Collectors had an app. Nine in ten ran their day off paper instead.
I designed the app that Piramal Finance's field collectors now use to plan their day, find customers, and record what happened at the door. Paper liners are gone, and the same team collects from a loan book that keeps growing.
A collector's job is easy to describe and hard to do.
Find the customer. Have the conversation. Record what happened. Collect if you can. A field collector at Piramal Finance does this for 40 to 60 cases a month, across every loan product the company sells, on a low-end Android phone, in traffic. The tool they had for it was so slow and so incomplete that most of them had quietly gone back to paper. I owned the redesign end to end: framing the problem, the field research, the design, and the handoff to a team of four engineers.
Walked the day with collectors
I shadowed collectors through real workdays in four branches before drawing anything. The problems were not where the product brief said they were.
Fixed the data before the screens
Phone numbers, locations, and past conversations already existed somewhere in the company. Getting them onto the collector's phone did more than any layout could.
One record for every case
Every call, visit, promise, and payment now lives in one place that the collector, the BRM, and the next collector can all see.
The official app was so slow that paper won.
Here's what a collector's day looked like before. The company's app was LCS, built on Microsoft Power Apps. It was slow to load and unreliable in the field. Customer phone numbers were often missing. Customer locations were never there, so the collector looked the address up on a maps app, drove to the area, and then searched on foot for the right building, which cost 15 to 20 minutes on every visit. The update form asked for far too much. And the case list had no order to it, so nobody knew which visit mattered most today.
So collectors did the sensible thing. Every morning they printed the month's cases from an Excel sheet, one case per row, and worked off that. These "liners" were the real tool. LCS was updated because it was mandatory, and hated because of how long it took. In the evening, each collector called their BRM and read out the day's updates case by case, 15 to 20 minutes per collector, while the BRM typed it all into another spreadsheet.
Speed was the complaint everyone led with. The bigger problem was that the real record of every case lived on paper and in phone calls. Nobody had the full picture of a case: not the collector, not the BRM, and not the next collector who picked it up the following month.
Some things were settled before I started.
Cheap phones, patchy network
Collectors use low to mid-range Android phones and spend the day moving between areas with uneven coverage. Whatever I designed had to be light, quick, and usable with one hand on the street.
Proof of visit was not up for debate
Business and compliance needed a selfie and a location match for every visit. That's an industry-wide norm and collectors can't opt out of it. My job was to make it take seconds inside the flow they were already doing, not to argue about it.
One app for the whole day: plan, find, talk, record, collect.
The new app opens on what matters this month and what to do today, in the right order. Every case carries the customer's location, photo, phone number, and past conversations. Recording a visit takes two taps and a short form. Payments happen by QR or link, and the customer gets a receipt straight away. Everything a collector does is written to a single case history that their BRM can read without a phone call.
Five screens carry most of the day.
The day, in the order it should happen
The top of home shows the numbers a collector actually tracks: cases resolved this month, where they rank, and what they've earned. Below that is the day's schedule, already sorted so that the customer who promised to pay today is first, not buried in row 37 of a printout.
Getting to the door without the guesswork
The RM captures the customer's location when the loan is taken. It was sitting in the origination system all along. The app now shows it, along with the present and employer addresses as backups, and opens navigation in one tap. The 15 to 20 minutes spent walking around asking for the house are gone.
Know the story before you knock
Every case leads with what the collector should know before the conversation: the last promise and whether it was kept, how this customer usually pays, how the last conversation went. A photo confirms they're talking to the right person. Then the phone number, the address, and the payment link are one tap each.
Recording the visit: two taps, then a short form
The old form asked for everything at once. The new one first asks a single question, "what was the result?", with four answers. The form that follows only asks for what that result needs: a promised amount and date for a promise to pay, a reason for not contactable. The selfie and location capture live inside the same flow, so proof of visit isn't a separate chore.
Taking the payment without cash
Collectors no longer collect cash. They show a QR code or send a payment link, and the moment the customer pays, a receipt goes to the customer and the payment shows up against the case. Before this, a customer who handed over cash at the door got nothing on paper.
I spent the first weeks on the route, not in Figma.
I visited four branches and shadowed five collectors through full workdays: the morning printout, the drive, the search for the address, the conversation, and the evening call. I sat with their BRMs to understand the other end of that call. I wasn't able to speak to customers directly, so what I know about the doorstep comes from watching it, not from asking.
5 collectors, 4 branches
Full days in the field. The printout in their hand told me more in five minutes than the product brief had.
BRMs
Each manages four or five collectors. Daily calls with every one of them, plus keeping their own Excel current, was a job on top of their job.
Product and business
What the app had to do, what compliance needed from it, and which numbers the business actually watched.
Three things the brief didn't say.
Mandatory use is not adoption.
