About Maatrum
Maatrum is an India-based full-stack legal verification and property valuation provider for banks and NBFCs. When a lender values a property before sanctioning a loan, Maatrum sends a civil inspector to the site to capture the ground truth a valuation report is built on.
Product
The Site Inspector app is the field arm of that operation. It’s a mobile web app that walks an inspector from recruitment all the way to their first paid visit, capturing geo-location, boundaries, dimensions, condition, photos and a selfie with the attendant, then sending a structured, geo-tagged and time-stamped record straight to the back office.
Core problem & pain points
Inspection used to be a two-step manual chain. Inspectors carried an A4 template, filled it in by hand on-site, then WhatsApped the photos and the scanned sheet to the office. A desktop team re-keyed all of it into the portal after that.
The pain was everywhere. Inspectors hand-wrote 50+ fields in the sun, sheets came back illegible, media went missing in chat threads and the office burned hours re-typing everything. Worst of all, there was no proof the inspector had actually been there. A desk-side inspection was easy to fake.
Challenges
Inspectors trusted paper. A screen had to earn that trust.
Connectivity is unreliable. A rural plot and a dense city lane both drop signal.
Units differ by region (feet, metres, sq. yards, cents).
One flow had to feel bank-grade to legal teams and rugged to field agents.
Who I designed for
Primary was the Site Inspector: a civil-diploma field agent who moves between sites on a two-wheeler, phone in one hand and a measuring tape in the other, squinting under the sun on a patchy network and motivated mostly by the weekly payout.
Secondary were the back-office valuers, who need clean geo-tagged data mapped to exact report fields, and the banks, who need the whole thing to be provably trustworthy.
Research
During a 6-week field pilot I kept seeing inspectors do “double work”, filling in the paper sheet and then the app. That turned out to be the most useful signal I got, because it showed me exactly where they didn’t yet trust the screen. Two findings changed the design. The app was too rigid: at a vacant plot, people were hunting for building fields that shouldn’t have been there at all. And they felt boxed in by the fixed photo slots, after years of just dumping 20+ images into a WhatsApp chat.
Information architecture
With 50+ fields, one long form was never going to work. So instead of grouping the data the way a report does, I ordered it the way an inspection actually happens, following the inspector from arrival through to investigation: Attendant → Verification → Land → Building → Additional Info → Caution Areas → Comparable Sites → Market Value. A progress stepper shows only the 3-5 fields tied to where they’re physically standing, and a percentage counter keeps a long visit feeling like it’s moving.
How I approached it
The goal I kept coming back to was to productize the workflow, not just digitize the form. Four principles ran through every screen:
Mirror the physical journey, so the app feels like an extension of the inspector’s movements rather than a chore.
Verify, don’t type. Most on-site data is checked against bank records, not discovered, so input should really just be a tap.
Glanceable and thumb-friendly: big touch targets, usable one-handed, readable in bright sun.
Offline-first, because a lost signal should never cost someone a re-measured property.
Sign Up & Onboarding
Inspectors here are being recruited, not just onboarded, so I treated the whole thing like a product landing page. It leads with earnings and makes them tangible through an interactive slider (visits per day times days per week gives a live monthly figure), then drops the barrier with “no fees, no experience needed” and a plain checklist of what you’ll need.
Sign Up
Most of the convincing is done by the trust signals, reports accepted by 40+ banks & NBFCs and coverage in 60+ cities. A two-field phone and OTP entry then keeps the sign-up itself almost frictionless.
Application & Status
An application is a form plus a waiting period, and the real risk is people getting anxious and dropping off while a human reviews them. So I split the form into two light tabs, Details and Work experience, with clear upload states for documents. For the wait, I borrowed the mental model people already have from tracking a parcel: a five-stage timeline (submitted → review → interview → verification → ready) showing the current step, an “est. update in 48h” and a “Message desk” button for when they’re unsure. Dead time starts to read as progress, and it all ends on a clear “You’ve been selected” screen.
Details & Work Experience
KYC is where any Indian gig platform bleeds the most users, so I leaned hard on the verify-don’t-type pattern. Every credential resolves to a clear state that also shows how it was checked. Aadhaar reads “verified via OTP”, the bank account is confirmed with a one-rupee penny-drop (“we sent ₹1 to your account”) and PAN is matched against the payout account. A selfie-liveness step locks the identity, and everything closes with a readable employment agreement and a draw-to-sign signature on a “Verified · Activated” screen. That’s the bit of closure I wanted them to feel before earning their first rupee.
Application Status
Identity & Payment Details
Home Screen & Earnings
The home is a field agent’s mission control, built to be read in seconds. An availability toggle hands the inspector control over their own workload, and nearby assignments show distance, payout and deadline together so they can pick by route and value.
Home Screen & Earnings
Resume-work cards make offline recovery visible: a crashed session comes back from a card that shows how far along it was. And the earnings view sits “inspections done” right next to the money, so the link between doing good work quickly and getting paid is always in view. For a high-volume job, that’s the thing that keeps people going.
On-site Property Inspection Flow
This is the heart of the product, and where most of the design decisions ended up. It opens with presence-first trust. Address and GPS get confirmed, then the owner reads out an OTP and the inspector takes a mandatory selfie with the attendant. It works like a handshake, and it’s what makes a desk-side inspection impossible. From there the stepper carries them macro to micro.
Three input patterns do most of the work here:
Verify, don’t type: boundaries, owner and address arrive pre-filled, and the inspector just confirms each one with a green check or flags it with a red cross.
Compute, don’t calculate: they enter length and width per element and the app works out the built-up area itself, labelled “Calculated from Building details” so nobody has to trust their own mental maths. And a modular flow drops whole steps that don’t apply.
Pick “Plot / Land” and the Building/Interior section simply disappears, which hands back the flexibility they had on paper. A unit toggle lets them work in feet or metres while the back office still receives normalised data.
On-site Property Inspection Flow
Underneath all of it, local storage acts as a vault. Progress caches between steps and resumes by MTR number, and when data is being written a bottom-sheet reports “Saving…”, tells people not to close the tab or go back and lets the heavy media finish uploading later from better coverage. A green “Inspection completed” screen gives the closure a paper hand-off never did.
Design system
Maatrum verifies legal and financial records, so the app had to feel bank-grade while staying legible on a bright plot. I built it from scratch on a warm beige base, borrowed from e-ink readers, which cuts the glare you get off a white screen and keeps the data as the hero of every screen. Amber carries the primary actions, with a bit of gradient depth so a tap target still reads as tappable to sweaty fingers, and green and red badges handle every verification state. I set everything in Plus Jakarta Sans for its high x-height, since a single misread door number or square-footage figure can invalidate a whole report. Data groups into rounded cards that echo the sections of an actual valuation document.
Results & impact
We tracked weekly usage right through the pilot, and once adoption crossed 95% it was clear the friction had gone. The paper and the whole re-entry team disappeared. Every tap now maps to an exact field in the automated valuation report, and the back office gets structured, geo-tagged and time-stamped data that’s ready for estimation.
95%+
adoption (6-week pilot)
1 Step
2 steps → 1 (capture, no re-entry)
40+
banks & NBFCs
60+
cities
