VS

03. Research

6 layers of evidence — each one surfacing what the last one missed. Applied in loops, not in sequence

I had worked on the legacy LOS for nearly a year before this project started. Enough time to observe its failures directly — not from a brief, but from being inside it. When the merger created conditions for a new platform, that firsthand knowledge became the starting hypothesis. Research didn't create the problem list. It validated, deepened, and in several cases overturned it. The approach combined Non-Linear Design Thinking with Systems Thinking — looping between empathy, ideation, prototyping, and testing as findings demanded. What drove it was curiosity: every answer raised a new question, and following those questions back and forth between departments revealed how each team's behaviour was shaped by — and in turn shaped — every other.

3.1 — Heuristic Evaluation

Evaluation of the legacy LOS against Nielsen's 10 usability principles.

3 CRITICAL VIOLATIONS - DROVE THE DESIGN SYSTEM

H1 - VISIBILITY OF SYSTEM STATUS

No loader after submission. No feedback on button click. Users couldn't tell if an action had been registered or was still processing.

H4 - CONSISTENCY AND STANDARDS

Sizes, labels, colours, and spacing inconsistent across every module. Disabled buttons identical to active ones. Every screen felt independently designed — because it was.

H9 - HELP USERS RECOGNISE, DIAGNOSE AND RECOVER FROM ERRORS

Error messages were technical and unactionable. When third-party APIs (Perfios, Signzy, FIS) failed, users had no recovery path. 10.7% of all tickets were UX challenges — error recovery was the leading cause.

5 SUPPORTING VIOLATIONS - DROVE IA, & ERROR PREVENTION DECISIONS

H2 - MATCH BETWEEN SYSTEM AND THE REAL WORLD

Navigation reflected database architecture, not user workflows.

H5 - ERROR PREVENTION

9.1% of all tickets were unforced errors: incorrect input, missed fields, requests to re-upload. The system accepted invalid or incomplete data without flagging it at entry.

H6 - RECOGNITION RATHER THAN RECALL

Users navigated by memory. No landmarks, no workflow-based grouping. New employees relied entirely on colleagues — the system had no mechanism for independent discovery.

H8 - AESTHETIC & MINIMALISM

Every screen was independently designed — inconsistent spacing, competing visual weights, irrelevant information surfaced alongside critical fields. No visual hierarchy meant every element demanded equal attention.

H10 - HELP & DOCUMENTATION

3.1% of tickets were false alarms — policy unawareness, process unawareness, vendor process gaps. No contextual help existed inside the system. Sanya wrote detailed guidebooks that users ignored because guidance wasn't where confusion happened — inside the workflow.

UX Designer learning from a Credit Manager on the credit underwriting process

Image Description -Inconsistent button styles placed next to each other

Image Description - Missing Input Validation

The findings gave us named, principle-grounded evidence to anchor every design system decision.

3.2 — Competitor Reviews & Walkthroughs

App store reviews coded across 6 competitors. Hands-on walkthroughs of LendingKart, Indifi, Flexi Loans, Zip Loan, and Bajaj Finserv — benchmarked against onboarding, status communication, and error handling.

Image of the playstore reviws of our competitors

Image Description : Competitor Analysis of Our Competitors

Image Description : Overview of the customers review of our competitors

⚠️ 42% of all complaints — Technical Issues

Reliability was the market's primary failure, not lack of features

✅ Top praise — speed and status visibility

Praised for being predictable and fast — not feature-rich

👍 Progressive document collection

YONO SBI's multi-column, structured form approach was the clearest positive

👍 Existing Customers preferred National Banks

I already know what fast, transparent lending felt like — and chose us despite it

3.3 — Support Ticket Analysis

1,455 tickets across all the modules. Treated as a dataset — analysed for patterns, not handled as individual incidents. Surfaced what observation and interviews hadn't, because people had stopped naming these as problems.

Image Description : Screenshot showcasing our analysis of support tickets, categorizing issue themes and extracting key insights to inform design improvements

🏦 Bank Statement Upload

27.3% (397) of all tickets. Not one interview flagged it. The system's biggest failure, normalised out of sight. Lack of Error Diagnosis & Recovery for Users

❓ Case Stuck — cause "Unknown"

63% (158) of the stuck cases: No traceable reason whether the error was technical, human, or internal/external 3rd party plugins- invisibly stuck

🪪 Name / DOB / Gender Mismatches

The bad day story was pattern, not exception. Basic sanctity check was missing

📋 Application Status Visibility

37 Cases: "I have to call the RM again to know the current status" — in ticket form

3.4 — Ethnographic Research: Apprenticeship Model

This method surfaces what interviews can't: the workarounds people have normalised and stopped naming.

Following are some of the problem we experienced and observed in action — but never heard named in interviews because they had become part of the normal way of working:

CMs doing transaction tagging and obligation sheet work entirely in Excel — the LOS lacked sort, search, filter, and edit. Two complete parallel workflows built to compensate
CMs carrying notebooks to PD visits — writing questions before, transcribing after. Standard practice. Nobody named it as a problem
OPS handling tasks that belonged with the RM — accumulated in their bucket with no process behind it
Employees discussing top performers informally, but no self-service view of their own performance data
System slowness at month-end: 30–45 min page loads. Consistent across every branch

Image Description : UX Designer collaborating with a Credit Manager to understand the credit underwriting process.

We worked alongside people — sat with CMs through case queues, shadowed RMs in the field, observed Operations at month-end. Didn't ask what the problem was but watched what people actually doing their job

3.5 — Stakeholder Interviews

Structured interviews across every level — National/Department Heads of Credit, Sales, Ops, and Sanction. Real ticket scenarios and competitor examples used as prompts. People rarely volunteer problems they've normalised; the prompts bypassed that. Where ethnography showed what people did, interviews named why — the incentive structures, the KPIs, the organisational pressures that made workarounds rational. The bucket blame for TAT emerged here

"The case will show in our bucket as we are dependent on the external agency for the CPV/FCU report and Sales will blame us for the delay— even though it's not in our hands"

Every team optimising to move cases out of their bucket, not ensure quality within it or improve the overall process.

3.7 — Primary Customer Research

The internal research told us how the system failed its operators. This told us how our organization failed to serve our current & potential customers.

🧑‍💼 25

MSME Borrowers across Tier 1, 2, and 3 locations

🏪 92%

Sole Proprietors who were running it alone

⛔ 83%

Micro Business Loan abandoned

💰 82%

Having an annual turnover under ₹1 crore

📣 80%

Apply via word-of-mouth, not comparison

💻 35%

Carry a laptop to work or while commuting

I should know at the starting itself whether I am eligible or not. So that I can look for arranging my financial needs elsewhere (Informal Market). Open Quote

The eligibility-first design principle in one sentence

You invest a lot of time providing documents and answering every clarification required and wait for weeks togther to be informed that my application is rejected. Open Quote

Time invested → late rejection → zero trust

From Kotak it just took 5 days to get my last Business Loan disbursed. Open Quote

The competitive benchmark. Cited unprompted

Even if it gets debited it doesn't reflect — and I'm being charged late fees. Am I responsible? Open Quote

Post-disbursal failure with financial consequence

3.8 — How the findings were processed - 4 Steps applied after every research phase

3.8.1 — Root Cause Classification

Every finding classified before any solution was proposed:

01. UX Issue

Bank statement upload (27.3% of tickets). Looked like user error due to incorrect input format.
Design fixed it - Error Diagnosis, and Recovery guidance built into the interface — reduced bank statement upload tickets by 70% in the first month alone.

