— Case 02 · 7 min read

One payment, 300 services

50 utility payments used to take 40 minutes, one by one. That was the obvious problem. The hidden one took longer to find.

Roles
Senior Product Designer (Lead)
Team
UI Designer (partner). Content Designer (partner).
Duration
~2 quarters
Format
B2B Payments · Web
Interbanking · redesigned bulk payment flow · two desktop screens showing service-add panel, multi-invoice table, and save-as-favorite drawer.
Fig 01Redesigned bulk payment flow · service-add, multi-invoice selection, save-as-favorite.
— Context

Interbanking is Argentina's main B2B financial platform — a hub where companies consolidate all their bank accounts and operate payments, collections, investments, and more in one place. Users range from SME owners handling everything themselves to enterprise treasury teams where preparing, authorizing, and sending a payment are three separate roles, handled by three separate people.

I worked in the payments vertical. Within it, utility payments — paying recurring services like electricity, telecom, or municipal taxes — was one of the highest-revenue operations on the platform. It was also, for companies paying dozens of services per month, one of the most painful.

300
Services / month for largest customer
40 min
Old time for 50 services
3 roles
In enterprise authorization flow
— Act 01

The wrong brief

The request came from commercial. The solution came pre-packaged.

A commercial lead flagged an insight: a large company with 300 monthly utility payments wasn't using Interbanking to pay them. They were using a competing platform — and the reason was simple. Loading 300 services one by one wasn't something they were willing to do. They wanted a file upload.

That was the brief: build a bulk import via file.

Before designing anything, I started pulling on the technical thread. Interbanking's utility payment search runs through an external API with strict lookup constraints — you can't search by free text or barcode, only by specific identifier fields defined per service type. Any file parsing solution would need an algorithm robust enough to map a client's file to that structure, handle errors correctly, and not create more problems than it solved. When I raised this with the tech leads, the estimate confirmed it: high development cost, real risk of poor data quality.

I went to the PO with a counter-proposal: split into two tracks. First, a manual bulk flow — let users search and add multiple services in a single session instead of repeating the full flow one by one. Faster to build, immediate value, lower risk. But before locking in either track, I wanted to talk to actual companies.

Original flow · five labelled steps connected with arrows: Search service, Select an invoice, Save as favourite, Add to cart, Prepare payment. Every service repeats the full chain.
Fig 02Original one-by-one flow · the full screen chain repeats per service. 50 services = 50 loops.
— Act 02

The discovery that changed the question

We went in to validate a solution. We came out with a different problem.

We spoke with several companies — including the original one with 300 services — to understand how they actually worked, what files they had access to, and what kind of solution would genuinely reduce their effort.

Two things came out of those conversations.

The segment split

Large companies (300+ services) were willing to download a template, fill it in, and upload it — once. They understood the upfront effort as a one-time cost. Medium companies (around 50 services) weren't interested in managing a file at all. They preferred to do it manually, as long as they could do it in one flow instead of fifty.

That confirmed the dual-track approach: two different solutions for two different segments, complementary not competing.

The insight nobody expected

During the conversations, we discovered that most users didn't know that saving a service as a favourite meant it would appear automatically on their payment calendar every month. They thought they had to search for and load every service from scratch, every single month.

For a company paying 300 services, that wasn't just a usability problem — it was the reason they hadn't adopted Interbanking in the first place. They were mentally pricing in a 300-service manual load every month. When we explained that the load happened once and the calendar handled the rest, the reaction was immediate: "If that's how it works, I'll fill in your template right now."

The feature that unlocked adoption already existed. Nobody knew it was there.

Segment A
Large companies (300+)
Willing to download a template, fill it in, upload — once. Treats upfront effort as a one-time cost.
→ File upload (Phase 2)
Segment B
Medium companies (~50)
Not interested in managing a file. Prefers manual loading — but in one flow, not fifty.
→ Manual bulk (Phase 1)
Fig 03Discovery insight · two segments, two solutions, complementary.
— Hidden value

The feature that unlocked adoption already existed. Nobody knew it was there.

— Act 03

The dual solution

Manual bulk first. Favourites visibility everywhere.

