Everything you need to know about chatbots


So firstly, what is a chatbot? In simple terms, chatbots give customers a live online chat experience that will enable them to ask questions using normal, everyday language and receive an answer that correctly responds to what was being asked.

Powered by artificial intelligence (AI), and specifically a subset of AI called natural language processing, the bot is able to understand the intent of a question by breaking it up, identifying the meaning and responding with an answer.

Over the last 10 years, natural language processing has improved to such a degree that adoption has been much more viable and additionally the barriers to entry have become a lot lower. As a result, more companies are using chatbot technology across industries.

The responses that are given to the user can be from a predefined list of scripts and phrases via machine learning applications and external data sources, which can include your internal systems.

Most common applications include answering frequently asked questions and helping users answer account specific questions.

Supporting staff

Chatbots can help businesses, not only by managing staffing resources more effectively by giving them time to help customers with more detailed needs or supporting those who are more vulnerable, but they can also help businesses manage peaks of enquiries, for example when Covid-19 Payment Holidays were introduced.

They are also multi-channel, so you can utilise them on your website and integrate them with chat platforms like Facebook Messenger, plus they can be plugged into voice assistants, such as Alexa and Cortana and others.

User experience

So what are the benefits, aside from managing peaks? Well, first of all, your customers can have a much better user experience – they can get an instant response, at any time on any day without having to pick up the phone and call your organisation or search out answers on your website.

But also, chatbots enable you as a business to have a much better understanding of your users on an ongoing basis.

The second you launch a chatbot, it’s going to be gathering a multitude of information by logging every query that’s coming through.

You’re going to understand on a much larger and more detailed scale, exactly what your customers want from you as an organisation and in real-time.

This can have benefits, not just for the team that is looking to optimise and refine the bot after launch, but also other areas of the business that can benefit from this data.

Chatbots in action

If we look at the financial services sector, there are a number of uses and examples of companies embracing this technology.

A prime example would be NatWest which launched  its Cora chatbot a number of years ago. I’ve actually interacted with it as a customer and can report having a good experience.

Effectively, it answers account queries that don’t need human intervention. A little nearer to home was a project that we helped launch last year which was for Kensington Mortgages to help mortgage brokers asking criteria questions.

This delivered quite a big business benefit and Kensington spoke at our mortgage transformation event last October and publicly shared that after launching the chatbot they had a near 20% reduction in queries over the phone related to lending criteria.

How do you get started?

A chatbot development project is no different from any other digital development project. So if you’ve been involved in building a website, portal or digital platform previously, you will be familiar with the process.

Firstly, you go through discovery and research, then design and prototyping and finally implementation, testing and launch.

Tips and lessons learned from projects

  • Start Small:

Obviously you want to launch a quality product, but I am a big believer in starting small. If you identify and focus on those low effort areas, maximum impact options first, your minimum viable product is very likely to deliver some tangible business benefits fairly quickly.

By identifying the right use cases for a chatbot, there’s an opportunity here to deliver some bang for your buck, some actual value, reasonably quickly, which can be really great at getting buy-in and building momentum for future phases.

  • Listen to your users

From the second your chatbot launches, it’s going to be logging all of the queries coming in and you’re going to be gathering a tonne of really interesting and useful data.

This should be used to regularly evolve and improve the chatbot, but also as I alluded earlier, there could actually be some really interesting information here that could be utilised by other parts of your business.

  • Have an open mind

So, it’s an exciting technology and there is opportunity here. But you need to ask yourself early on in the discovery process, whether this use-case is a genuine fit for the technology, or would it be better suited left to human interaction.

Poorly executed chatbots can have a negative impact on your customer experience, so you need to choose carefully. 

  • Expect different responses from different users

You also need to recognise that different cohorts, or different personas of users, within your customer base may respond differently to a chatbot.

We’ve seen in testing that particular groups of users are much more open to using bots than others. There are those that are more used to using conversational interfaces. For example, certain generations are a lot more used to using WhatsApp and Facebook messenger etc.

Therefore, you need to anticipate this and additionally, you should not expect a bot to solve all problems for all users.

During the research phase, for example, you may have identified that your password reset functionality is really hidden on your site and it’s causing lots of problems at the call centre by people calling in.

Actually, that could be solved and supplemented by a bot helping point users in the right direction or undertake that password reset for them.

But separate from that you should take those learnings and look to improve those other platforms, off the back of that research.

‘It doesn’t need to be complicated’

So to wrap up, a key takeaway that I am looking to communicate is that it doesn’t need to be complicated. We have helped some of our clients identify use cases and build bots within a number of weeks.

For those bots that don’t require systems integration, you can deliver a lot of benefit fairly rapidly using this technology and with the barriers to entry so low now, it certainly is a good time to at least consider a chatbot for your business.

Mark Lusted is the managing director of Dock9