02. Regulatory Requirements

Physical signature for loan documents. Design could not remove it.
Solution: e-sign (digital equivalent) — accommodated the mandate while eliminating the travel.

03. Internal Dependency

TAT bucket blame. CMs blamed for delays caused by Sales. Not a UX problem
Required a structural process change - real-time TAT attribution per department in the dashboard.

04. External Dependency

Perfios transaction tagging (7.2% of tickets — External Server Errors/Delayed Responses).
Solution - Recovery-first design (CM flag interface) and a pattern-submission interface so CMs could raise ticket with the right person

05. Internal Policy

"AML as a task could be moved earlier in the loan processing for early rejection as it contains FIR, Suits, Cases, Fraud List, Terrorist, Political Affiliation details"

Management Standpoint - AML costs ₹400 per applicant; CIBIL + Perfios together cost less than half that. Running them before eligibility confirmation meant absorbing that cost on applications that would never qualify. Reclassified as a cost-driven policy constraint.

Solution: optimise pre-eligibility for speed, post-eligibility for thoroughness. The right design only became visible once the classification was correct.

How Did We Classify ? - The Answer - 5 WHY'S

Bank Statement Upload worked through. Sample Error Message:

"Problem with provider id B7RE1714633381759"

27.3% of all tickets. Initially assumed to be a Perfios vendor failure or user error. The 5 Whys overturned the assumed classification — each Why answered by cross-referencing research methods.

01 - WHY

Bank statement upload generates 397 tickets — the single largest category across all 36 modules
Ticket Analysis

02 - WHY

Users can't tell if their upload succeeded, failed, or is still processing — no progress indication, no feedback on completion
Usability testing + UX Challenges 10.7% tickets

03 - WHY

When Perfios returns an error, the message is technical and unactionable — users don't know what to fix, so they raise a ticket instead
Heuristic evaluation — H9, ticket data

04 - WHY

Nobody called it out. Users raised tickets and moved on. The tech team treated the volume as operational noise — not a signal. Design was never brought in to look at it. The vendor was never asked to improve their error responses. Each team absorbed the friction individually and the failure was never named as a problem worth solving
Ethnography: "I just raise a ticket when it doesn't work".

05 - WHY

The system had normalised its own failure. 27.3% of all tickets were coming from one module — and it had become background noise. It took ticket analysis to surface what daily use had made invisible
Ticket Analysis — Surfaced what interviews couldn't

CLASSIFICATION RESULT

UX Issue — Not a Perfios failure or a user error

Fully fixable by design once classified correctly. The vendor was not the root cause. The absence of failed states were.

WHAT THIS PRODUCED IN BLUEPRINT

Plain-language error messages replacing the technical Perfios codes

3rd party recovery patterns standardised across every external vendor failure

Progress Indicators so users always know where their upload stands

3.8.2 — Thematic Analysis

Each classified finding was assigned a code using a consistent DESCRIPTOR CONCEPT convention. Codes that appeared repeatedly across different data sources — interviews, tickets, ethnography, competitor reviews — were grouped into themes. Four themes emerged before any group session began.

QUOTE -> CODE

"Case stuck — cause unknown" (158 Tickets)→ MISSING CASE VISIBILITY

"I have to call the RM to know my status"→ MISSING STATUS UPDATES

"I carry a notebook to PD visits, transcribe late"→ MISSING FIELD CAPTURING TOOLS

→

THEME 1

No actor has a complete picture at any stage

QUOTE -> CODE

"Customer fails one product — I restart the entire application for the next"→ CROSS PRODUCT ELIGIBILITY MISSING

"When Sales take 5 days to revert, TAT shows in our bucket"→ MISALIGNED ACCOUNTABILITY MEASURE

"158 cases stuck — cause listed as unknown"→ MISSING OWNERSHIP HANDOFF

→

THEME 2

Nobody owns what happens between stages —
or when a stage fails

QUOTE -> CODE

"Perfios incorrect transaction tagging"→ VENDOR FAILURE UNHANDLED

"No recovery steps when 3rd party plugins fail"→ MISSING FAILURE RECOVERY PATH

"Written guides but people ask the same questions"→ MISSING CONTEXTUAL HELP

→

THEME 3

The system was built to succeed — not to fail gracefully

3.8.3 — Affinity Mapping

The themes were brought into cross-functional workshops. Each department placed their observations against the themes. Clusters emerged — note that Cluster 3 is fed by two themes combined, which is what affinity mapping reveals that individual analysis cannot: that two apparently separate problems share the same systemic root.

THEME + DEPARTMENT OBSERVATIONS

Theme 1 — Manual effort creates errors the system cannot catch

Sales: "I fill in the same customer data correctly — Credit still sends it back"

Credit: "Incomplete submissions arrive — I correct and return"

Ops: "I find the error at disbursal — too late to fix without restarting"

→

CLUSTER 1

Every department thought the errors were someone else's
fault

THEME + DEPARTMENT OBSERVATIONS

Theme 2 — No actor has a complete picture at any stage

Customer: "I call the RM to know where my application is"

RM: "I call the CM to find out why a case is stuck"

Support: "I track every query in a shared Excel — no system"

→

CLUSTER 2

The same question was being asked five times a day
across five different channels

THEME + DEPARTMENT OBSERVATIONS

Theme 3 — Nobody owns the transitions between stages

Theme 4 — The system was built to succeed, not to fail gracefully

Sales: "Customer fails one policy — I restart from scratch for the next product"

Credit: "The case came incomplete — I lose days correcting it"

Ops: "By the time it reaches me, the time is already gone"

→

CLUSTER 3

TAT wasn't lost in the steps — it was lost between them

These three clusters — not one per department, not one per method — became the design brief. Cluster 3 emerging from two separate themes shows that an accountability failure and a system resilience failure were both expressions of the same absent handoff design.

Image Description : Affinity Mapping Session

Clusters → Insights — the meaning extracted before HMW was written

CLUSTER 1

Every department thought the errors were someone else's fault

→

INSIGHT 1

The system accepted incorrect data without validation — making downstream blame rational and inevitable. Accountability without visibility is blame without resolution.

CLUSTER 2

The same question was asked five times a day across five different channels

→

INSIGHT 2

Status communication had become a distributed full-time job nobody had named, measured, or designed for.
The absence of a system answer created a human workaround at every level.

CLUSTER 3

TAT wasn't lost in the steps — it was lost between them

→

INSIGHT 3

Handoffs had no system. Every transition between departments was an unowned gap where time, accountability,
and context disappeared. Digitising the steps would have digitised the delay — not eliminated it.

Insights named the systemic truth. HMW statements then reframed each insight as a design question — without prescribing a solution. The insight is the constraint that makes the HMW specific enough to be solvable.

3.8.4 Interconnectedness Mapping

How does each department's failure shape every other's? A CM rushing to clear their bucket wasn't a Credit problem — it was Sales TAT measurement creating pressure that rippled downstream. Tracing those ripples until they connected made the systemic root visible.

WHAT THE 6 LAYERS PRODUCED TOGETHER

Ethnography showed what people did. Classification told us why. Interviews named the incentive failures. Tickets quantified what had been normalised. Competitor walkthroughs set the market bar. Customer research showed who bore the cost. Heuristic evaluation named the usability violations formally. No single method produced the full picture. The combination made the reframe in the next section not just defensible — but undeniable.

Separately, each finding pointed at a symptom. Together — first through thematic analysis that coded the patterns across all data sources, then through affinity mapping workshops where every department clustered and owned the same evidence — they pointed at the same root: a system with no mechanism to surface a failure before it became someone else's problem. Thematic analysis produced the patterns. Affinity mapping turned them into a shared diagnosis the whole organisation could own.

