a trusted agent guided them through it.
27%of Tier 3 would try purely digital
Barely a quarter of Tier 3 users would try a product with no human anywhere in the loop.
What makes each tier try a new product is irrelevant to the others.
Tier 1
Interface
Tier 1 users check app store ratings, watch YouTube reviews, and ask tech-savvy friends before trying anything.
Tier 2
Social
Tier 2 users rely on recommendations from their CA, accountant, personal banker, or family members who already use it.
Tier 3
Relational
Tier 3 users need a trusted agent or clerk to recommend it, a nearby branch, and a family member who already uses the same agent.
Every tier starts small. How much they risk depends on who they trust to catch them if it fails.
Tier 1
Digital-first adopters in metros
₹5-10K
First transactions. Scales to ₹1-1.5L. Above ₹50K reverts to branch.
Tier 2
Peer-influenced users in emerging cities
₹5-10K
First transactions. Scales to ₹1-1.5L. Above ₹50K reverts to branch.
Tier 3
Non-digital users in semi-urban and rural areas
₹3-5K
Caps at ₹36K a year. Above ₹10K requires an agent.
Trust is not granted. It is earned through months of incident-free performance.
Tier 1
4-6 months
Cross-check digital activity against bank statements. Monitor transaction speed and consistency before increasing capital.
Tier 2
6-8 months
Consult a CA or accountant, verify with a personal banker. Wait for a trusted professional to confirm before scaling.
Tier 3
12+ months
Visit the branch in person, speak with the agent, observe family using it. Trust only builds through repeated face-to-face interaction.
Things that surprised us
Not every finding was expected. A few patterns cut against common assumptions about how trust works in financial products.
Complexity as security
A minority of participants interpreted a confusing interface as proof that it's secure. The assumption: if it's hard to use, it's harder to hack.
Self-blame over accountability
When fraud happens, users blame themselves, not the product. This internalisation of responsibility masks systemic product failures.
Permanent trust breakdown
Tier 1 users switch apps. Tier 2 and 3 users blacklist the brand and warn their entire network. Recovery is nearly impossible.
The regulation paradox
Users trust RBI regulation as a safety guarantee, yet simultaneously expect those same regulated institutions to obscure fees. Both beliefs coexist unresolved.
Zero social media trust
Not a single participant across any tier trusts financial advice from social media. Peer trust is strong, but only from known individuals.
"Nervous" is universal
Every new user across all tiers described their first digital deposit with this word. It appeared unprompted, and it reflects the real stakes of handing money to an unfamiliar system.
Study details
- Role
- Research Lead: framework design, literature audit, study design, participant criteria, discussion guide
- Method
- Semi-structured interviews, 60 minutes per session, generative and exploratory
- Platform
- poocho.co for recruitment and facilitation
- Conducted by
- NetBramha Studios and poocho.co
- Sample
- 15 participants across Metro, Emerging, and Semi-urban/rural India
- Status
- Complete. The full report is ready. Email hello@shankar.design to read it.
Next
01
Overhauling a 100-year-old bank’s mobile app
Fintech · Mobile App
Read case study →