Chatbots are conversational agents; software that is designed to help achieve a task using natural language. By understanding your query; whether a review, concern, question or complaint, chatbots can provide the answer to you or your customers in real-time. They are widely used for customer service, being able to answer questions pertaining to order updates, refunds and general QA. More advanced systems, such as those employed by Dominoes and Booking.com also allow you to order/book directly without leaving the chat interface. This is what we refer to as “conversational commerce”.
Conversational commerce is a term coined by Uber’s Chris Messina in a 2015 piece published on Medium. It refers to the intersection of messaging apps and shopping. Meaning, the trend toward interacting with businesses through messaging and chat apps like Facebook Messenger, WhatsApp, Talk, and WeChat. (credit: Shopify Encyclopaedia – https://www.shopify.com/encyclopedia/conversational-commerce)
Conversational marketing is the word | Data Driven Investor
Before making a purchase, talk. Marketers who do that will be ahead of the game. That's the premise of what Paul Ince…
Chatbots can be hosted as stand-alone web app interfaces, on chat apps such as Facebook Messenger or Telegram, or on your own business app.
These aren’t difficult to make either. If you wanted to take a go at making a chatbot, you could use one of the many platforms out there (of which there are plenty). We also produce a custom API based chatbot solution at hypermind technologies.
Research is now indicating that 62% of consumers would prefer talking to a bot over a human (https://www.drift.com/blog/state-of-conversational-marketing/). And this boils down to frustrations consumers have about finding information, with the biggest concern of not being able to get answers to simple questions, including a hard-to-navigate site, being the toppers of consumer frustrations.
Consumers have a desire for information being provided in real-time. In the same Drift report, the researchers found that consumers would expect a near-instant response whether speaking to someone over the phone/face2face or via online chat.
If you run a business, you need to be prioritizing the development of chatbot into your systems. With a bot, you don’t need to fill out any forms, you can validate the user quickly, provide information in natural language and real-time.
The Technology behind Chatbots
At the heart of a chatbot lies Natural Language Processing. The ability to understand the semantic connection of words uttered by a user and hold a natural conversation as you would with friends.
This is where the problem of setting up your own chatbot comes in.
The human art of conversation is filled with nuances and colloquialisms which aren’t easily interpreted or understood, even among one another. To successfully build a natural language processing algorithm which can detect nuances requires a lot of training. And that requires three things; 1. Data, 2. Processing power and 3. Time.
Small businesses tend to run into all 3 issues; not having enough data (particularly start-ups), the processing power (GPUs, 1tb+ of storage space and ability to deploy to a cloud host – AWS/ Azure to ensure super quick speeds) and time.
So how can I leverage chatbots?
For something simplistic, such as an order tracker or FAQ bot, it can be simple to determine the questions users will ask your bot. which means you can build up a list of questions and their corresponding intents.
For something more advanced and potentially transactional (shopping bot), you’ll need to work around the legalities surrounding payment, data, security etc… while needing to react to changes in decisions, session timeouts or situations where payment can’t be captured. These need dedicated dev work.
The suggestion is, as always, to begin small.
Start small with a customer service bot on your website. Let it link to your FAQ for customers to ask questions and get their required answers easily. Build a fail-safe to move the convo to a human agent if the bot can’t answer.