Like it or not but chatbots are coming and for good. They will automate or at least reduce many of our day to day work. Banking is one of them. I don’t remember when I have to physically go to bank’s branch for my routine banking stuff after I get the app in my phone. How easy it is now to transfer money, pay utility bills, manage stocks and what not. But all this is about to change, just how apps reduce our effort, bots are going to make the technology disappear from our naked eyes.

Similar article: Make your chatbot fun to talk with, the right way.

But chatbots are not new, they are within us for years. Then why suddenly so much talking on it. Because technology went out of the way and developers now can focus only on the actual business logic of developing a bot. Microsoft bot framework, Facebook Messenger bot, Telegram bot, Motion AI and list goes on. Usually building a chat bot requires creating or integrating with chat app or communication channel, integrating Artificial Intelligence to understand and respond to user message, take action based on user intent and continually improving the knowledge base to server more and more user queries. This wasn’t easy until all these big players started offering bot frameworks.

Talking about how bots will change banking sector, consider you want to send money. Typically you will follow below steps.

• Open your bank’s app
• Login to your account (typically a pin or more tedious two factor authentication involving credentials and then entering code received in SMS or Email.)
• Understand navigation to reach Transfer Money page.
• Enter payee details and amount.
• Submit request.
• Confirm it.
• Receive multiple annoying notifications both as SMS and Email. Yikes.

Now consider the same happening inside Whatsapp, Skype or Slack. You just open app and say “Send 2000 bucks to Joe”. How easy, isn’t it? Bot must be able to look at my payee, locate Joe and from my primary bank account (if there are multiple) send him 2000 bucks. Saves me from tedious app navigation.

Banking Bot

As I said before due to internet banking and mobile banking we no longer need to visit the branch, in fact in my opinion physical branches are no more required, as chatbot can automate most of the task and it is easy for customer to interact with is as well due to natural language communication.

Bot will simplify bank’s digital presence, hence encouraging more and more customers to fully utilize it. Banks don’t need to develop all different applications, banking app, mortgage calculator, stock portfolio management. Everything can be combined in single interface and when new modules are added, it don’t require app update, or publishing to App Store, making whole development life cycle easy to maintain.

Following are few banking stuff chatbot can handle now.

Customer Alerts: Many banks already send app notifications of important transactions, but this can be simplified using chatbot. User can ask for daily account balance to be received as message in their choice of messenger. Same way any important message like low balance alert, debit alert, credit card payment alert, pin requests, all these can be done through chatbots.

Money Transfer: Money transfer is one of the basic functionality of bank. As of now we do it from app which requires significant steps to complete. This can be replaced with easy to follow chatbot messages as we discussed earlier in this article. User just send simple message “Send 2000 bucks to Joe” and that’s all it takes.

Deposit Checks: One of the reason we need to visit physical branch is to deposit checks. But chatbot can help with that as well, you just send image of the check and cognitive api from Google, IBM Watson or Microsoft Azure can help understand check image, amount needed and account to which you want to deposit into. As always there are security measures which needs to be taken care of like authentication, mistake in cognitive API, user generated errors and privacy, but thanks to Microsoft Bot Framework and Facebook Messenger API, we can address many of these concerns relatively easily.

Inquiries: Every now and then user opens banking app to figure-out his balance or check last transactions to see whether he received money from specific account or not. Let’s say on every month, user checks whether salary is credit or not. For this simple inquiry user needs to login into app, navigate through few screens to see last credit transactions and logout. Rather user can use bank’s bot on Skype by saying, “Is salary credited?” and bank bot can respond with salary amount if it is credit by user’s employer. And to support new inquiries bank don’t need to update app and as bot will reside inside Skype or similar channels (Facebook messenger, Slack, Kik, SMS etc) they don’t need to worry about the UI/UX part, neither they need to manage user sessions, once account is linked to a Skype or Facebook account then user can make inquiry without authenticating again and again.

FAQs and Search: Let’s face it, it is really hard for user to find answers to most of their FAQs due to nature of website search and most of the time banking app lacking FAQ section. But it’s totally opposite in case of chatbot. User can simply type a question and developer just need to configure FAQ backend to the bot’s search module, and that’s it. Match made in heaven. Not only user get answers to their problem right away, bank also get to see every request that comes in and if there are good enough request for a question and there is no relevant section in FAQ then bank can easily add relevant answers. Also user have chat history where he can see previous responses making it easier to find information when needed.