04. Reframe

The brief was three things. The research changed all three.

Management's direction was clear: reduce TAT to 7 days, replace weekly Excel reports with real-time dashboards, and build the platform as B2B SaaS. The research confirmed all three were the right destinations. But for TAT — the most critical of the three — it revealed that the assumed route would not work.

Once we take everything online, the TAT will come down automatically. That's what being a technology-driven company means — we don't need people chasing each other over phone calls and emails. The process does the work. Open Quote

CEO, Ashv Finance — Team meeting

It was a reasonable belief. Confident, visionary, and shared by almost everyone in the room. Then the research produced a moment that changed the conversation.

THE MOMENT THE RESEARCH STOPPED BEING DATA COLLECTION AND BECAME A DIAGNOSIS

It was the last working day of the month. The customer had agreed — 21% interest, 2-year tenure — but only if funds arrived before Friday(Last Working Day of the month) evening. Three customers, same commitment, each 50km apart. The RM needed physical signatures from all of them before cutoff. At the third stop, the customer caught it himself: the name on the application didn't match his ID. One field, wrong from the start. The case went back. The entire cycle restarted — Credit, Sanction, Operations. The RM drove back, got the corrected signature, walked in at 1:30pm. The disbursal didn't make Friday. Four checkpoints. Not one caught it. This happened every month in some form — because the system had no validation at entry, no cross-field checks, and no way to surface an error before it became a physical journey.

The name mismatch in that story was not an exception. The bad day story was a pattern, not an incident. Every failure had a direct design response.

Direction - Reduce TAT to 7 Days

Assumption : "TAT is 34 days because steps are offline. Digitise them and it comes down to 7"

Automating a broken process digitises the delay. It doesn't eliminate it. The restart points had to be removed first — and none of them would have been touched by digitisation alone.

The field research, ticket analysis, and stakeholder interviews surfaced 6 failures that explained precisely why the assumed path would not work.

📋 01 - PROCESS

Eligibility loop — hours wasted per customer

No product-switching logic. 3–4 full re-entries per customer. 2–3 hours each. Normalised — no interviewee named it as a problem.

🕸️ 02 - SYSTEM

No feedback loops — errors discovered late

No validation at entry, no cross-field checks. A name mismatch at step 1 invisible until step 10. Fixing it meant a full restart.

⚖️ 03 - ACCOUNTABILITY

No deviation workflow — every exception untracked

158 cases stuck, cause "Unknown." Approvals over email — no trail, no resolution path. TAT measured against the team holding the case, not causing the delay.

🏗️ 04 - ARCHITECTURE

Rigid architecture — nothing self-serve

Every change required engineering. RBI mandates became crises. B2B SaaS ambition structurally blocked.

🪛 05 - FRICTION

No design system — IA built around the system, not users

No shared components, inconsistent states, navigation by memory. Every screen independently designed.

🧠 06 - HUMAN

No guidance, no recovery — users were on their own

Steps, rules, exceptions — all carried in memory. No in-system guidance, no recovery path. Every case, every day, from scratch.

" This wasn't a usability problem. It was a system whose parts had no memory of each other "

What Changed as a Result ?

SCOPE

UI redesign → full platform rebuild. Restart points became primary targets. Accountability layer added: FTR, deviation workflows, in-platform approvals, real-time TAT attribution.

INTENT

Not just faster TAT — but the same people, same hours, processing more cases with greater accuracy. Efficiency in the backend, not just speed at the surface.

One reframe. One question that followed every subsequent decision: does this change the system's behaviour — or just the surface of one of its parts?

5. Personas, Journey Mapping & HMW

5 People. 4 Departments. 1 Broken System

We created personas by combining qualitative insights (ethnographic studies, interviews) with quantitative data (support tickets, customer reviews) to define user needs and workflows. Using journey maps, we uncovered opportunities to enhance transparency, automation, and process efficiency, resulting in streamlined operations, reduced manual interventions, & an improved experience for all.

Kredible Kiran

Credit Manager

I have to carefully read the body language of the customers and his intent to pay irrespective of the situation quote

Resilient Ravi

Relationship Manager

I have to be on the field most of the day looking for new leads while handling existing customers & reach the monthly target. quote

Diligent Deepa

Operations Executive

It would be great if the process and the system could somehow prevent me from working late nights during the month end quote

Savvy Sanya

Senior Support Executive

I’ve written everything in emails and guides, but people still come back asking the same questions repeatedly! quote

Veteran Vijay

Wholesale Trader & Business Owner

My previous business loan from Axis was disbursed in 5-6 days—that’s the kind of service I expect! quote

6. Blueprint

How the platform was built — and why in this order

6.1 - How we actually worked - a repeating quarterly loop

There was no fixed phase plan. Every quarter ran the same cycle:

PRIORITISE

Dept. heads + C-suite inputs → MoSCoW → RBI overrides if active

INVESTIGATE

Ethnographic deep dive — UX issue, policy change, or dependency?

SYSTEMS MAP

Map causes and cross-dept. impact before proposing anything

CO-DESIGN

All stakeholders in the room — sign-off before any prototype

TEST & PILOT

Select branches, 1 month — fix UX + tech issues before scale

RELEASE & MONITOR

National rollout after pilot sign-off. Loop restarts next quarter.

With the personas validated and the reframe accepted, one principle governed what came next: understand the system before designing any part of it. Nothing was committed to engineering without co-ownership from the people who would use it.

6.2 - Design System

Shared components, standardised interaction states, single source of truth for engineering. Three decisions required platform-wide pattern enforcement — which is what the design system made possible.

Plain-language errors 10.7% of tickets were unreadable API error states. Without a design system, each developer handled each vendor failure differently across 36 modules. One pattern — plain cause, specific action, no jargon — applied everywhere simultaneously.

3rd Party Error Recovery Patterns Perfios, Signzy, FIS, Karza each had different failure modes. The design system standardised one recovery pattern per dependency — what the screen shows, what the user is told, what action they can take. Consistent across every external failure, not decided screen by screen.

Contextual Onboarding 9.2% of unforced errors came from legacy system habits. A one-time tutorial would have been skipped. The design system built onboarding as a layer — appearing at the moment of confusion on any screen that needed it, not as a one-off feature in a single module.

Image Description - Typography section showing Plus Jakarta Sans as the chosen typeface, selected for legibility and trust in financial interfaces, with a defined font scale (12–32px), line height ratios, and a two-weight principle (Regular 400, Semi Bold 600) that kept the interface structured without visual noise.

Image Description - Tables are molecules — search bar, filter, pagination, and bulk action assembled into one reusable component. Every data-heavy screen in the DLP used this molecule as its base. Change the component once, update everywhere.

Accessibility
Some of the prominent principles used in our website are listed below
01. WCAG AA contrast
02. 44px tap targets
03. Colour paired with labels
04. Font size adjustment
05. Theme Selection for the PWA for the RMs to use in the harsh field sunlight.

Outcomes
01. Fewer engineering bugs
02. Faster Dev cycles
03. Direct component reuse without new HTML
04. WhenAadhaar masking (2022) and digital consent (2023) arrived mid-project, both were absorbed in weeks — not months.
05. Ashv Gati launched as a new product, it was live in under two weeks — the platform's configurability meant no engineering tickets, and the design system meant no new components

6.3 - Information Architecture - Built for How Users Think

The legacy navigation was built around system architecture, not user tasks. Users navigated by memory. The IA rebuild started from user mental models.

Image Description - A snapshot from the open card sorting session

Two methods in sequence:
Open card sorting (5–8 members) produced the groupings.
Tree testing (30 members, no UI) validated them under task pressure.
Every new module went through this before release.

