Customer Support Automation Requires Cognitive Platforms, Not Bots

With the boom of e-commerce amid pandemic, there is a dire need for consistently efficient customer support that can deal with the high volume of tickets. Early chatbots introduced five years ago aimed at coping with the problem but they failed to efficiently replicate human customer service agents.

This story was written by Annie Clain, Outreach Manager at BTCWires.

Source: Customer Support Automation Requires Cognitive Platforms, Not Bots (zee5.com)

Customer Support Automation Requires Cognitive Platforms, Not Bots

With the boom of e-commerce amid pandemic, there is a dire need for consistently efficient customer support that can deal with the high volume of tickets. Early chatbots introduced five years ago aimed at coping with the problem but they failed to efficiently replicate human customer service agents.

This new technology caught a lot of attention and more and more companies started to adopt it, however, it didn’t take long for users to realize they were talking to a machine, and soon enough these robotic agents due to their incapacity to troubleshoot the way normal agents would lead to a search for a solution to this growing problem.

An Exclusive Interview With Leading E-Commerce Expert and Astute Entrepreneur Hamiz Khan

In an effort to solve this problem, Maxence Bouygues decided to take a different approach and started to create a deep learning algorithm that would be able to replicate humans and could learn from everyday experience. By contrast with standard approaches, these virtual agents created by Bouygues automatically learn from data (emails, tickets) that is already there and therefore provide about 95 percent accuracy while the industry oscillates in the 15 percent.

These virtual agents, built on large enterprise customers have automated over half a million tickets per year with a model that took no longer than 4 hours to train and deploy. “This is a true revolution compared to bots, which would have taken months to deploy, at a high cost for the customer care team. True deep learning can automatically understand and automate tasks by mimicking what humans have been manually doing for the past years,” says Bouygues.

Failure of fragmented customer support backend processes 

The ability to deliver consistent customer service is the key to effective customer acquisition and retention. In fact, according to a recent study 86 percent of customers are willing to pay more for a better customer experience.

However, consistently delivering efficient customer service is not an easy task, although there are plenty of customer services platforms businesses can use to improve front-end customer communications, many companies are still finding it challenging to deliver a consistent customer experience, because of fragmented back-end processes. For instance, let’s say that a customer calls to inquire about the availability of a product, to find this information the customer service representative will have to contact other departments to find out if the product is available. Hence, delivering high-quality customer service requires a high level of inter-departmental coordination and collaboration. Here are some “broken backend process issues” that one needs to take into consideration.

Poorly interconnected systems:

Systems are fragmented and poorly interconnected together. Sometimes due to mergers or acquisitions. One department may use a legacy system, or multiple systems depending on the provider with which they work.

• Communication issues between different entities. Tickets are sent to departments that aren’t competent to resolve. Sometimes, part of the process is outsourced by a BPO. The back and forth between BPOs and internal support teams, working with different tools is causing a lot of friction and sometimes multiple days of delay.
• Documentation is also spread across multiple knowledge bases. Some examples are internal documentation centers, wikis, external websites, legal documents, training documents, email templates. All hosted on different platforms.
• Agents use documents that they know of, or sometimes create their own. It’s for instance very common to see agents create their own “email templates” to respond faster to customers with similar issues. The problem is that when the support organization updates the official material and official templates, agents keep using their outdated versions, causing problems in the processes.

Cognitive platforms with RPA could be a solution

To overcome such challenges, companies need to take a holistic approach when creating processes throughout the organization in order to meet customer needs efficiently. This is where automation can help to tie up all the loose ends and execute a customer-centric business strategy.

As they are equipped with Robotic Process Automation (RPA) , a feature which allows these AI powered platforms to integrate with external systems such as order tracking tools, so as to perform automated actions (or automated processes). It’s the combination of the intelligence (the brain) and the RPA (arms) that allow for a full solution to work for a company, and enable real automation, and ultimately hit above 50 percent of ticket volume automation. “What’s important about these solutions is that they can easily integrate with multiple backend systems at the same time, in a timely fashion. Once this is done, the platform should provide enough configurability to easily program and re-program the tool by the company itself,” says Maxence Bouygues, product engineer of Forethought AI-powered agents like Agatha.

Agatha can provide businesses with peace of mind as it can provide them with the support customers require without the hassle of manually contacting other departments and the long and strenuous search process, though automated routing to provide customers with real-time assistance and insights. Hence, strategically and accurately supporting the customer service team.

Forethought has been able to provide its clients including fortune 500 companies AI-powered customer service agents who use Natural language processing to understand and immediately respond to customers to resolve simpler issues while also being able to do the repetitive search and routing that customer service agents spend valuable time on. Thanks to Bouygues who developed and built cognitive virtual assistants for his business clients and has been able to earn the company over $750,000 in annual recurring revenue in 2020.

Final Thoughts

Although the implementation of AI can be extremely advantageous, it is important for companies to choose the right tool that suits them best. “You need to understand that choosing the right tool is a very complex process and it requires a highly specialized team that can understand the company’s needs, and come up with and develop unique features and solutions that are required to address the unique problems faced by each company.”

The installation of High tech, AI software is an extremely complex process and requires a specialized team that is capable of understanding the competitive landscape that the particular company is operating in.

Maxence who manages such a highly talented team, who have been selling such technology to Fortune 500 companies, feels that jumping to just any AI solution will not solve the problem. “AI tools using NLP can be extremely helpful, as long as they meet the company’s requirement, but going for a solution that may be cheap but does not solve the problem will only be a waste of money,” he said.