3 Ways That Chatbots Fail
Unless you’ve been hiding under a rock for the past two years, you’ve probably either heard of or tried a chatbot. Whether on Facebook Messenger, Skype or WeChat, you’ve most likely encountered at least one across myriad messaging platforms.
In a customer first era, chatbots a great way to automate more routine tasks efficiently. They’re also really good at delivering a consistent customer experience. But, as with all things, the devil’s in the detail, which is why it’s worth understanding where chatbots fall flat.
The promise of chatbots
Chatbots enable one-to-one conversations with customers through automated interfaces – and are a great way to deliver immediate responses. Depending on the queries you’re dealing with, there are essentially two types of chatbots to choose from: those using pre-programmed scripts and decision trees (that can be adapted easily to new questions and scenarios), and others that are integrated with AI (artificial intelligence), machine learning and natural language processing (NLP).
3 ways chatbots miss the mark
With efficiency front and centre, chatbots have the potential to add real value to your customer experience. But before you race to implement a chatbot in your business, it’s worth weighing up the benefits and risks carefully. Chatbots can enhance your customer experience, but because ‘smarter’ bots are still fairly rare, ‘pre-trained’ bots are only suited for particular scenarios and only as good as the technology and design behind them. So where do chatbots fail?
From a practical perspective, it means using market intelligence through data. By implementing tracking systems across every customer channel and touchpoint, organisations will be able to share information across the business to deliver a consistent experience and enable powerful internal shifts in culture.
1. Understanding
Because chatbots use open source libraries, most won’t be customised to your specific industry and customers. Pre-trained bots will be limited to their pre-programmed decision path and are limited by your designer / programmer’s understanding of your customers’ behaviours and requests. While chatbots don’t reason, smarter bots can cope better with some language nuances. At the same time, without human judgement, chatbot accuracy will always be limited.
2. Context
Pre-trained chatbots follow a structured conversation plan and can lose the flow fairly easily. With more access to customer history and data, smarter chatbots can ‘learn’ customer preferences. To keep context, chatbots need every possible response to every possible customer request.
3. Consistency
Chatbots work well if customers start their conversation as a chat, before being transferred to a human agent if needed. The problem comes in where customers have started a conversation on another channel such as the phone or email. Without consistency across all your channels, chatbots will force your customers to start over – and risk damaging your brand and customer experience.
How to get chatbots to work for your business
Merchants UX & Collaboration Solutions Architect Jean-Louis Viljoen explains that any virtual assistant technology solution has the potential to help improve the customer experience and deliver efficiency through automation. But, Viljoen warns, you need to be able to adapt to change, ‘Whether it’s a pre-trained or smart chatbot, you’ll need a clear implementation plan, with a future roadmap to manage its evolution in line with your evolving customer experience.’
Before you make the most of any chatbot solution, he advises ensuring it’s fit for purpose and spending time on:
- Design – make sure it’s enhancing your CX and brand.
- Scope – rather than be tempted to get your chatbot to perform many tasks, focus on performing simple tasks well.
- Consistency – ensure a consistent level of service across all your other customer service channels.
- Personalisation – if you’ve chosen a pre-trained chatbot, keep conversations as human as possible and try to integrate with customer data for an improved personal touch.
- Testing – spend time making sure your chatbots are able to handle real customer requests before launch. Nothing annoys a customer more than being rerouted or forced to start again.
- Failure handling – when chatbots fail, make sure you can escalate the chat to a human quickly.