8th July, 2020

Chatbots UX – friend or foe?

by Larry Brangwyn

Recently, a client asked us to explore the possibility of adding a chatbot to their website to handle incoming booking requests. Having extensive knowledge of planning and managing live chat experiences for our clients over the years, we decided to throw ourselves into the task and explore the possibility of using this exciting new technology. Because everything new has exciting potential, right?

We approached this from a number of angles.



The chatbot immediately tries to emulate a conversational dynamic, by breaking down the exchange into micro-transactions and funneling users by categorising their enquiry.  It also often tries to emulate a person by introducing humanistic phrasing and a credible delay to user input and bot response, just like seeing 3 dots when you’re waiting for someone to reply to your iMessage. 

This isn’t entirely unreasonable, but it’s a far cry from live-chatting with a human being, where the primary question is open – how can I help you?

As current chatbot technology isn’t capable of interpreting complex and unexpected requests, the developers tend to try and corral users within the bot’s areas of understanding – getting the necessary information piecemeal in a question-by-question exchangeWhen it comes to asking an open question, the natural language programming often can’t handle the input properly and reverts to a default option list, or gathers your contact details so a human can follow up with you. This issue leads nicely to the motivations of the chatbot.



Our test of different chatbots highlighted that as well as bulking out the conversation with questions, the bot often tries to collect user data under the guise of personalisation, and of contacting you in case your chat session is interrupted. While this is arguably reasonable, it’s not always proportionate to the nature of the request. 

Firstly this is usually in the form of two additional, personal questions (your name and email address) that are often requested before you even tell the bot what you’re after. A lot of questions don’t genuinely require this information in order to answer a general enquiry, and run the risk of alienating users who don’t want to hand over personal information. The more cynical of these users might think this is a clumsy attempt at lead capture and an opportnity for companies to spam people with marketing messages down the line.

This may be a feature that is offered by chatbot developers to entice marketing teams and decision makers to invest in the platform, but to counter any mistrust from cautious users, this should be opt-in rather than a feature the users must use in order to continue.



The above issues put the spotlight firmly on the primary reason for engaging in chat – to get answers to your questions or to get the company you’re contacting to do something you need. When the somewhat cumbersome and artificial nature of a chatbot conversation is combined with the need for additional steps to compensate for the chatbot’s limited abilities and the conflicting priorities of the technology/company vs the user, the end result is that it would ultimately be quicker, easier and less stressful for some users to just phone a company and get an answer/result in real-time.



The other angle is the urgency of the request – a lot of chatbots balance out the technology with human teams, helping to triage the requests and handing off when necessary. It sets up the expectation of being able to resolve an issue quickly, but doesn’t divulge upfront that the humans might be offline (asleep). A user with an urgent request might get a few questions into a chatbot before learning that the bot can’t assist and the humans won’t be back until 9am. This doesn’t answer the user’s query and adds wasted time into the bargain.


The verdict

Unfortunately, new technologies get instant hype as people clamour to find the ‘next big thing’ and this finds these technologies implemented too quickly without enough thought put into the use case. There haven’t been any groundbreaking paradigm shifts in the web in the last couple of years. In fact, we as human beings are playing catchup with technology.

The reality is that chatbots are trying to replicate the experience of talking to a person by way of periodic conversational exchange, but are falling far short of being fast, useful or convenient. In some cases they actually make things more difficult and time comsuming than they need to be.

We’re sure there are realistic use cases for this technology, but without significant customisation and interaction planning, the average website isn’t one of them.

If you’re interested in using Chatbots on your website, or want to augment your business with digital automation, we can help you plan and select useful technology solutions.

Tags: AI, Automation, Chatbot, Machine Learning

Follow us on Twitter for more

Want our musings in your mailbox?

That’s great! Complete the form below and we’ll occasionally send you news and ideas. We send updates infrequently and we hate spam.