On paper, LCS had 100% usage. Every collector updated it, because they had to. In practice, nine in ten were running their day off a printout and treating the app as an evening chore. So "are they using it" was the wrong question. The one I designed against was "have the liners disappeared?" That set the bar: the app had to be faster than paper at the moment the collector reaches for it, or it would lose again.
The most expensive 20 minutes of a visit happened before the conversation started.
Reaching the area was fine. Finding the actual house wasn't. Collectors were asking shopkeepers and neighbours, which is slow and awkward for everyone. Meanwhile, the customer's location had been captured at origination and was sitting in another system. Surfacing it, along with the present and employer addresses as fallbacks, turned out to be the single biggest time saving in the whole project, and it needed almost no new design, just the right data on the right screen.
Every case was being worked from memory.
A case can pass between collectors from one month to the next, and a BRM or ARM may step in on a tough one. But the history of a case lived in a printout, a phone call, and a BRM's spreadsheet. Whoever picked it up next started from zero. The app now keeps one case history, every call, visit, promise, and payment, that everyone on the case can read. The evening call became unnecessary because the BRM can just look.
Most of the design work was in the order of things, not the look of them.
A queue, not a list.
The liner had no order. A collector picked cases by instinct, or by whoever was closest. From the field days I worked out which cases actually move the month, pitched an order to product and business, and they took it:
A customer who promised to pay today is the most likely money of the day, so they come first. Then the customers who haven't promised anything yet, then the ones nobody has reached, then fresh cases. The home screen is built on this order, so a collector opens the app and simply starts at the top.
Ask what happened before asking for details.
The old update form put every field on one screen for every outcome. Most of them didn't apply to most visits, and collectors were filling them in from the car at the end of the day, if at all. I split it in two. A bottom sheet asks one question with four answers: not contactable, promise not given, promise received, payment received. Only then does a form appear, and it asks only for what that answer needs. Selfie and location are captured in the same pass, so compliance rides along instead of being a second task.
Let collectors leave notes for the next collector.
Because different collectors can handle the same customer in different months, each update asks how the conversation went, whether the customer was polite or rude, and whether they're likely to be difficult. That's written by the person who was there, at the time, and it shows up as "before calling" context for whoever comes next. It replaces the version of this that used to happen over the phone, from memory, a week later.
What I tried before this.
Several versions of the home screen, the case card, and the update flow didn't survive contact with collectors. The ones below show where I went wrong and what the fix was.
The happy path is one visit. Most days aren't.
A lot of visits don't end in a payment, and a lot of cases don't stay with one person. These were designed up front rather than patched in later.
What if the stored location is wrong?
The app doesn't pretend to know. It shows every location it has, numbered and labelled with where it came from: captured at loan time, the present address, the employer's address. The collector picks the one that makes sense and navigates to it.
What if the customer can only pay part of it?
A promise to pay carries its own amount and date, not just a yes. If a customer can only manage a portion this month, that's recorded as such, and it's what the next visit is measured against.
What if the collector is absent, or the case is too tough?
A BRM can go along on a hard visit, or handle a case alone if the collector is out. They use the same app, see the same history, and write to the same record, so nothing is lost in the handover.
The app didn't need to be beautiful. It needed to be faster than a printout at the moment a collector reaches for one.
Piloted in one branch before it went to a thousand people.
I worked with one PM and four engineers for about six months. The field findings shaped the scope before there was a spec, and the priority order, the two-step update, and the location data were all argued out in product discussions with prototypes rather than documents. We piloted with 11 collectors at the Vasanth Nagar branch first. The aim was an app that needed no onboarding, and we still gave a short demo in the branch to be safe. What we learned in the pilot went into the build before the wider rollout, and I did design QA on every screen before it shipped.
The printouts are gone, and the same team collects more.
- →Almost no collectors on paper liners, down from around 90%. Usage was always 100% because it was mandatory. This is the number that actually changed.
- →15 to 20 minutes saved on every visit, because the customer's location is on the screen instead of found on foot.
- →Around 2% more collected, from the same collectors getting through their cases faster.
- →A growing loan book, handled by roughly the same number of collectors. The efficiency is what makes that possible.
- →No more cash at the door. Every payment goes by QR or link, and every customer gets a receipt.
- →One case history that collectors, BRMs and ARMs all read and write, instead of an evening phone call and a spreadsheet.
What I learned
The month still happens in its last week.
Collectors do most of their work in the final days of the month, because that's when the numbers get counted. The app made each visit faster but hasn't yet changed when visits happen. The next things on the list are about that: daily incentives so effort spreads across the month, an ordered route for the day so the schedule is also a map, and instant reconciliation so a payment shows up the moment it's made.
The best screen was the one with the right data on it.
The biggest win in this project was a map pin that came from a system the collectors had never seen. Most of what made the app faster than paper was information that already existed, put where the collector needed it, in the order they needed it. The interface mattered, but mostly by staying out of the way.