Sravan.
Reimagining the field verification experience for on-ground verifiers — reducing cognitive load, cutting training time, and improving data collection accuracy.
OnGrid's address verification product relies on a network of field verifiers who visit candidate addresses, collect information, and submit reports through a native Android app. The redesign was an opportunity to make this experience significantly more effective and easier to use.
Verifiers visit individual locations and use the native app to fill in information, upload documents, and take pictures. In observing this process, three major drawbacks became clear.
Many verifiers were abandoning work or were reluctant to engage with the app and learn the verification process
Verifiers were frequently submitting wrong or inconsistent data, compromising report accuracy and downstream audit quality
Ops executives had to spend multiple days training each new verifier — creating a bottleneck that didn't scale
Getting aligned with the team. Learning about the current experience through internal documents and an app audit
Understanding product vision, constraints, and verifier needs and behaviour in the field
Identifying user needs and pain points, communicating where we needed to focus design effort
Generating and exploring solutions. Creating an interaction framework for the proposed flow
Final UI for the solution, with an interactive prototype for validation
Measuring impact of the new designs on training time, data accuracy, and verifier completion rates
After the kickoff meeting with the PM and Ops teams, I learned about the goals and timeline for the project. I then conducted an initial audit of the current experience to surface insights that would guide deeper research.
Training verifiers takes a disproportionate amount of time from Ops executives — this is the most urgent cost to address
The audit of existing screens surfaced four structural issues that were contributing directly to training burden and data errors:
To understand why operations executives spent days training each verifier — and why verifiers were abandoning the process — I conducted interviews with both groups, focusing on the specific moments where the current experience broke down.
Questions focused on why they find it difficult to understand and retain the LAV process and scenarios
Questions focused on why verifiers face difficulty during onboarding and what repeated corrections they make
"It is hard to understand and remember address verification processes"
"Sometimes I mistakenly enter wrong results"
"Gets worried about whether I will get paid after getting work done"
"Verifiers don't grasp the process quickly — we repeat the same training multiple times"
"Some verifiers try to ask candidates for money"
Verifiers face significant difficulty during the onboarding process — the app gives them no guidance on how to analyse a scenario and select the right result
Having conducted interviews and combined findings across participants, I developed a persona to represent the verifier user group and communicate needs and pain points to other stakeholders.
Works part-time with OnGrid and full-time with a food delivery platform in Gurgaon. During onboarding, finds it tedious and sometimes very hard to learn verification scenarios and analyse situations on field.
"OnGrid earnings support part of my livelihood"
Beyond the persona, I developed a customer journey map for the verifier — communicating exactly where in the experience we needed to focus design attention.
The amount of mental effort required to learn new information overwhelms most verifiers — causing them to abscond. We decided to focus on reimagining the onboarding process, alongside other usability problems.
From interviews with verifiers and operations executives, it became clear that verifiers need to understand three distinct scenarios and correctly analyse the situation to select the right result — without any guidance from the current app.
The candidate stays at the given address — verifier confirms presence and collects the required documentation
The candidate does not stay at the given address — verifier must note this accurately without ambiguity
The address provided by the candidate is invalid or doesn't exist — verifier needs a clear way to report this outcome
The current app gave verifiers no structured guidance for making this determination — they were expected to remember the logic from an in-person training session.
After identifying where the problem existed, we explored ideas rooted in a single core principle: rather than requiring verifiers to memorise complex decision logic, the system should be able to determine the outcome based on the verifier's inputs.
Humans have a limited capacity for processing new information — especially in unfamiliar, high-pressure field situations. Requiring verifiers to recall abstract scenarios from memory creates unnecessary failure points.
By redesigning the flow as a guided questionnaire, the system can analyse the verifier's answers in real-time and determine the correct scenario classification — without the verifier needing to know the terminology at all.
A questionnaire-like experience will allow the system to estimate the scenario based on the answers submitted by the verifier on field — removing the need for verifiers to memorise outcome categories
The redesigned user flow guides verifiers through a structured decision tree — branching based on whether the address was found, and then further based on address type (Business or Residential).
Verifier opens the app and sees their assigned requests in a clear, scannable list
Before beginning a verification, the verifier sees context about the request — removing the need to recall this from memory
First branch point — if No, the verifier selects a reason (incorrect number, left the job). If Yes, the flow continues
Business or Residential — the app routes the verifier into the correct data collection flow based on their answer
Type-specific questions with camera capture, landmark upload, and document verification — with clear progress indicators throughout
Based on the verifier's inputs, the system classifies the outcome and generates the verification report — verifier reviews and submits
After defining the user flow, I built wireframes in Balsamiq to establish the interaction framework. The visual design referenced OnGrid's branding — using teal blue as the primary CTA colour — while fixing the hierarchy, spacing, and interaction pattern issues identified in the audit.
Replaced overwhelming forms with a guided, single-task sequence to eliminate choice paralysis and ensure focus on one step at a time.
Hiding non-essential fields reduces overwhelm and prevents skipped steps by focusing the user only on the immediate requirement.
A real-time camera layer automatically scans documents and provides instant feedback—like "Too blurry" or "Fit ID within the frame"—to ensure high-quality uploads on the first try.
In-app capture guidance prevents poor-quality photos at the source, eliminating audit rejections, costly site re-visits, and turnaround time (TAT) delays.
Based on the verifier's field inputs, the system automatically classifies the verification outcome and generates the report — the verifier reviews and submits without memorising scenario rules or outcome categories.
Moving outcome classification from the verifier to the system eliminates guesswork at submission, reducing report errors, audit rejections, and the training burden on operations teams.
When the request is finally submitted, positive audible feedback confirms for the verifier that their actions were correct — reassuring verifiers that they have successfully submitted the request.
Auditory confirmation at the point of final submission reduces cognitive load and reinforces correct action completion — giving verifiers confidence that the request went through without requiring them to visually re-scan the screen for status cues.
Agent drop-off plummeted from 60% to 10%
Verification capacity increased, driving a 50% rise in revenue.
Shifted operations from "firefighting" and recruiting to high-level quality audits.
Minimise cognitive load to maximise usability — the total mental processing required to use a product directly determines how quickly users can complete tasks and how likely they are to make errors
Moving decision logic into the system (rather than requiring users to memorise it) is one of the most impactful UX interventions available — particularly for infrequent or complex tasks
The relationship between usability problems and business problems is direct — high drop-off, poor data quality, and heavy training burden were all downstream of the same root cause: an app that gave users no guidance
