ING Help Center
Making ING's support experience a differentiator
How can we reduce the time it takes to resolve a customer issue?
Improved FAQ accessibility and content, added chat history, made chat asynchronous, and added a chatbot
23% reduction in customer waiting time. 15% reduction in average handling time.
Lead designer, worked on content analysis, information architecture mapping, visual design, prototyping, and usability testing
ING Philippines' has two 24/7-customer service channels.
First is In-app Messages. This functions a lot like emails or SMS messages where customers can send a message and can come back to it later. These in-app messages are usually responded to within 24 hours.
The second channel is Live Chat. This is the channel for urgent concerns where customers can get the help of a customer care agent to resolve their issues.
The problem with Live Chat is customers often wait ~10 minutes (or more) to talk to a customer care agent. This wait is a very unpleasant experience, especially if you have an urgent issue that is time-critical, like reporting fraudulent or unauthorized transactions.
Another problem is that reaching out via Live Chat became the preferred channel. This happened because customers were often unable to find solutions in the FAQ or in the website. And, since there is no prioritization functionality in the chat, a lot of the "trivial" queries sent through the Live Chat occupied customer agents' time and effort.
Because of the growing concerns around providing quality customer service, ING's leadership challenged us with the following:
How might we reduce the time it takes to resolve a customer's issue when they reach out via Live Chat?
- Primary: When a user with an issue contacts support, I want to address their issue as fast as possible so that they can get back to doing whatever they need to get done.
- Secondary: When a user contacts the bank about an issue, I want the experience to be pleasant and seamless so that they feel that they are taken care of by ING.
- Provide support as fast as possible: Customers should not be spending too much time waiting for an answer.
- Provide correct and helpful answers the first time: We don't want customers repeatedly coming back for the same issue. As much as possible, we want to solve it the first time around.
- Minimize the time it takes to provide a solution: We want to resolve the issue as fast as possible. If the issue has no true solution, we try to alleviate the customer's concern.
- Drive activation, retention, and engagement metrics: Our solution should prove that it can impact the metrics that matter.
Frequently Asked Questions (FAQs) are not easily accessible
When we interviewed a few customers, 11 out of 12 said that they'd look at the available self-service options first before reaching out to customer support.
But when using the ING app, customers often went to ING's communication channels, Live Chat and Send a Message, and barely used the self-service option.
When I investigated why customers barely used FAQs, I learned that the entry point to FAQs was barely noticeable by customers. There were several reasons for this:
- Bad placement: The entry point was at the top right of the screen. This part of the screen is often overlooked.
- Bad contrast: The thin line icon contrasted horribly against the bright orange background of the header. It even failed contrast tests, yikes.
- The floating action button has more affordance: The call-to-action for the communication channels is a very visible bright orange button contrasting against a white background. It is also placed at the bottom left of the screen which is very near to the thumbs making it very easy to click. tldr; very very very easy to click
Self-service content and structure are not updated
The content in the app's FAQ section was made by the Marketing team. They made it in anticipation of what customers might ask. It also hasn't been updated since the app was launched two years ago.
The Help content in the app was also different from that in the website, because the website was more frequently updated by the Marketing team.
Chat transcripts between bank and the customer are not automatically saved
As a regulatory requirement, the bank needs to be able to keep records of its communications with a customer.
Because the chat tool doesn't automatically save the transcript of a conversation between bank and customer, agents need to copy and record this manually. This creates a lot of extra work for customer agents and adds an unnecessary 2-3 minutes to their average handling time.
Agents can only talk to one customer at a time
The existing chat platform allowed an agent to speak to only one customer at a time. The bank usually had 3-4 customer care agents available at any time and these agents could answer queries at an average handling time of ~15 minutes. This meant that, if more than 4 customers sent a question via Live Chat at the same time, some of these customers would need to wait to talk to a customer care agent.
Customer's messages are only received by the bank once a "connection" has been established
This was a technical limitation of the chat platform we were using. As an effect of this limitation and the previous cited issue—i.e., being able to talk to only customer at a time—customer's often need to actively wait for an agent to be connected.
This "active waiting time" meant that customers needed to keep their app session alive by tapping on the screen. Otherwise, they get disconnected from the chat and they have to reconnect and queue again.
There wasn't one solution that would solve all the issues we have. The leadership team wanted to have a chatbot but it wasn't something that could be implemented quickly.
So, we looked at several improvements which varied in terms of impact and effort required. When thinking about possible solutions, we considered the following:
- How effective is it against the jobs-to-be done we set?
- What's the level of change necessary to our existing product?
- What's the effort to build this?
