Martina Lanzi

Santander · 2021

Digital Help Center

Support was branch-first, by default. We set out to change that.

Role

UX Designer

Scope

Discovery · Information Architecture · Workshops · User Testing · Salesforce

Duration

6 months

1,000+

Help Center articles — grown from an 80-article MVP

140

Self-service flows

4M+

Visits every quarter

Digital Help Center

Overview

Santander's Help Center had to move onto Salesforce. I used that integration as a chance to rethink how customers get help — because too many were calling or visiting a branch for things they could have solved on their own.

The problem

Customers defaulted to calls and branch visits — even after trying to help themselves online.

What I did

Research, a co-creation workshop, wireframes and user testing — from discovery to launch.

Why it matters

An MVP of 80 articles grew into a Help Center with 1,000+ articles and 4M+ visits a quarter.

01Part 1 of 3 · The problem

Help existed. People just couldn't get to it.

  1. 01The challenge
  2. 02Research

01 · The challenge

Fewer contacts. More problems solved without them.

Moving the Help Center onto Salesforce came with a business goal: fewer contacts across every channel, and more customers solving things on their own. The numbers showed how far off that was.

KPI · Target6.7%Minimise contactability

Reduce the contact ratio in contact centers and branches — 77,545 fewer contacts a month.

Baseline: 1,181,131 contacts a month (2020)

KPI · Starting point55.6%Reduce failed self-management

of people who booked a branch appointment online had already tried to solve it themselves.

Baseline: 76% self-management contacts a month (2020)

Takeaway

More than half the people who booked a branch visit had already tried — and failed — to solve it online.

02 · Research

What customers do, and what the traffic says.

I interviewed customers who had used the Help Center at least twice, benchmarked Mercado Libre and Despegar, and analysed a year of Help Center traffic.

  • Contact details topped the traffic all year — people were looking for a way out, not an answer
  • Scheduling a branch visit fell from 4th to 13th; starting a claim climbed from 16th to 10th
  • The best help centers lead with one question and a search bar, then a few task-based categories
Useful phone numbersSúperLíneaAttention channelsHow do I schedule a branch visit?How can I start a claim?
#1#4#8#12#16JanFebMarAprMayJunJulAugSepOctNovDec
Help Center articles by monthly ranking (#1 = most visited). Rebuilt from a year of traffic data.
02Part 2 of 3 · The process

From findings to a shared direction.

  1. 03Direction
  2. 04Design

03 · Direction

Mapping the journey with the whole team.

I facilitated a co-creation workshop with Business, Product, Engineering and UX. Together we mapped where the journey broke, prioritised content, named the product and agreed on the main flow — within Salesforce's technical limits.

The customer journey

Illustrated customer journey: a customer with a question, the Help Center, being routed to phone, branch, chat or email, getting lost in articles, the impact on contacts and operations, and finally a clear answer
  1. 01A simple questionBook a branch visit, find a document.
  2. 02The Help CenterSearch and categories — but answers are scattered.
  3. 03Routed outPhone, branch, chat or email.
  4. 04Lost in articlesLong content, still unsure what to do.
  5. 05The business sideContact volume, operations and teams.
  6. 06ClarityThe answer — without having to call.

Takeaway

The gap wasn't missing content — it was the path between a question and its answer.

04 · Design

Search first, then the shortest path to an answer.

I sketched the experience in low fidelity and iterated with the same group. Every page opens with one question and a search bar, then the most-consulted categories, with every other topic one tap away. Search surfaces self-service actions before articles, so people can act instead of just reading.

Two grayscale wireframes of the Help Center: the home with a search bar, most-consulted categories and other categories; and a category page for Cuentas y Tarjetas with a side menu, featured articles and more articles
Home and category page — search first, the most-consulted categories up top, every other category one tap away.
03Part 3 of 3 · The outcome

Validated, launched — and still growing.

  1. 05Testing
  2. 06Launch
  3. 07Three years later

05 · Testing

Five customers, one end-to-end flow.

I tested the MVP with five customers, including an A/B test on the product's name.

Validated
  • Found the entry point
  • Navigated the IA without friction
  • Completed the flow end to end
Learned
  • Keep the name “Help Center” (A/B test)
  • Expect to schedule a branch visit from it

Takeaway

Every participant found the entry point and finished the flow — and all five wanted to book branch visits from it.

06 · Launch

Employees first, then customers.

The new Help Center launched to Santander's own employees first — to catch pain points, fix inconsistencies with Flame (Santander's design system) and bring in flows other squads had already automated. It also checks back on a customer's last query.

07 · Three years later

From an 80-article test to a product the bank runs on.

The MVP started small: 80 informational articles, with support routed by customer type and query. It worked — and showed we were on the right track. Three years and many challenges later, the Help Center lives across the bank's digital channels. Most recently, we launched a Help Center Backoffice: an internal tool to create, edit and retire content, with request inboxes for every area of the bank — so the product can sustain itself over time.

MVP
80informational articles
Three years later
1,000+articles
140self-service flows
4M+visits every quarter
Live inMobile appOnline Banking · PersonalOnline Banking · BusinessPublic website

Takeaway

The key isn't creating something — it's making it work, and making sure it gets used.