Concrete example: card sorting placed the Deviation Workflow under "Case Actions" — not "Approvals" where the system had put it. Tree testing confirmed: 80%+ first-click accuracy on "Case Actions", under 30% on "Approvals." The module shipped in the right place. Repeated across every module over four years, this prevented navigation debt from accumulating.

Image Description - IA of the legacy system. User had to click on each card to look for a specific sub task which were not placed in a logical hierarchy

Image Description - Recommended sample IA for the new DLP system

6.4 - Co-Creation — Built Over Four Years, Not One Session

Eight half-yearly cycles. Each cycle: review what was delivered, measure what changed, co-create the next six months with every department in the room.

From Clusters to How Might We - Closing the Loop from Research to Design

The clusters produced insights. The insights fuelled the HMW statements. The insight is what makes an HMW specific enough to be solvable — without it, HMW questions become generic.

Cluster 1

Every department thought the errors were someone else's fault

→

Insight 1

Accountability without visibility is blame without resolution

→

HMW 1

How might we prevent errors from entering the system
in the first place — and surface them immediately when
they do?

Cluster 2

The same question was asked five times a day across five different channels

→

Insight 2

The absence of a system answer created a human workaround at every level

→

HMW 2

How might we give every actor — customer, RM, CM, Ops,
& Support — a real-time view of exactly where a case
stands, without anyone having to ask?do?

Cluster 3

TAT wasn't lost in the steps — it was lost between them

→

Insight 3

Digitising the steps would have digitised the delay — the handoffs needed to be owned

→

HMW 3

How might we make every handoff between departments
fully traceable — so TAT can be attributed to where the
delay actually happened, not where the case happened to
be held?

Each session: journey map opened the room → HMW reframed pain as a design question → dot-voting surfaced what teams believed in → roadmap agreed before anyone left. SMS/email milestone alerts received 9 votes — the highest in the programme. Nobody could claim it was the design team's preference.

Image Description - Snapshot from the Co-Creation Workshop conducted with the departments Heads

The CIBIL First Debate Sales wanted to pull CIBIL using PAN and name only — before collecting full documents — to qualify customers faster and reduce wasted field visits. Credit refused: without complete documents, the CIBIL pull was context-free and increased the risk of misassessment. Neither position was wrong. The resolution was progressive validation gates — CIBIL pulled at the right stage once minimum qualifying data was confirmed, giving Sales speed without giving Credit incomplete cases. Neither team got everything they asked for. The system got what it needed.

The Video PD Proposal Credit Managers in the field surfaced a competitor doing Video Personal Discussions for loans below ₹5 lakhs — eliminating the physical visit entirely. The opportunity was significant: RMs were travelling 50km+ for PDs on small-ticket loans, consuming time disproportionate to the loan value. The Risk team raised fraud concerns in the same room — could identity and intent be verified reliably without a physical meeting? The debate ran across two sessions, with the Risk team reviewing other competitors practises and data before approving. Signed off in a month — a decision that would have taken a quarter through a traditional approval chain, and one that would not have surfaced at all without CMs and Risk in the same co-creation room.

6.5 - As Is -> To Be : What the Process Became

AS IS - Thirteen largely offline steps became a fully digital workflow across four departments.

Mapping the Current Journey i.e. As Is Process

Image Description - The As-Is loan origination process — 13 stages across five actors (Customer, DSA, RM, Credit Manager, Ops, Treasury), with multiple offline handoffs, policy decision points, and manual workarounds at every stage.

TO BE - A Near Digital Process for a modern B2B SaaS platform. The To-Be process — fully digital, with real-time visibility, built-in guidance, and accountability at every handoff.

Mapping the Future Journey i.e. To Be Process

Image Description - The To-Be digital workflow — the same loan origination process rebuilt across six role types with automated triggers, in-system approvals, digital consent, e-sign, push notifications, and structured deviation paths replacing every offline handoff.

6.6 - Mobile App - Designed for the Field

Three independent sources pointed at the same decision: field research (4 CMs, 4 cities named the gap unprompted), co-design votes (highest priority in HMW 1), and customer data (91% of borrowers already used GPay, PhonePe, or Paytm).

Customers were mobile-first. The platform wasn't. The app gave RMs everything in the field: PD notes captured live, documents uploaded on-site, case status checked between visits. No more driving back to complete what started in the field.

6.6 - Executive Dashboard - From Weekly Excel to Real Time Data

Image Description - Summary Dashoboard for the CEO

Management was receiving weekly Excel reports pulled from the backend. The ask was real-time. My role here was different — no reframe needed. I worked directly with the CEO, COO, and Risk Head to understand what each report needed to reveal, iterated on their direct feedback, and sought sign-off at each stage.

The NPA finding — Express Loan at 12.4% , traced to a specific segment and region — surfaced in the first review cycle. It would have taken days of manual analysis without the dashboard.

6.7 - Usability Testing - Validating the Field Experience

Every module went the same process

Co-Creation

Hi-Fi Wireframes

Usability Testing

Management Review

PILOT (1 Month)

PAN India Release

Low-fi in co-creation: Signals nothing is fixed — stakeholders give flow feedback, not polish feedback.

Hi-fi in testing: flows already validated by co-creation, so testing was about execution. The design system made hi-fi production cost low — screens assembled from existing components, not built from scratch. Think-aloud protocol mitigated the polish effect.

Management review was a gate, not a formality. When changes were significant, the cycle restarted. Nothing moved to pilot without both users and management signed off.

6.8 - Prototypes

📱

RM Mobile App

Field Sourcing to Submission

🎛️

Product Configuration

Onboarding to Deploying Products

📊

Executive Reports

Performance, NPA, Pipeline

📱

Customer PWA

Status, Repayment, Top-Up

📋

CM Underwriting

Review to Recommendation

🏦

Operations Flow

Case Verification to Disbursal

7. Decisions

The hard calls, not the frameworks

Every decision here traces back to specific evidence — research findings, usage data, or field observation

7.1 - In What Order - E-Sign Vs Pre-Disbursal Verification

⚖️ WHAT WE BUILT FIRST - AND WHY?

Pre-disbursal form verification with auto-population. The existing process required manual form fill, physical signature, scan, upload, and Ops head verification. End-to-end: 5–6 hours. 28% of cases were being processed repetitively due to upstream errors the system made invisible.

" A digitally signed incorrect application is a faster failure — not a better experience. Fix accuracy first. Then accelerate."

⏳ WHAT WE DELAYED - WITH A COMMITMENT?

E-sign was the most visible signal of a digital-first platform — commercially rational for B2B SaaS investors and customers alike. Management's case was strong. We didn't dismiss it.

" E-sign was sequenced as Q+1 priority #1 — not killed. When it shipped, it landed on a process that actually worked. The digital-first signal management wanted arrived on a stable foundation."

7.2 - What We Removed - 2 Features the Data made Indefensible

🚫 REFERRAL FEATURE KILLED

220 applications over 6 months through the referral programme. 2 met eligibility criteria — and both were rejected. Conversion rate: 0%.

The informal referral network already existed — 80% of MSME borrowers came through word of mouth anyway. The feature formalised a behaviour that was already happening, added engineering overhead, and produced nothing.

🚫BHARAT SETU INTEGRATION — KILLED

EMI payment via Bharat Setu was retained — an effective fallback when E-NACH failed due to bank defaults. What was removed: utility payments (electricity, gas, bills) added as a delight feature.

Fewer than 10 customers from 3,000+ ever used it. A lending app is not a bills app. Removed to simplify and cut maintenance overhead.

