Automate your whole customer service, not only chats

In 2016, we saw chatbots emerge as one of the hot topics and since that many companies have improved their customer service with chatbots but time has moved forward and companies are discovering that it does not solve all customer service problems.

Why automating customer service with a chatbot is only part of the solution

By our statistics, a company with 1M users gets about 150K customer contacts a month and out of those contacts 130K are calls and 20K are chats. So, although chatbots enable a high-level of automation, there are still more incoming calls compared to chats and this probably will not change in the foreseeable future in many industries.

Many organisations are focused only on chatbots rather than on full customer service automation. The reason is that there are no service providers who offer a “full package” and implementing separate services to automate customer service is rather complicated due to integration issues. Using multiple service providers is a problem since it duplicates knowledge bases and it is tricky to cross-train machine learning models on different data sources. 

Even if companies have implemented different solutions from different providers then making representatives and the whole organisation work with multiple systems is a pain.

This is where MindTitan differentiates as we are the first to combine both channels and target full customer service automation. What we have observed is that higher level of AI automation can be achieved in customer service by combining both solutions to work in concert. One such example is MindTitan’s client Elisa, a Northern European telecom, that is using AI to automate handling incoming chats, route incoming calls to the appropriate customer service specialists, suggest the correct dialogue flow to representatives and gather a bunch of useful information about customer service in the process. Also, the customers appreciate it as well by giving it an NPS of around 50.

Call center pain points 

With a great number of customers comes a great number of customer queries on a wide range of topics. As an example mentioned before, an organisation with 1M users gets about 130K customer contacts each month and has about 130 dedicated representative dealing with customers who wait in line for 2.4 minutes on average. This 2.4 minutes is obviously a problem as customers don’t like to wait and, with an average 10-15% abandon rate, leave companies clueless on their issues, lose sales opportunities and lower customer satisfaction. Our statistics also shows that about 30% of customers choose the wrong option from the IVR menu which leads to even longer wait time and takes about 110 seconds of customer representative time to first understand what the customer actually wanted and then transfer the call to the right person.

An additional issue is the 14 month workforce turnover time which means many representatives are not competent since they first have to spend a lot of time reading manuals to answer customer questions, but since they are human and they do not remember everything, they remember it incorrectly or the content of the manuals changes over time. It takes months before you can be confident that a new hire is good at answering a wide variety of questions, but even then you can never be sure because 30% of customers press a wrong button in IVR and representatives have to redirect those calls. There is also a poor overview for managers even with long-term representatives to know how competent they are as there are not very good ways of checking it, besides listening to calls. 

Customer pain points

When a customer reaches out to a service provider, they want a solution now and expect companies to communicate with them on their preferred channel, be it in person, online, or on the phone and they want problems to get solved within first contact.

A company could double or triple the size of their customer service departments to achieve faster response & answer rates, but it’s not a feasible solution, as the company would have to hire more people as response times crept up.

So how to improve the current approach?

Don’t get me wrong chatbots are still important and should be implemented as they can fully resolve customer contacts if done correctly. Go beyond just giving some answers to questions – make your focus to fully resolve their issues on the first try. Chatbots also enable to take the human customer service representative fully out of the loop but as we saw from statistics real problem lies with calls that make up to 85% of contacts. And this is where CallBot can help.

Stage 1: Get rid of IVR with CallBot

Customers usually call because they have an urgent issue and while in a hurry to speak to someone who could help them, they often don’t select the best possible option from an IVR menu. Also, let’s not forget that IVR is designed for push button phones, not smartphones. Inability to unlock their screen locks mid call  and find buttons to press on a smartphone is a top complaint from customers. By the time they find a way, they just wish to speak with somebody and press something randomly. This often results in them being forwarded from one department to another to reach someone who can assist them.

For a customer service specialist, this means a lot of repetitive work with impatient people while for customers, it is an irritating and time-consuming process which often leads to hang-ups and bad customer experience.

It has been shown that customers spend about 30% of their total call time selecting options and getting routed to the correct place if something goes awry in the process.

So what have we done and learned? Well, MindTitan developed a CallBot solution that simply asks the person “What can I help you with?”. The customer then describes the problem and the CallBot forwards the call to the appropriate specialist.

As the company, you also have a bunch of benefits. You reduce the wait times and finally easily meet your SLAs as less time is wasted on routing and representatives can actually focus fully on resolving the issues of your customers.

Also, CallBot routing enables to make your new recruits immensely more efficient. It is now possible to train your new recruit in only one topic and route them only those calls. This means that new recruits are already productive on the first week. Once one topic is mastered, train the representative with the next topic and start routing them additional calls. This is the dream for onboarding new recruits in the contact center.

You can also do prioritization – if the AI detects sales opportunities or serious technical issues, those can be handled as top priority. If your abandon rate is 15% and out of those 10% could have had a sales opportunity, you can calculate how much revenue you are losing. With CallBot prioritization, you can catch most of them.

Stage 2: Empower your representatives with dialogue flows

Dialogue flows can be a powerful tool for customer representatives. Suggesting representatives the correct dialogue flow means they can give more consistent answers faster and with less training. Dialogue flow is basically a smart script which gives instructions to the representatives and does all the necessary background checks itself. With a well constructed dialogue flow, the representative simply has to say and ask what the AI suggests them, simple as that. 

However, it is important to track the performance of every dialogue flow to make sure that it actually solves customers’ issues. If customers’ issues are not resolved by the dialogue flows, the dialogue flow owner needs to do some investigative work to figure out the issues – are the answers confusing and too long, are there cases we are not covering, is the vocabulary we are using confusing for our customers etc. Once the issues are detected, dialogue flows need to be updated accordingly. Getting the dialogue flows correct might take a few iterations – we have seen cases where only 20% got their issue solved with iteration 1 and more than 90% with iteration 5.

Stage 3: Automate flow

Stage 3 is the automation of stage 2. Not all the dialogues should be automated as we believe there are cases where representatives will keep on doing a better job for a very long time – for example, calls with a sales opportunity. On the other hand, if the dialogue flow is not very complicated and the representative is basically just asking and saying what the AI suggest, we have an opportunity to fully automate that. 

From stage 1 IVR automation we pick out those calls which can be handled fully by AI and direct the rest to representatives who can focus their attention on those calls where human touch is essential and needed.

What value CallBot can bring

Deploying intelligent CallBot can enormously reduce waiting times for customers, reduce hang-ups from customers, automatically answer the most questions and forward the calls to the appropriate customer service representatives fast and precisely.

And that is not all! The AI models for customer service are dependent on the data inputs, so the customer service-related models are amplified and get a lot better as they get more inputs. Calls transcribed to text can help chatbots and vice versa. 

While a single machine learning model improves the customer experience, combining multiple models across customer service and bringing in data from outside the contact center brings true efficiencies while vastly improving the customer experience and also additional revenue possibilities.

Contact us to find out how automation of all customer support with a personalized AI assistant Titan CS solutions has benefited MindTitan clients.

Related posts