Manual bulk flow

The core change was letting users search and add multiple services — across multiple companies — in a single session. Instead of repeating the full flow fifty times, a user could build a complete payment batch, select invoices across all loaded services, and prepare the payment in one go.

The complexity was in the details: how favouriting logic worked when multiple invoices were selected at once, how the review table communicated what was selected, and how to handle the authorization flow downstream — where in enterprise companies, preparing, authorizing, and sending are separate roles with separate permissions.

We ran usability tests focused on flow comprehension, the multi-service summary, and the updated favouriting logic. Edge cases around the review step were caught and resolved before handoff.

— Before · One-by-one

~40 min
For 50 services
Search × 50
Add identifier × 50
Retrieve invoice × 50
Prepare × 50

— After · Bulk flow

< 5 min
For 50 services · scales to 300
Search batch · Select invoices across services
Single review · Single prepare
Fig 04KLM analysis · cognitive walkthrough estimating task completion.

To demonstrate impact before the feature reached full production, I applied KLM analysis — a cognitive walkthrough method that estimates task completion time based on individual interaction steps. Modelling the old flow against the new one for a 50-service payment batch: from approximately 40 minutes of repetitive one-by-one operations to under 5 minutes in the redesigned bulk flow. The analysis became a reusable methodology reference shared across other verticals in the platform.

Multi-service search · web · empresa picker open with Edenor, Metrogas, Telecom, Edesur, Personal.
Step 1 · Searching for services across companies.
Adding multiple services in the same screen · empresa form on the left, services-added panel on the right with Edenor and Metrogas grouped, inline error state.
Step 2 · Adding more than one service in the same screen.
Invoice selection · cross-service batch · Edenor expanded with two invoices selected, Metrogas collapsed, total summary header.
Step 3 · Reviewing invoices from multiple services side by side.
Saving as favourite · drawer overlay on the right with name field and 3 included services (Personal, Telecom, Edenor) listed with their identifiers; informational note about automatic invoice lookup next month.
Step 4 · Saving invoices from different services together as a favourite.

Favourites visibility initiative

The discovery about "save as favourite" became a separate workstream. We designed educational wizards and contextual messages to surface the feature for new users — explaining clearly that loading a service once saves it for future months. These were implemented in both the payments sub-home and the main Interbanking dashboard, targeted at users who hadn't yet engaged with the feature. The initiative drove more traffic to the payments vertical.

A file upload option using a downloadable template is also in progress for large-company use cases.

Educational wizard · 'Guarda tus facturas' tooltip pointing to Guardar como favorito button on the invoice selection screen, explaining how saving makes invoices appear automatically next month.
Educational wizard · in-flow tooltip on the invoice selection screen, surfacing what "Guardar como favorito" actually does for next month.
Contextual message · 'Tus facturas favoritas' tooltip on the Pagos sub-home, pointing to the next-due-dates table and explaining how favourited invoices auto-appear in the calendar.
Contextual message · on the Pagos sub-home, the upcoming-dues table makes the payoff of saving favourites visible at a glance.

The result

50 services: 40 min → < 5 min

At 300 services, the reduction compounds further. The manual bulk flow is in production. Because the platform is mid-migration to a new architecture, rollout is currently limited to users on the updated version — full adoption metrics will consolidate as the migration completes. The expected impact is direct: utility payments are one of the highest-revenue operations on the platform. Reducing friction at this scale drives adoption, and adoption drives revenue.

— Closing

What I learned

Questioning the solution before building it is part of the design work.

The brief said "file upload." The constraints said the algorithm would be brittle and high-risk. The research said most users didn't need it — they needed something simpler. Delivering the brief as received would have taken longer and solved less.

Hidden value is still value — it just needs to be visible.

The most impactful discovery in this project wasn't something we designed. It was a feature that already existed, working correctly, completely unknown to the users who needed it most. Sometimes the right research question isn't "what should we build?" — it's "what do users not know we already built?"

— Closing line

Questioning the solution before building it saved six months of the wrong work.

— Next case · 03 / 03

The pieces that didn't sell

Continue →