Both removed because the data made the case unambiguous. The ability to remove features is as important as the ability to build them

7.3 - What Research Found : The Feature Nobody Asked For

During field sessions, employees were already discussing top performers in casual conversation — who had the most disbursals, which branch was leading. Monthly performance emails were passive, delayed, and ignored. CMs had no self-service view of their own performance — tracking their numbers in personal MIS Excels.

Since I am a CM,I don't have any tool or place from which I can calculate my performance or how many manager assesses my performance. I am dependent on my manager's quarterly reviews. Open Quote

Satish Ramachandraiaha, Bengaluru

🎨 WHAT WE DESIGNED

Real-time leaderboards — top 5 PAN India per department, updated live. If employees were already talking about who was top, the platform should make that visible, real-time, and earned.

🔀 THE DESIGN TENSION - HOW WE RESOLVED IT

Volume-only rankings would reward Ravi for closing cases fast — even if Deepa absorbed the errors downstream.
The fix: the Executive Dashboard surfaces average TAT per employee alongside disbursal volume. Speed is visible. So is the cost of cutting corners. The leaderboard drives the behaviour. The dashboard holds it accountable.

The mobile app, real-time leaderboard, deferral marking digitisation, and deviation workflow automation were not on the original brief. All four came directly from field research and co-design sessions. The users who needed them most had never been asked — and wouldn't have known to ask.

08. Leadership

How the work was led — not just what was built

8.1 - Evidence before argument

The reframe that changed the brief — and how it was won — is documented in full in the Reframe section. The principle it demonstrated applies to every significant push-back in this project: come with the data that makes the cost of inaction visible. Not the UX cost. The cost to revenue, to targets, and to the people whose incentives are at stake.

8.2 - Designing the conditions for good evidence

The most impactful leadership move wasn't a design decision — it was structural. Research findings weren't presented to department heads — they were given to them to use before their quarterly meetings. Journey maps, ticket patterns, and field observations became the evidence they used to argue for investment, process changes, and policy decisions. One example: the National Credit Head used CM journey mapping data to make the case for a new pre-assessment process to the CEO — without the design team in the room. Design stopped being the team that executes after decisions are made. It became the team whose evidence made decisions possible.

8.3 - Building the team - and what it built in me

WHAT I BUILT FOR THE TEAM?

01. Standardized research templates for ethnography, card sorting, tree sorting, and presentations - sessions could run without me

02. Broke department silos — replaced requirements documentation with joint ideation sessions where teams co-owned what was being built

03. Built design system with documented usage principles — component decisions not reinvented every sprint

04. A co-creation format the team could facilitate independently

WHAT THE TEAM BUILT IN ME?

01. Giving ownership before someone feels fully ready — and trusting the scaffolding to hold them

02. Shifting from doing the work to creating conditions for the work to be done well

03. Knowing when to step back — the hardest part of leading a small team

By year four, the Junior Designer owned modules end to end. The Senior Designer ran stakeholder sessions independently. Neither was true at the start. Both required deliberately creating space — and resisting the instinct to stay involved longer than necessary.

09. Outcomes

Measurable impact — including what didn't land

📈 59%

REDUCTION IN TURN AROUND TIME

34 days → 14 days. Measured across 175 Disbursed Loans in the last quarter

📋 3.2 -> 0.74

TICKETS PER CASE PROCESSED

Pre DLP - 3.2 tickets per application across 1,290 applications over 6 months

Post DLP - Last Quarter - 129 tickets across 311 cases processed = 0.74 tickets per case

⭐ 3.9 / 5

SATISFACTION SCORE

Up from a baseline of 2.6

Half-yearly evaluation across 100 respondents — internal employees (CMs, RMs, Ops, Support) and external lending partners.

Method: structured questionnaire administered at the end of each 6-month performance cycle, covering system usability, task completion confidence, and overall experience.

📈 15%

SUPPORT TEAM PRODUCTIVITY

Freed approx. 10 hours of team capacity

Team of 4 handling three request categories: DLP support tickets, platform configuration requests, and customer queries.

10 hours calculated at ~4.5 min average reduction per ticket (15% of ~30 min average resolution time) across 129 tickets. Per-task efficiency — not volume increase.

What We Didn't Achieve - And Why?

The target was 7 days. The roadmap to close the remaining gap was scoped and sequenced — but two things consistently interrupted it. Firstly, RBI compliance mandates arrived without negotiable timelines — Aadhaar masking, digital consent, and other regulatory changes required immediate engineering attention, consuming sprints that were planned for TAT-reduction features.

Second, in the final 6 months, leadership began exploring a platform sale, roadmap features were deprioritised, and the company was eventually shut down. The 7-day gap wasn't a design limitation. It was a combination of regulatory reality and a business decision the team had no ability to control.

9.1 - Outcomes beyond the headline metrics

Platform Scalability

Ashv Gati — a small-ticket product up to ₹5 lakh — launched PAN India in under 2 weeks. Previous benchmark: 3 months. Multi-tenant architecture made the platform licensable to other NBFCs — each with independent configuration. 45 configuration support tickets in the analysis period became zero. That ticket category no longer exists.

Structural Change - Support Result

Shared Excel replaced with a 3-tier CRM: auto-routing, L1→L2→L3 escalation, root cause analysis on every unresolved case, knowledge base updated after every resolution. CSAT, handle time, and incidents tracked for the first time. Not a feature — a structural change to how the organisation handled its own failures.

Customer Outcome — Post-Disbursal Self-Service

CAMS Finserv integration gave customers a single place to track all loans, repayments, outstanding balances, SOA access, top-up eligibility, and CIBIL health — without calling an RM. Reduced multiple hard CIBIL enquiries. Returning customers requesting top-ups became a measurable signal of service quality.

Strategic Intelligence

Express Loan NPA at 12.4% traced to a specific segment and region — enabling a targeted eligibility adjustment, not a blanket product restriction. Regional performance visible in real time for the first time. Decisions that previously required weeks of manual analysis now happened inside the platform.

The numbers measure what changed. The people show why it mattered.

Ravi's last working day of the month is no longer the bad day story. Deepa no longer enters the same data into two systems side by side. Sanya no longer tracks queries in a shared Excel with no escalation path. Vijay no longer has to call anyone to know where his application stands. The platform didn't just move faster. It changed what it felt like to work inside it — and what it felt like to be served by it.

10. Reflections

What this project actually taught

A design that works for one user can break the system at scale

RMs were happy with the design — it showed what was recommended and why. At approval, it was rejected. Credit heads saw what testing couldn't: at scale, across hundreds of RMs under month-end pressure, showing detailed eligibility criteria would turn the screen into a coaching tool. Rebuilt to show only breached criteria, no score. Enough for RMs to explain the outcome honestly — not enough to engineer it. User research shows what people need. Policy review shows what it produces at scale. The designer's job is to hold both before anything ships.

Solve the right problem first

Pre-disbursal verification before e-signature. Design system before any screens. The instinct was always to ship what felt ready. The discipline was to ask: which problem must be solved before this one can be solved well? A digitally signed incorrect application is a faster failure — not a better experience.

Users don't flag what they've normalised

Bank Statement Upload: 397 tickets, 27.3% of all volume. Not one interviewee mentioned it. They'd stopped noticing. Ticket data surfaces what interviews can't. And 54 "User Error" tickets were actually validation gaps — renaming the problem changed the solution entirely. What users call their mistake is usually the system failing to catch it earlier.

Test with the real thing, not the easy version

Single-column forms passed on a simplified prototype. Failed on the actual 50+ field form — excessive scrolling, errors impossible to find. The prototype removed the complexity that made the problem real. Always test in the context where the design actually has to work.

