Banking Chatbots Need More ‘Advisor’ and Less ‘Robo’

When it comes to chatbots, banks and credit unions think the technology is tricky but underestimate the importance of designing quality conversations. Don't let a new chatbot with the wrong tone undermine your experience and hurt your brand.

Today’s customer expects seamless service. Chatbots increasingly seem to promise that — they’re fast, knowledgeable, always on. But when it comes to customer service, chatbots are making the financial services industry seem more faceless and distant than ever.

It’s estimated that by 2024 the chatbot market will be worth an incredible $1.34 billion. It’s no wonder 80% of financial services brands see these robo-advisors as an opportunity.

Chatbots are everywhere — even in mental health services where the Woebot, founded by legendary AI superstar Andrew Ng, received another $8 million of funding last year.

Chatbots can offer enormous cost savings to the financial services firms that use them. So, what’s not to love?

Quite a lot.

In the digital age, speed, knowledge and availability is the floor of acceptability. What differentiates a bank or an insurer is humanity. Or at least the illusion of it.

Read More:

“One global bank’s chatbot has a creepy avatar. Its responses are corporate speak.”

I recently used a global bank’s chatbot. It is proof that giving a bot a woman’s name and a (decidedly creepy) moving avatar do not make it seem any more human. Its responses are corporate-speak — slightly impenetrable and often dull. And when it can’t answer, it throws in a saccharin apology so at odds with its usual tone that it seems wholly insincere.

Technology isn’t the problem. The chatbot’s tone is. This is why financial institutions should make “conversation design” their biggest priority when developing a chatbot.

If you want to make a successful chatbot, there are two major considerations: task and tone.

With any task, you need to set a user’s expectation for what the bot can and can’t do. So, a good rule to adhere to is the “POD” rule. Only give bots tasks that are:

  • Predictable. If it’s a task that recurs again and again within your organization, and there are more often than not a set number of outcomes, then it’s a probably a good idea to delegate it to a chatbot.
  • Outsourceable. Would you be comfortable outsourcing this task to another company? If it’s a transaction with high-value or that needs a lot of oversight it’s best to leave it in the hands of a human.
  • Defined. Only give bots a task that has clear parameters and an obvious exit point for handing over to a person.

Four Factors Point to the Right Tone

Aside from task fulfilment, it’s language that decides whether an interaction with a bot is positive or negative.

Certain rules apply to conversation design no matter what your brand is. We think that the “Cooperative Principle,” developed by renowned philosopher of language Paul Grice, is the clearest. It’s what Google has used in developing its voice and text-based chat functions.

Four maxims govern a successful conversation design, according to Grice:

1. Right Amount of Data. Your institution’s bot should provide enough information to carry the conversation forward and serve the implicit or explicit purpose of the interaction. But no more than that. Several money management bots do this well. Cleo and Plum, for example, provide prompts to guide users through the interaction.

2. Quality Advice. Your bot shouldn’t lie or mislead the user.

3. Only What’s Relevant. Don’t include unhelpful information or steer the conversation away from the user’s need in that moment.

4. Adjust Manner as Appropriate. Be clear and simple in your language and structure and adjust your bot’s tone depending on the situation. The disruptive insurance app, Lemonade, does this cleverly by switching between bots. One bot, Jim, deals with sensitive issues like claims and another, Maya, deals with more general inquiries. One can be calming and concerned while the other can be more jokey.

Getting the task right and setting expectations build the foundations for a satisfactory interaction. But what really sets a brand apart is creating a voice that makes your customer feel valued.

Read More:

What kind of ‘Warm’ Do You Want?

It’s no good having a UX team if they’re all moving in different directions. Clear and memorable voice guidelines are essential, ones that give designers the freedom to build in small empathetic touches while still staying true to the brand’s personality.

It’s also important for the bot designers to understand the institution’s DNA. What are the traits that set a bank or credit union apart? And how can we pinpoint them as accurately as possible? Yes, you might want your brand (and by extension your chatbot) to sound “warm.” But what kind of warm? Reassuring? Optimistic? Paternal?

Once you’ve focused on a few key tonal qualities that accurately reflect the brand, it’s time to bring it to life. We think of a brand voice on three different levels:

  • 10,000 feet — your overarching narrative. These are the things the financial institution, stands for, and stands against.
  • 1,000 feet — the personality. This is where those adjectives really come into play.
  • Ground level — the real detail of what you say, from word choice to sentence structure and length.

Consistency across those three levels is crucial. But to really achieve that, writers need training. Expansive verbal guidelines are helpful, but only if they’re not left in a desk drawer. Supplement brief and practical guidelines with real life examples and workshops that let your writers get to grips with your new voice.

Quickly, it will become clear that value doesn’t just lie with what you say, but how you say it.

Many of the issues with the global bank’s chatbot I mentioned earlier, for example, arose from a problem with consistency. It was obvious that there were a team of writers bringing in different styles without a central brief to guide them. Sometimes there was a hint of personality, sometimes there wasn’t, and the contrast between these different modes made not just the chatbot, but the bank look sloppy.

Clear parameters would have helped massively. It would have forced their writers to stop and consider:

  1. Does this reflect the bank’s narrative? In this case, is it something a globally minded bank, who is for thinking big and connecting customers to opportunities, and against insularity and small-mindedness, would say?
  2. Does this sound like the help or advice someone ambitious, authoritative, pragmatic, etc. would give?
  3. Are the individual words and structure consistent with that?

Ultimately, chatbots can be a great way to deliver a bank or credit union’s brand promise to customers. It’s another touchpoint that’s only going to grow in importance. But only if institutions keep humanity at the forefront of this technology.

This article was originally published on . All content © 2024 by The Financial Brand and may not be reproduced by any means without permission.