“Bots” is a term used everywhere in the industry of customer service. They come in the form of chatbots, virtual customer assistants, avatars, intelligent assistants, conversational agents, and basically whatever else a company wants to call their product. So what do these terms refer to, exactly? Is there a universal way to classify bots based on their features or internal technology? Here’s how you can think about them:
Primitive Marketing Bots:
The most simple form of bots are primitive marketing bots. These are often used in digital marketing to simply send out messages on chat platforms such as Facebook Messenger. They fulfill the very basic requirements of being called a bot, which means they allow you to enter a question to respond back to you. They fall short on basically everything else – how long they take to respond, how well they understand your question, and how intuitive their responses are. A good start on the promise of chatbots, but nothing close to their potential.
Do-It-Yourself (DIY) Chatbots:
The next step in the evolution of chatbots is Do-It-Yourself, or DIY chatbots, which represent the majority of products referred to when speaking about bots. These bots must be developed over a period of months by companies’ customer service teams to include every possible user input and its corresponding bot response. “I would like to track my shipment” is different from “Where’s my package?” within these systems because the phrases contain different words despite having the same intent, so building these bots to be effective can be cumbersome. Some DIY chatbots do advertise the use of artificial intelligence or machine learning, but realistically its power is limited to synonyms of words, not the intent of a user’s question.
When a product needs to be trained rather than built from DIY software, a product crosses into the realm of customer-service chatbots. These are “pre-built” in the sense that a company doesn’t need to spend hours programming the software on how to respond to every question. It can decipher intent regardless of how a user asks a question. Due to the advanced artificial intelligence technology that these bots require, there are very few fully developed customer-service chatbots in the market yet. The main difference between these and DIY chatbots is on the back-end – are companies building or training the product?
Although they aren’t used in the realm of customer service, virtual assistants represent the most advanced artificial intelligence technology today. Software such as Siri, Google Home, and Amazon Alexa can interpret voice, hold conversations, retain information, and integrate across different applications and platforms.
Where Bots Fall Short
So if chatbots have so much potential, why do 43% of people still prefer dealing with an actual person over a chatbot? Or to put it another way – when is the last time you had a great experience with a chatbot? The technology clearly has a long way to go, and here’s where they fall short:
- They need to be built manually: The majority of chatbots in the market are at best DIY, so customer service teams need to build out decision trees for every possible customer request, as well as every way that request can be worded. If you want a bot to be conversational, then that introduces a full decision tree, increasing input possibilities exponentially.
- They need ongoing maintenance: They need to be reconfigured every time new information is added to the knowledge base. Chatbots do not automatically crawl company knowledge bases once information is added. Every new product comes with a new set of questions (and new ways they can be worded!) and a new set of answers.
- They require heavy integration: Every industry has unique words and phrases, from “tickets” and “NPS” in customer service to “accounts” and “sourcing” in sales teams. Chatbots need to be configured from the ground up to understand every use of these terms.
Bottom line: They’re cumbersome. To build an effective bot using a DIY chatbot product would take a huge amount of both time and resources. Even for large corporations that might be able to afford that, the technology will be left in the dust when pre-built bots are released.
We’re often asked if Solvvy is a bot. The short answer? No. We’re an intelligent conversational platform focused on helping users self-service their questions. We fundamentally differ from chatbots because we are built around artificial intelligence, not business rules. This means that our customers don’t build our product, they train it. They don’t make long lists of inputs and outputs, they guide our learning engine, which can understand questions however they’re worded.