Catching a problem early is always cheaper than fixing it late

A four-week pilot costs weeks of monitoring. A broken national launch costs months of rework, a ticket spike, and damaged trust. Bank statement upload launched nationally without a pilot. That was the only time it happened.

11. What's Next

The work continues. The target is 7 days

TAT — From 14 Days to 7 Days

The original target is still open

Obligation sheet and bank transaction tagging still consume several hours each. Both are automatable. Hunter check integration — highest-voted in co-design — would compress pre-assessment further. Blockers are known. Target is achievable.

Banking Transactions Tagging Accuracy

From 78% to 90%

Wrong tagging produces wrong eligibility values. The hypothesis: give CMs a pattern-submission interface to flag transaction types the system misclassifies — closing accuracy from the inside through domain expertise the algorithm can't generate alone.

From Weekly Reports to Live Signals

Proactive signals for policy makers

The dashboard currently surfaces patterns only when someone investigates. The next layer: proactive notifications to the Risk Head when NPA signals cross thresholds, regional underperformance before it compounds, deviation delays before they hit TAT.

Build Feasible Modules Internally

Reduce vendor dependency

AML, VCIP and VKYC depend on external vendors — cost per transaction, failure points outside the team's control. Building internally means better failure-state design, lower cost, and full roadmap control.

The Design Question I'm Still Thinking About

The platform still requires humans at multiple points — hygiene checks, credit assessment, deviation approvals, sanction sign-offs. Which of those genuinely require human judgement, and which are manual only because the system hasn't been built to handle them? If most can be automated, the platform could serve a significantly larger loan book without increasing headcount. That's a systems design problem — and where the next version should start.

User Archetype - Credit Manager

Kredible Kiran

Credit Manager, Mumbai | 10+ Years of Experience in Credit Assessment | Post Graudate in Commerce

Male | 38 years | Married

" I have to be on the field most of the day looking for new leads while handling existing customers and reach the monthly target ""

👤 About Kiran

Kiran, a Senior Credit Manager at Ashv Finance, brings over a decade of experience in finance, handling credit assessments, loan approvals, and coordination with sales teams. A commerce graduate from a government college, he is fluent in both vernacular and English. Married with two kids and caring for his aged parents, he balances a demanding job with family life. To save time, he commutes by bike and often works late nights and weekends. Dedicated and meticulous, Pruthvi navigates the fast-paced world of finance with pragmatism and precision.

💡 What Surprised Us

Kiran wasn't failing at his job. He was running two jobs simultaneously — the LOS for submissions and Excel for every analytical task the LOS couldn't do. Obligation sheet calculation, transaction tagging, CAM generation: all manually reconstructed in Excel every single case. The system had been built around a process that assumed human cognition would fill its gaps. It had been doing that for years.

🎯 Goals & Motivations

✅ Enhance Turnaround Time (TAT) - Reduce delays in loan processing
✅ Minimize Delinquency - Improve approval accuracy to reduce default rates
✅ Career Progression - Seeks leadership roles as a validation of his perseverance
✅ Work-Life Balance - Wants to optimize work time to dedicate more to family

🔍 Needs & Expectations

🔹 Automation & AI Assistance - Reduce manual intervention and repetitive data entry
🔹 Customizable Workflows - Ability to edit loan applications to improve efficiency
🔹 Mobile-Optimized Tools - Faster approvals via DLP on mobile
🔹 Better Task Visibility- Clearer categorization of tasks with reduced redundancy

😤 Pain Points & Frustrations

🚨 Inefficient Digital Tools - Slowness, functionality issues, and manual data entry
🚨 Complex Task Management - Redundant tasks, confusing categorization, and multiple tabs
🚨 Process Delays - Postponements, cancellations, and lack of an individual TAT calculator.
🚨 Sales Coordination Issues - Incomplete loan applications lead to rework.

💡 Preferred Tools & Experience

💻 Experienced with - FinnOne, Salesforce
📲 Uses Extensively - Excel for Calculation & Word for Documentation

😊 What Delights Him?

🌟 Performance Dashboards - Helps track KPIs effectively.
📲 Mobile-Friendly Solutions - Enables efficient task handling while commuting.
⚡ Automated CAM (Credit Appraisal Memorandum) Generation - Saves time on loan approvals.

📌 Key Design Considerations

✔ Performance-Optimized Digital Tools - Reduce load time and increase responsiveness
✔ Task Simplification - Streamline categories, automate obligation checks, and remove redundant steps
✔ Seamless Mobile Experience - Ensure full functionality in mobile workflows
✔ Pre-Validated Loan Applications- Reduce back-and-forth with Sales teams

🚶‍♂️ Kiran's Journey

Based on the insights gathered from our research, we created detailed journey maps for each persona, capturing their goals, pain points, and key interactions across the touchpoints. This helped us visualize the end-to-end experience and identify meaningful opportunities for design improvement.

👓 Kiran's Point of View & "How Might We.."

Through immersive field research and user shadowing, we uncovered overlooked pain points by mapping real-world workflows and emotions. These insights helped us craft clear Point of View statements and translate them into focused "How Might We" questions, guiding purposeful design.

User Archetype - Relationship Manager

Resilient Ravi

Relationship Manager, Bengaluru | 6+ Years of Experience in Sales & Lending | Commerce Graduate

Male | 30 years | Married

"Targets come and go, but relationships stay. I just wish things were faster, smoother, and less stressful."

👤 About Ravi

Ravi is a dedicated Relationship Manager who hustles to meet monthly targets while balancing personal responsibilities. Living in Bengaluru for work, he visits his wife, daughters, and elderly parents in Gulbarga every month. Despite the pressure of closures, manual paperwork, and high ROI hurdles, he remains jovial and energetic, though he hides the stress. Not tech-savvy, he relies on intuition, people skills, and his manager’s support to navigate tough cases. He dreams of becoming a Branch Manager but struggles with manual processes, last-minute dropouts, and intense performance pressure.

💡 What Surprised Us

Ravi wasn't inefficient. He was operating a system with no product-switching logic and no field tool. Every eligibility failure was a full restart — data entry, documents, submission — from scratch. Four CMs in four cities independently named this as their highest time drain. Nobody had told them to manage it a different way. There was no other way.

🎯 Goals & Motivations

✅ Career Growth - Wants to be a Branch Manager
✅ Higher Conversions - Wishes for automation to reduce manual work
✅ Recognition - Aims to top the branch rankings every month
✅ Work-Life Balance - Wants less month-end stress

🔍 Needs & Expectations

🔹 Automation - Reduce manual data entry & paperwork
🔹 Simplified Processes - Quicker approvals & loan disbursals
🔹 Support for Sales Pitch - Clear ROI breakdown & selling points
🔹 Confidence Boost - Better understanding of credit criteria to fight for cases
🔹 Branch Manager’s Help - Easier ways to escalate tricky cases

💡 Preferred Tools & Technology

📲 WhatsApp & Calls - Primary mode of communication with DSAs & customers
📲 Basic Banking Apps - Used for transactions & customer insights
📲 Social Media (YouTube, FB, Insta) - For leisure & occasional learning
📂 Offline Documentation - Still depends on physical forms & signatures

😤 Pain Points & Frustrations

🚨 Month-End Chaos - Works late nights chasing closures and signatures
🚨 Manual Data Entry - Cumbersome form-filling and physical paperwork slows him down
🚨 Customer Hesitation on ROI - Struggles to convince MSMEs due to high interest rates
🚨 Target Pressure - Faces scolding from Branch Manager & Regional Head if numbers are missed
🚨 Lack of Automation - Wishes the loan process was more digital & efficient
🚨 Confidence Gap in Credit Discussions - Needs manager’s help to push tricky cases
🚨 Last-Minute Case Dropouts - Customers backing out impacts his credibility & commissions

😊 What Delights Him?

