Automated customer engagement solutions are not getting much love from consumers because many of them are simply poorly designed.
With most customer interactions shifting online pretty much overnight, brands have been scratching their heads a lot since the start of the pandemic to define what their online experience should look like, and most importantly, how they could bring a fresh touch to the customer experience in a digitally saturated world.
Making a long-lasting impression on consumers offline was already a challenge, and the equation most companies now had to solve, was about answering a large volume of customer queries, and differentiating their online experience from the competition, to engage consumers who were bombarded with brand signals, and couldn’t bear looking at a screen for another minute after days spent on Zoom.
Among some of the technologies brands have prioritised in transforming the customer journey are chatbots. Unfortunately, their implementation is often short sighted, and is being counterproductive, chasing visitors away rather than engaging them efficiently.
The human touch
As the world went into lockdown, customer services and call centres were quickly saturated with calls from worried customers which were not being answered as contact centres were being shut-down. Chatbots were added to company ranks to take some of the load and remove pressure from those teams. But there was an equally important objective to offer visitors something different and more innovative than another FAQ. When done well, chatbots are supposed to give customers an interactive experience that felt conversational and as close as possible to interacting with a human being.
Indeed, despite years of increasing their digital savviness, consumers, especially in Australia, are still very much in need of a human touch. We recently conducted a study looking at consumers’ preferences when interacting with brands, and half of respondents said they still prioritised face-to-face or phone interactions. The other half that mentioned digital channels, still want a human being on the other end, prioritising emails or live web chats with agents.
Unfortunately, consumer experiences which a large number of inadequate chatbots mean they are not convinced that our chatbot friends are succeeding in creating an engagement medium that is quickly and effectively solving their problems or queries. Only 2% of Australians said chatbots were their preferred method to communicate with brands, showing how inadequate many chatbot implementations are.
Where is the love?
A plethora of organisations claim that they deliver intelligent bots and assistants, but customer feedback shows that many lack the simplest conversational and behavioural design, thereby giving customers a disappointing experience and frustrating them with the brand and turning them away from engaging further in these channels.
A typical example is when chatbots systematically pop up as soon as visitors navigate on a website and only provide a very narrow set of pre-scripted options for enquiry . Do we want a sales person jumping at us every time we enter a store, and only allowing 3 choices of question? I don’t think so.
Unfortunately, many companies are being lured by cheap chatbot solutions and the promise it will modernise their customer experience, when it is only damaging it. The industry needs to improve their standards if they don’t want chatbots to become an annoyance customers want to avoid at all costs. Organisations need to look at chatbots as true business applications and not just “science experiments”.
Who’s a good chatbot?
So why are most chatbots failing to live up to our expectations? And what should companies look for in a good chatbot?
1. The AI is too basic to handle human language.
Most chatbots are very narrow it what they allow customers to ask. Any questions outside of the scripted options (if allowed at all) is often responded with a “I don’t understand your question, please select one of the 3 options….”. Few are a step further up the maturity curve and are able to recognise some keywords and customer intents, and deliver scripted answers. But for anything more complex than this, such as conversational and transactional self-service experiences, (which means allowing customers to ask their questions in the way they want) the quality of the AI and conversational design underpinning the solution is essential, and usually requires an advanced Natural Language Understanding engine, with deep neural networks and a complex dialog engine that can converse with the customer depending on what questions they ask and how much information they provide. In other words, an AI properly trained to understand natural human language and conversation and one that can adapt its responses based on how the conversation proceeds
2. There is no clear business objective behind the chatbot.
There are still too many companies that fail to clarify exactly what they want their chatbots to achieve before deploying them. What does success look like? Whether it is reducing customer service costs, or driving more sales leads, the answer to this question should be the blueprint for designing a chatbot. Also what data and measurement systems have they put in place to measure that success or otherwise of the chatbot? ow to these feedback mechanisms influence the chatbot optimisation and future capability. What is the optimisation strategy and how quickly can your chatbot adapt?
3. Customer preferences are not factored in the design.
The way customers interact with a brand is an ongoing source of insight that should guide chatbots’ design and capability before and after their deployment. This ongoing analysis allows us to identify roadblocks and pain points in the experience and bring necessary changes to the bot’s design. For example, if customers are constantly coming to a chatbot with needs it is not designed to address, how do you quickly assess and adapt to meet that customer need?
4. The chatbot isn’t adapted to the user journey, channel or platform.
Interacting with customers on a website, via a smart speaker, or on WhatsApp is different, and replicating self-service interfaces from one channel to the other is complex. Whether it works on a website, for example, doesn’t translate to a voice channel. Think about the ability for the customer to share a picture to help get their issue addressed. Not all channels allow this. Efficient chatbots have custom designs for each channel that factors in each platform’s specificities and how customers engage in them.
5. The chatbot isn’t designed for authenticated use cases and fraud prevention.
Online fraud and scams have increased significantly since the start of the pandemic, and it is every brand’s responsibility to give their customers the confidence that they take their privacy and security seriously when interacting with them. Smart chatbots, digital or voice-based, now embed automated authentication features that allow them to automatically identify a customer based on their voice, the way they type, swipe or hold their device when interacting with them. This “ zero effort” authentication based on behavioural biometrics can ensure customers’ protection whilst keeping the customer experience intact.
A properly trained chatbot can do marvels to the customer service and experience. One example of a chatbot that ticks all of the above is Commonwealth Bank of Australia’s award-winning virtual assistant Ceba, which has evolved instantaneously understand and answer 450 types of customer enquiries—that is, 90% of customer enquires—and approximately 70,000 different questions since its deployment in 2018.
The necessary strategic thinking and intricacies of building a successful chatbot or virtual assistant shouldn’t be underestimated. If they are, organisations will end up with dumb chatbots that can’t solve the simplest of customers queries or engage them efficiently, and in the worst cases will frustrate them and impact negatively on the brand. Once customers have had a bad experience with a chatbot, it is a long time before they will try again.
Excellent digital experiences are now more common and as consumers’ experience these their expectations of other brands increase. Their tolerance of poor customer experiences fades to a similar extent.