Some conversations are dredded (like the birds and the bees) while others are nonsensical (like when you’re playing with your 10-month-old kid). Then there are conversations that have been ongoing for decades, increasing in intellect and sophistication. This last type could apply to two philosophers of different schools. Alternatively, it could be applied to spoken conversations of an artificial flavor, due to the exponential rise of robots… well, chatbots to be more specific.
They came in the darkness of night, filtering through our internet cables, and squeezed themselves onto the bottom right corner of our screens. A second theory (possibly more plausible than the first) is that ecommerce has greatly benefited from ever-increasing AI development, and the expanding intelligence — and ballooning benefits — of chatbots has meant that by the end of this year 80% of companies are planning to be using them.
And the benefits of using bots are clearly labelled on the box: they are available to work 24/7; they can provide a customer with a quick and comprehensive answer; they can resolve problems; they reduce the cost of customer support, etc.
One ecommerce platform that works with the brainiest of chatbots is Salesforce Commerce Cloud (Demandware). Its bot-building capabilities are constructed via its Einstein artificial intelligence platform that is built into the platform.
SFC chatbots allow you to interact through a conversational interface and make use of natural language processing within the platform’s Einstein intelligence. This uniquely means that you are able to train the bots to pick up the intent of the customer — because as humans we are not always spot on with our use of language.
Yet before we get into the mechanics and benefits of Einstein bots, let’s take a look at its limitations.
Does. Not. Compute
They are certainly not here to do everything. To parallel the fear of many communities about foreigners coming and taking people’s jobs, chatbots are not here to replace humans. They are here to boost productivity by freeing up space for customer service agents to focus on tasks that require human input.
Despite the advancements of AI, chatbots have yet to pass the Turing test — the point at which AI is able to maintain a conversation that is indistinguishable from human conversation. So it is incapable of long and complex conversations with users. Don’t expect your enquiry about a pair of jeans to delve into the social impact of denim, or whether they are culturally appropriated from the working class. But if it’s specific problem-solving you’re after, that’s their daily grind.
The great benefit of chatbots is that they can resolve customers’ issues and queries in a smooth and fast process. Do you need to check the availability of an item? Or change the shipping address for your order. The bot will retrieve the relevant data and instantly respond. These are the repetitive, day-to-day tasks that you don’t need your customer service agents to undertake. That type of work would suck the life out of a section of your team.
And because bots lack certain flaws of the natural world, they don’t feel the effects of insomnia, so they’re available to work all day, every day. If you’re a company in one time zone, but have customers across the pond shopping on your website, and they want to ask something, instead of waiting 8 hours to put their question to you (or email you and wait hours for a response), they can easily ask a bot who is wide awake and ready to answer.
Chatbots on Salesforce Commerce Cloud websites are the first responders on the scene, aiding users in their needs. But if things get messy, and the convo gets clogged, the bot can transfer to an agent. For example, if a request is too complex, or the sentiment expressed by a user is poor.
Within the platform there are four parts to build a functionable bot. First is the dialog, which is how it interacts with the customer. You can create a welcome message that pops up in the bottom corner of your site. You can decide whether you want to be transparent that it is, in fact, a robot and not a human, to signify clearly to customers what type of interaction they are about to engage in. Perhaps you want your bot to be less intrusive, and quietly wait in the corner for customers to make the first move. It’s all up to you and your style of interaction.
Another part is the intent, which is what we spoke of earlier about the Einstein AI capabilities, and how it interprets what the user is trying to say or ask (A customer types “Where is my order?”). The third part, called entities, are categories you want to collect data for, such as a customer name, date, product number, email address (The bot replies “Let me see! What is your order number?”). And finally the fourth part, variables, acts as a vessel for capturing this data, storing it, and using it in conversation (When the customer replies with the order number, the bot retrieves this data from the Salesforce database, and responds “Your order was shipped to the US 3 days ago”.)
And importantly, any changes that you make on the bot builder can first be previewed before making it live.