🌟 Topping the Leaderboard - Feels motivated & recognized when his name is on top
🌟 Quick Loan Approvals - Gets excited when cases are approved fast without friction
🌟 Happy Customers - A smooth closure with a satisfied customer makes his day
🌟 Branch Manager’s Support - Feels confident when his manager backs his case
🌟 Surpassing Targets - Loves getting rewarded for exceeding expectations
🌟 Festive Incentives & Bonuses - Enjoys extra perks during peak business seasons

📌 Key Design Considerations

🎯 Mobile-First Design - Should work seamlessly on a smartphone
🚀 Fast & Simple UI - Minimize data entry, reduce friction in processes
📢 Guided Sales Support - Provide talking points & ROI justifications
🔄 Offline Sync & Auto-Save- Helps when network issues arise
🛠️ Role-Based Access - Easy escalation workflows for approvals
📈 Gamification & Leaderboards- Motivates to top the charts

🚶‍♂️ Ravi's Journey

👓 Ravi's Point of View & "How Might We.."

User Archetype - Operations Executive

Diligent Deepa

Operations Executive, Ahmedabad | 5+ Years of Experience in Operations | Graduation in Arts

Female | 32 Years | Married

"Why are we doing these tasks? In other companies, RM’s handle them since they coordinate better with customers!"

👤 About Deepa

Deepa is a dedicated and meticulous Operations Executive who ensures seamless loan disbursals while juggling manual processes, coordination issues, and last-minute pressures. She took up this job to become financially independent and provide a better future for her son. Despite working in a fast-paced, high-pressure NBFC environment, she sticks to policies and prefers structured, predictable workflows. Month-end is a nightmare for her due to overloaded cases, delayed approvals, missing documents, and security concerns when working late. She has worked on advanced LOS systems like Perfios and Newgen, making it frustrating to handle manual, redundant tasks in her current role. Though she is tech-savvy, she doesn’t trust new systems easily and likes to gradually adapt to changes.

💡 What Surprised Us

She wasn't failing at her job. She was doing three of them. CPV chasing, FIS parallel entry, Regional Head reminders, pre-disbursal form completion, deferral email chains, VKYC restarts — none of it was in her job description. All of it was in her calendar. The system had no deviation workflow, no audit trail, no in-platform approval chain. Every exception became Deepa's problem by default.

🎯 Goals & Motivations

✅ Seamless Operations - Wants better process distribution across the month, rather than a last-minute rush
✅ Job Security - Prefers stability over risky career moves due to financial sector uncertainties
✅ Efficiency & Automation - Wishes for fewer manual interventions in the disbursal process
✅ Work-Life Balance - Wants better work hours & safer late-night travel options

🔍 Needs & Expectations

🔹 Distributed Workload - A balanced month-end process to avoid last-minute crunch
🔹 Better Coordination Tools - Digital workflows for tracking approvals, deviations & missing signatures
🔹 Automated Data Validation - Ensuring sales enter correct details upfront to reduce rework
🔹 Tech-Enabled Efficiency- Wants a modern LOS system like Perfios or Newgen for faster processing
🔹 Workplace Safety Measures- Needs cab support for late-night shifts

💡 Preferred Tools & Experience

📂 Loan Origination Systems (LOS) - Familiar with Perfios & Newgen, struggles with manual processes
📧 Email & Excel - Spends time tracking approvals & managing deviations
📲 Social Media (Insta, FB Reels) - Loves entertainment & music in free time
🚀 New Digital Apps - Willing to try new software if it’s useful & user-friendly

😤 Pain Points & Frustrations

🚨 Month-End Chaos - Workload spikes at the month-end, making it stressful & unmanageable
🚨 Lack of WFH Support - No printers, biometric devices, or digital approvals, forcing office visits even during crises
🚨 Inefficiencies in Coordination - Sales don’t enter accurate details, leading to repeated corrections
🚨 Approval Bottlenecks - Has to search through emails for approvals & deviations, causing delays
🚨 Redoing Work Due to Credit Issues - Credit Manager’s missed due diligence forces her team to redo the process
🚨 Security Concerns - No cab service for late nights, making travel unsafe
🚨 Short-Staffed Team - Frequently overburdened due to limited staff

😊 What Delights Her?

🌟 Organized & Predictable Workflows - Loves when everything is structured & documented properly
🌟 Automation & Less Manual Work - Feels empowered when systems handle repetitive tasks
🌟 Efficient Coordination - Seamless approvals & accurate sales data make her job smoother
🌟 Safe & Secure Work Environment - Prefers better support for late-night shifts
🌟 Time with Family & Travel - Loves spending weekends exploring new places with her son

📌 Key Design Considerations

✔ Process Simplification - Minimize redundant steps & ensure data accuracy from sales
✔ Approval Tracking System - An easy way to access deviations & approvals instead of emails
✔ Progressive Change Management - Introduce new features gradually as she doesn’t like sudden changes
✔ Automation & Workflow Optimization- Introduce paperless documentation & pre-validated entries

🚶‍♂️ Deepa's Journey

👓 Deepa's Point of View & "How Might We.."

User Archetype - Senior Support Executive

Savvy Sanya

Senior Support Executive, Chandigarh | 8+ years of Experience in Support Operations | Graduation in Computer Science

Female |32 Years | Unmarried

"I’ve written everything in emails and guides, but people still come back asking the same questions! A CRM and self-service portal would solve half our problems"

👤 About Sanya

Sanya is a seasoned support operations expert who has spent over a decade resolving customer queries, managing legacy LOS systems, and improving service quality. She is the go-to person for troubleshooting but often finds herself answering the same queries repeatedly, despite creating detailed guidebooks. Her work is overwhelming, tracking customer requests manually via Excel and handling oral requests from colleagues, leading to errors and inefficiencies.

She works even on weekends and holidays, ensuring seamless loan processing and customer query resolution. She advocates for automation, having worked with ZOHO & LeadSquared, and pushes for a CRM system to streamline her team’s workflow. With rising customer impatience, she believes self-service options in apps and portals could ease the burden. She aspires to lead at a higher level, leveraging her deep industry knowledge and customer insights.

💡 What Surprised Us

The system was routing every problem that had no other channel directly to Sanya — and calling it her job. Bank Statement Upload failures, configuration change requests, case escalations, user training questions — all landed in one shared Excel. She had built the institutional knowledge the system never encoded. The system relied on that knowledge being stored in one person rather than in itself.

🎯 Goals & Aspirations

✅ Process Efficiency - Wants a structured support system to reduce redundant manual work
✅ Automation & Self-Service - Aims for digital self-service tools to reduce query load
✅ Clear Communication - Desires defined SLAs so customers have realistic expectations
✅ Career Growth - Aspires to move up the org hierarchy leveraging her domain expertise

💡 Preferred Tools & Experience

📝 CRM & Ticketing Tools - Prefers Zoho, LeadSquared, but currently stuck with manual tracking in Excel
📧 Email & Internal Communication - Handles policy changes & troubleshooting via email
📲 Self-Service Portals - Believes an app-based self-service model can reduce query volume
🎧 Social Media & Music Apps - Uses entertainment apps (Insta, YouTube) in free time

😊 What Delights Him?