Content Delivery channel:Banks can use chat bot as an effective content delivery channel. Let’s face it, financial education is not an interesting topic and most of the users won’t search about it until absolutely needed. So banks can use chat bots to send messages (and not notifications, we are getting good at ignoring notifications from app) according to their past inquiries, taking into account what are their current financial goals. This is the best way to feed user’s preference back into machine learning algorithm to find persona of a user and then deliver highly personalized content to them making it less annoying. Say customer is searching about different credit cards then sending them content about interest rate comparison of different cards would definitely grab their eye ball.

Customer support: It is almost natural to pick a phone, open a messenger app and talk to someone, then why not with customer support agent from bank. Many banks do have chat based support system. But problem with this approach is that bank have to train many agents to fully support this system. If banks replace human based chat system to bot based, they will not only effectively reduce cost but also human error in responses they give. Also this will standardize the process so that it gives exact same response for same query. Most frustrating part in human based support system is that user needs to wait for their turn to come in and for a large bank this can span upto 10mins before someone can attend user.Switching to chatbot will remove the wait time altogether and this can be biggest factor in adoption of chatbot for support systems.

Offers:What a great way to send offers on product customer is using or searching for than sending them a personalized message. There is almost guarantee that customer will see these messages as they normally use messenger to talk with their friends, family and business contacts.

These are few of the ways banks can leverage chat bots to server customer in better way. Believe it or not, chatbot will be future of banking and sooner or later banks have to build their own bots, just like internet banking and mobile apps.

How bots benefit banking

Chatbots will have big impact on the banking sector, as we already saw few benefits in article so far, below are few more significant benefits.

Superior user engagement: Using communication channels like Skype, Facebook messenger, Slack, kik, telegram to engage customer will become easier for customer to reach to banks easily. Now they no need to worry about automated call answering machines, ambiguous menu transversal, waiting for support person to get hold of it, difference in timezones, different business hours or festival holidays. On the other side banks will benefit by reducing overall response time and reduced human operator cost. Customer experience will be as simple as typing something and sending it to bank, while bank will take care of understanding it and acting upon.

Better customer understanding: As messenger data can be stored and later processed through machine learning algorithm by data scientist, will reveal new hidden facts about customers. Which can then be feed back into system for more engaging and simpler systems. As a human (and soon by machines) it is easier to understand intend of a customer from plain text messages rather than reading telemetry data.

Reduced IT costs: Developing and maintaining a chatbot will reduce overall IT cost, as it is easier to develop support for new inquiries than building same feature on website/app and then deploying, testing and maintaining it.

Similar article: 6 Precautions to take before starting you next bot project.

Challenges with bots for banking

If we don’t talk about the other side of the topic it won’t be considered complete. So here are the few hurdles in implementing bots for banking.

Authentication: Internet banking have significant stronger authentication mechanism, now to add complexity to it bot needs to be authenticated that too from other systems (messengers). Where Facebook Messenger don’t provide any authentication mechanism, Microsoft Bot Framework does provide way to integrate authentication and authorization. Still authentication is in early stage to be used in banking bot.

Still these limitations are not deal breaking, and can be worked around with some technical fiddling. And as per my knowledge soon all bot framework will provide a way to do authentication and authorization natively.

Natural Language understanding and user error: Computers are learning faster than ever before, they can recognize almost every sentence, image and object. But it is still machine, so it won’t be foolproof. Bot needs to be taught how to understand a message and then act upon it. This means every possible combination of a sentence have to be feed into system before it can be used. There are alternative to this approach which doesn’t require every possible combination before hand, just few examples and bot will improve itself as it sees more messages. Either way it is not easy for an experienced machine learning expert to program it for everything that is possible inside banking app. Also user may type wrong message or a message which can have two or more different meaning.

All these problems can be addressed. User errors are easier to fix using some UI pattern like showing a message what bot is going to do with the last message (only for messages having DML statements). The other problem with NLP will get resolved as time passes by and more better cognitive system takes place. Even today we can implement fairly satisfying NLP using Microsoft Cognitive service or Google Prediction or IBM Watson or Amazon’s service.

What's Next

As users don’t need to download or install new apps or softwares, adoption is going to be skyrocket for chatbots. Bots will be most disruptive that any past software or hardware device. Bank of America has been already planning for launching bot with Facebook messenger as they said in their press release (http://newsroom.bankofamerica.com/press-releases/consumer-banking/bank-america-collaborating-facebook-help-clients-stay-connected-thei) “We want to serve clients wherever and whenever they choose, and Messenger is a perfect complement to our connected banking proposition through 2016 and beyond.”

Want to jump on the bot wagon but don’t have enough resources, let us help you build bot for your business and get early adopter advantage over your competition. Contact us at chatbot@veloxcore.com.

Like reading our article and don’t want to miss next article in the chatbot series, subscriber to our newsletter below.