While integrating chatbots into customer service has compelling benefits — banks could save $1 trillion by 2030, according to Autonomous Research; plus, they enable asynchronous communication and are available 24/7 — research from Zendesk found consumers still prefer speaking to humans.
The good news is that most are open to being persuaded of chatbots’ benefits. So much so that, according to Zendesk, 70% believe they can make customer support more personalized and effective, especially if they use AI.
So what are bank chatbots currently able and unable to do? And what enhancements would bring them more in line with consumers’ expectations?
Here’s what the data from FinTech Insights, our digital banking research platform, told us.
The Great, the Satisfactory and the Basic: A Roundup of Chatbot Functionality
To explore the current state of chatbots, we looked at 149 national and regional banks, credit unions, and challengers in North America and Europe. Our sample consisted of 90 firms in North America (76 in the U.S. and 15 in Canada) and 58 firms in Europe (41 in the EU and 17 in the UK).
We grouped the chatbots in our sample into three categories:
Tier 1: The chatbots in this tier are the most basic — more akin to an interactive search tool than a full-blown chatbot. While some can answer simple questions, most are only capable of explaining how to perform a specific action or linking to articles which explain how to do so.
Tier 2: Alongside answering basic questions, Tier 2 chatbots can also provide the customer with additional information on their query, as well as linking to specific pages that might help.
Tier 3: These are the most advanced chatbots. Usually AI-powered, these chatbots can process natural language and handle complex queries.
Banking Chatbots in Numbers
Out of the 63.1% of firms in our sample that have chatbots, the majority, 38.3%, are Tier 1. 22% are Tier 2, while a very small minority, 2.7%, are Tier 3.
From a geographical perspective, the EU is most behind. 43.9% of EU firms don’t have a chatbot. Of the firms that do have a chatbot, around 39% are Tier 1 and 12.2% are Tier 2. Two banks have Tier 3 chatbots: bunq and KBC.
The U.S. — the location with the second biggest percentage of banks without chatbots (40.8%) — has just one bank with a Tier 3 chatbot — Bank of America. The majority of banks that have chatbots, 38.2%, are Tier 1, and 19.7% are Tier 2.
Canada also has one bank with a Tier 3 chatbot — TD Bank. It stands out from the rest of our sample for being the only location where the majority, 60%, are Tier 2, not Tier 1. Only 20% are in the latter category.
At the other end of the spectrum, the UK is the only country in our sample with no Tier 3 chatbots. The majority, 52.9%, are Tier 1 and 23.5% are Tier 2. Another 23.5% don’t have a chatbot.
Given these numbers, it’s not surprising that chatbots are particularly unpopular in the UK.
Research from RFI Global found customer satisfaction with chatbots in the UK is under 50%. Worse, only a minuscule 1% prefer chatbots to other contact methods.
The reasons for customer dissatisfaction are all rooted in poor user experience. Thirty percent blame their dislike on lack of personalization, 25% are worried about security, and 23% say their bank’s chatbot doesn’t have sufficient capabilities to make using them worthwhile.
What makes Tier 3 chatbots stand out?
So, what separates Tier 3 chatbots from Tier 1 and Tier 2?
We’ve taken a closer look at the four Tier 3 chatbots in our sample to find out.
Bank of America (Erica)
Launched in 2018, Erica is by far the most advanced chatbot in our sample and the most widely used, with 1.5 million customer interactions a day.
Erica has proved popular because it can perform several actions without forcing the user to leave the chat. These include making transfers between accounts or via Zelle, blocking or unblocking cards, searching for and viewing transactions, and performing personal finance management tasks such as analyzing and viewing spending by category.
TD Bank (Clari)
While not quite at Erica’s level, Clari is a step ahead of Tier 1 and Tier 2 chatbots because of its ability to answer complex queries without redirecting users elsewhere. Its capabilities include showing balances, recent transactions, and spending insights by category.
bunq (Finn)
Alongside showing spending insights by category and IBANs for different currencies, Finn distinguishes itself through its ability to offer non-banking support, such as recommending restaurants within easy reach of users’ last known location.
KBC (Kate)
Kate can process natural language and ask follow-up questions, but stops short of resolving queries within the chat, redirecting users to a relevant section of KBC’s app instead.
That said, it does have several useful capabilities, including transfers between accounts, blocking and unblocking cards, adding cards to Apple Pay and starting a loan or insurance application.
Banking Chatbots Must Step It Up
In an episode of RFI Global’s Banking Uncovered podcast, Standard Chartered’s global head of digital banking Mohamed Keraine observed: There is no value in digital itself, its value is in its ability to scale.”
To which we’d add: for scale to be truly valuable, it’s not enough to improve internal metrics like operational efficiency. Just as — if not more — important, it must meet those it’s intended to serve where they are.
As things stand, banks still have a long way to go when it comes to chatbots.
That customers can see the potential is half the battle. But, for a critical mass of consumers to become willing to use them, they must stop behaving like glorified FAQs and deliver on their promises.