🌟 Proactive Solutions - Loves when issues are fixed before customers even report them
🌟 Efficiecnt Collaboration - A well-structured support process reduces chaos
🌟 Recognition & Impact - Feels valued when her expertise is acknowledged
🌟 Automation & Digital Tools - Prefers smart workflows over outdated manual work

🔍 Needs & Expectations

🔹 Centralized Ticketing System - A CRM like Zoho or LeadSquared for tracking and assigning requests efficiently
🔹 Automated Self-Service Options - Wants apps & portals for customers to handle basic troubleshooting
🔹 Standardized Communication - Prefers predefined SLAs & clear response timelines for customer requests
🔹 Better Collaboration Tools - Needs structured workflows to minimize manual coordination & errors
🔹 Recognition & Growth - Seeks career advancement as she holds critical domain & customer expertise

😤 Pain Points & Frustrations

🚨 Manual Workload Overload - Tracking requests via shared Excel leads to duplication & inefficiencies
🚨 Ignored Documentation - Despite detailed troubleshooting guides, users still ask the same questions
🚨 Legacy System Bottlenecks - Managing a tightly coupled, outdated LOS system is challenging
🚨 Lack of CRM Tools - Handling requests manually without a proper ticketing system is exhausting
🚨 Customer Impatience - Faces pressure from impatient customers who expect instant resolutions
🚨 Limited Team Strength - Short-staffed, making it harder to keep up with growing requests
🚨 Unstructured Requests - Oral requests from colleagues make tracking even harder

📌 Key Design Considerations

🎯 Automate Routine Tasks - Reduce repetitive queries by implementing self-service solutions
🔄 Integrated Ticketing System - A CRM-driven workflow to replace manual email tracking
🛠️ User-Friendly Guides - Interactive step-by-step troubleshooting instead of static documents
📊 SLA-Based Response System- Customers should receive clear expectations for resolution timelines
💡 Scalable Team Workflows- Improve collaboration & efficiency with structured request tracking

🚶‍♂️ Sanya's Journey

👓 Sanya's Point of View & "How MIght We.."

User Archetype - Wholesale Trader & Business Owner

Veteran Vijay

Wholesale Trader & Business Owner, Bengaluru | 15+ years of Experience in Business | Graduation in Commerce

Male | 48 Years | Married | Annual Turnover : 10 Crores

"My previous business loan from Axis was disbursed in 5-6 working days—that’s the kind of service I expect!"

👤 About Vijay

Vijay started his business with just ₹30K and, through hard work and perseverance, has grown into a reputed wholesale trader catering to top brands like Cloud9, Apollo, Jain, Akshayapatra, Iskcon, Motherhood, and Fortis. His day revolves around client servicing, managing logistics, and ensuring timely deliveries. Vijay is a practical and patient businessman who is always on call, even while driving. He supervises every aspect of his operations, finding it hard to trust others, as his reputation is built on reliability. While he has an admin for accounts and calls, he personally handles finances, deliveries, and critical business decisions, often working till 9 PM or even later during festive seasons.

💡 What Surprised Us

Vijay's benchmark wasn't another NBFC. It was Axis Bank — 3–4 days, no status calls, money in his account. He didn't compare us to competitors who were similar to us. He compared us to the best experience he'd had. His loyalty was conditional: we had it, we kept losing it every time a process delayed him. He came back despite the experience. Not because of it.

"You invest a lot of time providing documents and answering every clarification required and wait for weeks together to be informed that my application is rejected." — Customer research respondent. The platform's late-rejection problem had a human cost Vijay named before we asked."

🎯 Goals & Aspirations

✅ Financial Growth - Wants to surpass last year’s revenue targets
✅ Expansion Plans - Aims to open new branches & become a brand like Metro & D-Mart

💡 Preferred Tools & Technology

🏦 Banking & Payments Apps - Uses Banking Apps, PhonePe, GPay for transactions
📲 Communication - Heavily relies on WhatsApp & Email for business operations

👥 Influencers

👥 Financial Growth - Relies on word-of-mouth recommendations
📊 Expansion Plans - Keeps an eye on industry shifts & financial opportunities

📌 Key Design Considerations

📱 Digitized Loan Process - Reduce manual steps & paperwork
🔍 Loan Status Transparency - Real-time application tracking system
🛠️ Personalized Loan Offers - Tailored credit options based on repayment history
📱 Mobile-First Approach- Easy access via apps & digital platforms

😊 What Delights Him?

🌟 Fast Loan Approvals - Appreciates seamless disbursals like his previous experience with Axis Bank
🌟 Trust-Based Relationships - Prefers working with reliable partners who deliver on promises
🌟 Operational Efficiency - Values quick, error-free services that don’t interrupt business

🔍 Needs & Expectations

🔹 Quick Loan Disbursal - Wants a faster & hassle-free process
🔹 Real-Time Loan Tracking - Needs a transparent status tracker for applications
🔹 Fully Digital Process - Prefers a complete online loan journey without manual interventions
🔹 Efficient Banking & Finance - Expects NBFCs to match corporate bank service levels
🔹 Seamless Loan Experience - Needs quick access to funds with low ROI & top-up options

😤 Pain Points & Frustrations

🚨 Loan Processing Delays - Despite a strong CIBIL score & past loan repayments, he doesn’t get the desired loan amount
🚨 Manual Loan Tracking - Has to call/text sales executives to track his application
🚨 Document Hassles - Some NBFCs still collect physical documents, making the process cumbersome
🚨 Inconsistent Service - Corporate banks offer faster loan disbursals than most NBFCs

👓 Vijay's Point of View & "How Might We.."

My Role

01. Executive Summary

1. 1 - What This Case Study Covers

1. 2 - Where This Started

1. 3 - The Problem

1. 4 - What We Achieved

02. The Challenge

03. Research

3.1 — Heuristic Evaluation

3.2 — Competitor Reviews & Walkthroughs

3.3 — Support Ticket Analysis

3.4 — Ethnographic Research: Apprenticeship Model

3.5 — Stakeholder Interviews

3.7 — Primary Customer Research

3.8 — How the findings were processed - 4 Steps applied after every research phase

3.8.1 — Root Cause Classification

How Did We Classify ? - The Answer - 5 WHY'S

3.8.2 — Thematic Analysis

3.8.3 — Affinity Mapping

3.8.4 Interconnectedness Mapping

WHAT THE 6 LAYERS PRODUCED TOGETHER

04. Reframe

THE MOMENT THE RESEARCH STOPPED BEING DATA COLLECTION AND BECAME A DIAGNOSIS

Direction - Reduce TAT to 7 Days

5. Personas, Journey Mapping & HMW

6. Blueprint

6.1 - How we actually worked - a repeating quarterly loop

6.2 - Design System

6.3 - Information Architecture - Built for How Users Think

6.4 - Co-Creation — Built Over Four Years, Not One Session

6.5 - As Is -> To Be : What the Process Became

6.6 - Mobile App - Designed for the Field

6.6 - Executive Dashboard - From Weekly Excel to Real Time Data

6.7 - Usability Testing - Validating the Field Experience

6.8 - Prototypes

7. Decisions

7.1 - In What Order - E-Sign Vs Pre-Disbursal Verification

7.2 - What We Removed - 2 Features the Data made Indefensible

7.3 - What Research Found : The Feature Nobody Asked For

08. Leadership

8.1 - Evidence before argument

8.2 - Designing the conditions for good evidence

8.3 - Building the team - and what it built in me

09. Outcomes

What We Didn't Achieve - And Why?

9.1 - Outcomes beyond the headline metrics

The numbers measure what changed. The people show why it mattered.

10. Reflections

11. What's Next

The Design Question I'm Still Thinking About

User Archetype - Credit Manager

User Archetype - Relationship Manager

User Archetype - Operations Executive

User Archetype - Senior Support Executive

User Archetype - Wholesale Trader & Business Owner