Make FAQs more accessible
I redesigned the FAQs to be the first touchpoint for customers looking to get support. This way, customers are encouraged to look for answers to their issues first before contacting support. This is aligned with our guiding principle of providing support as fast as possible and minimizing the time to provide a solution.
A few stakeholders wanted to actively discourage customers from using Live Chat. I disagreed with this. We should not discourage customers from immediately reaching out to get support from agents.
While we want customers to self-serve first, we also recognize that there are occasions where getting human assistance is better—like, reporting a fraud or blocking a card. In these kinds of "high-touchpoint, high-urgency" scenarios, customers should be getting the correct help as fast as possible.
This solution was a 2/5 on the level-of-change scale, a 2/5 on the effort scale, and a 5/5 on the impact scale.
Improve Help content and structure
We studied how different apps and digital banks structure their help sections. We looked at apps like Grab, Citibank, Revolut to see how they structured their Help content. We did this to understand what patterns already exist and are well-understood by our customers.
And then we analyzed the current state of our Help content—what customers were frequently asking and what was currently on our website—and compared it to what we wanted the structure to be.
This solution was a 1/5 on the level-of-change scale, a 2/5 on the effort scale, and a 3/5 on the impact scale. This was also a pre-requisite to the Chatbot effort.
Add chat history
Customer experience-wise, adding chat history allowed customers to view their past chats with the bank. This was helpful because, previously, chat conversations containing instructions or possible solutions would disappear when the chat ended. Now they are kept in a single, easy-to-access location.
For banking operations, adding chat history allowed agents to view past conversations the customer had with the bank without navigating away from their current conversation with the customer.
Adding chat history also reduced each agent's after-chat work like copying the chat transcript or re-sending instructions to customers via in-app message so that it didn't disappear after the chat ended. In total, after-chat work took up ~1 hour of each agent's time every day.
Overall, this improvement helped decrease the average handling time which, in effect, allowed customer agents to service more customers per day.
This solution was a 2/5 on the level-of-change scale, a 3/5 on the effort scale, and a 5/5 on the impact scale.
Make the chat asynchronous
We want to eliminate the time our customers spend actively waiting for an agent to get connected to them. Customers should be able to put aside their phone and continue on with their life while waiting for a response from the bank.
Synchronous chat means that customers and customer care agents have to be "connected" to send or receive messages successfully. Making the chat asynchronous means that customers can send their messages without waiting for an agent to get connected and that agents—once they are available and are connected to the customer—can read the messages the customer sent.
This solution was a 5/5 on the level-of-change scale, a 5/5 on the effort scale, and a 5/5 on the impact scale.
Add a chatbot
We made a chatbot! (This was a lot of work, so it really needs a separate case study.)
Customers still tend to reach out to customer support even if answers to their questions are available in the FAQs section.
Adding a chatbot allowed ING to surface relevant FAQs in a user's chat conversation with the bank. This aligns with our guiding principle of providing support as fast as possible.
And, this was actually one of the main reasons why we needed to restructure our Help content. We needed Help content that was concise and reusable across the Help Center, the website, AND the chatbot.
ING Philippines' Live Chat functionality also integrates with FB Messenger. In the ideal end state, the chatbot is be able to surface FAQs in FB Messenger as well.
Finally, adding a chatbot allowed prioritization of customer issues or concerns by "filtering" inquiries that are answerable through the FAQs. With the chatbot, we could provide faster support to concerns like someone hacking a customer account over less urgent ones like inquiries about a delayed check deposit. This will help minimize the time to provide customer's a resolution.
This solution was a 5/5 on the level-of-change scale, a 5/5 on the effort scale, and a 5/5 on the impact scale.
- Sitting beside the customer care team and seeing what customers ask for help with is one of the fastest ways to learn about gaps in communication, design, and operations. Really good way to empathize with the customer care team too!
- Collaborate with designers, engineers, and product stakeholders (including the customer care team) early and often.
- Test early and often with customers, especially for high-effort features that take months to build. We want to make sure that what we're building merits the effort we put into it.
- Micro-interactions are extremely helpful in communicating small, important nuances—like switching from chatbot to customer care agent—which would have otherwise needed detailed communication in the app.
- Using highly-detailed prototypes in usability tests reduces confusion during tests, helps the user better understand the experience, and produces more effective insights.
- It takes significant effort to avoid bank jargon in communications. But it usually results in clearer, more effective communication. It's easy but often wrong to assume that customers would know what "KYC" or "fraud" is.
- For customer support features, it's important that the experience is fast and the communication is clear. For a frustrated customer, nothing is worse than not being able to find the help you need to deal with an issue.
- Having excellent customer support in a heavily-regulated industry is difficult but, when done well, can be a differentiator.