What it means
Early chatbots followed fixed scripts, offering menus and matching keywords to prepared answers. Newer ones use artificial intelligence (software that learns patterns from large amounts of text) to understand ordinary language and respond more flexibly, though they can still make errors.
In a business, the appeal is cost and speed. A chatbot can answer thousands of routine questions at any hour, which reduces waiting times for customers and frees staff to handle harder cases that need judgement.
Finance teams meet chatbots in several places. Banks and payment firms use them for balance enquiries and card problems, finance departments use them as internal helpdesks for expense and policy questions, and advisory firms use them to collect information before a human meeting.
Good deployment depends on three things: a clear scope, reliable data behind the answers, and a smooth path to a person. A chatbot that tries to answer everything and cannot escalate will frustrate customers and may create complaints.
Risk and control matter in finance because wrong answers can have consequences. Organisations need to protect personal data, keep records of conversations, test responses for accuracy, and make sure the bot does not give advice that requires authorisation.
The business case should be tested with numbers. Compare the cost per contact handled by a person with the cost per contact handled by the bot, take account of the share of enquiries the bot can fully resolve, and include the cost of building, running and maintaining the system.
In practice
Real-world examples.
Example
A neobank adds a chatbot to its app so that customers can freeze a lost card, check a payment and find a branch or fee. Simple requests are resolved in seconds, and complex disputes are passed to an agent with the conversation attached. Waiting times for human staff fall noticeably.
Example
A multinational's finance team launches an internal chatbot to answer questions such as how to submit an expense report or what the travel limit is. Employees get instant answers from the approved policy. The accounts payable team spends less time answering the same emails.
Example
A wealth advisory firm uses a chatbot to collect a prospect's goals, age and savings before the first meeting. The adviser starts the meeting with a summary and spends the time on advice. The chatbot is clearly labelled and does not recommend products.
Formula
Calculation
Monthly saving = Contacts resolved by the bot x (Cost per human contact - Cost per bot contact)
Suppose a payments company handles 20,000 customer contacts a month, and each human-handled contact costs $5. A chatbot can fully resolve 60% of contacts at a cost of $0.50 each. Contacts resolved by the bot = 20,000 x 60% = 12,000, so the saving is 12,000 x (5.00 - 0.50) = 12,000 x 4.50 = $54,000. As a check, the old cost was 20,000 x 5 = $100,000, and the new cost is (12,000 x 0.50) + (8,000 x 5) = 6,000 + 40,000 = $46,000, which is a saving of $54,000 before the cost of building and maintaining the bot.Case study
Seen in the real world.
Quickpay Remit is an illustrative, fictional money transfer company that received about 30,000 customer messages a month. Most were requests for the status of a transfer, and its small support team struggled to keep up.
The finance director approved a chatbot connected to the transfer tracking system. She set three rules: the bot would answer only status, fee and document questions, it would pass any complaint or fraud concern to a person at once, and its answers would be sampled each week for accuracy.
Within three months the bot resolved just over half of all contacts and average reply time dropped sharply. The illustrative lesson is that a narrow scope, strong controls and an easy route to a person produced better results than a bot that tried to do everything.
Watch out
Common mistakes.
- Launching a chatbot without a clear scope, so that it tries to answer questions it cannot handle and frustrates customers.
- Counting only the savings and ignoring the cost of building, training, monitoring and updating the bot.
- Failing to provide a quick route to a human, which can turn a small query into a complaint.
Questions
People also ask.
Can a chatbot give financial advice?
Only if it is designed and authorised to do so under the relevant regulation, and many firms limit bots to general information and process help for this reason.
Are chatbots safe for sensitive financial data?
They can be, provided the organisation uses strong security, limits what the bot can access, and complies with data protection law, but this must be designed and tested.
How do I measure whether a chatbot is working?
Track the share of contacts resolved without a human, customer satisfaction, escalation rates and the cost per contact, and review a sample of conversations for accuracy.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%Related