Behind the Scenes: Let’s talk Chatbots

When you visit a corporate website these days, you’re sure to be greeted by a little chatbot at the bottom of the screen, ready to help you. These chatbots are making a huge difference on how customer service is being perceived.

They are the first interface between the company and a prospective customer, and a visitor can directly converse with the bot. This self-service application is making waves in the SaaS industry.

“Chatbots have the ability to improve the customer service experience by leaps and bounds. Chatbots can be contextual, sensitive, and personal at the same time to offer a never-before kind of experience.” –Ajeet  Kushwaha, ML/AI/Bots Strategy Lead at Freshworks

Being a business software company, Freshworks primarily has three kinds of users: end users, internal users, and our agents. The prospects usually visit our website to evaluate our products. Chatbots come into the picture to help them during their evaluation phase, when they visit the site.

From the sales perspective, our chatbot might say, “Hey you are trying to sign up, you might want to know more about the product or “Hey, since you are coming from an e-commerce industry, here are the five features which other people from the same industry are looking at”. With chatbots you can extend a lot of help to the customers and eventually turn them into power customers.

How do the chatbots work?

To understand this better, we got a lowdown on chatbots from Ajeet Kushwaha who leads ML/AI/Bots Strategy at Freshworks and Rahul Agarwal, Director of Technology at Freshworks.

Bots engage with the visitors contextually by leveraging visitor journey from tools that show the visitor’s google search or organic, navigation path, landing pages, and so on.

Chatbots have become quite popular amongst small and medium businesses as well. Despite not having high traffic, it gets hard for employees to respond via chat 24×7. Bots makes their job considerably easier. When we look at larger business, mid-market or enterprise, they require bots even more because they have a lot of customers and fewer agents.

For instance, we have integrated one of our products, Freshchat, with advanced bot building capabilities, powered by Freshbots. Apart from this, we’ve also integrated Freshdesk as a support bot.

We are offering this to a lot of small and medium businesses as well. This makes it easier for them to retain the interest of their visitors, and the response tells us that we were on the right path.

There is always one primary focus that we emphasize on for the bots – offering instant help to the customers for predictive/repetitive use-cases.

What we need to remember is, when you build a bot, make the context of the issues/queries of visitors/prospects more solution-driven rather than just vanity-driven. It has far more efficacy when you add value to whatever the business objectives are.

Take the Pre-Sales team for instance – through the bots, they acquire more leads from support. But how do you ensure that the prospect goes through without losing interest? Letting the end customer find answers using FAQs or by navigating through complex decision flows (DT) to know the status of an order might seem tedious.

How can this be resolved with the help of bots, you might ask. Well here’s how:

If someone says, “Hey, where is my order?” the bot understands the context/ intent. If customers’ phone/email is available (if not it will ask “hey, may I know your registered mobile number?”) it can make an API call to the order management system, fetch the latest order and say, “Hey, here is your order status.”
At the same time, it can again make an API call to the logistics module or management system and say, “It will be delivered, say, day after tomorrow.”

Innovation is everything. The rate at which chatbots are growing is extremely high. There have been instances where we have had trouble but through research and development, we have always managed to formulate a solution. The key is to understand where the bot is not able to perform by analyzing the funnel of conversations between the bot and the visitor.

Sometimes the built intents can also be recognized as the negative intents. For example, if the bot is asking the visitor for an email, and he/she isn’t comfortable with disclosing that information, a normal bot would go in and ask, “This is not the right email format, could you please enter your email again.” This is the bot just asking the visitor the same question in another fashion. But can you actually recognize that negative intent? We’ve built on this technique to generate information in the easiest manner, without letting the visitor leave with a negative experience. So, there’s some bit of intelligence that comes into play if a user does not want to give that information.

Remember, chatbots aren’t replacing our agents:

What we do at Freshworks is allow the chatbots to add more value to agents, not replace them. Hence, the concept of ‘Agent Assist’ was derived. We are able to cover how the same bot technology built for end customers to facilitate self-service can also be used by an agent to interact and access information, follow processes and be more effective. This allows them to solve more tickets in less time.

This process has helped us become more effective and appreciative of bots as not only has it led to better retention and CSAT but, and more importantly, customer happiness and satisfaction.