What the shift to an “AI-First” world means for the Call Center?

It must have been about 2 years ago when I made my last (outbound) call to a call center. I don’t even know if it was a “call center”, “support center” or an “advice center”. In my perception, they are all the same… but anyone reading this blog, feel free to send me a tweet to enlighten me.

I can hardly recollect the reason for calling, but the experience of wasting my time – waiting, while listening to the most dreadful tune – remains engraved in memory. As the operator finally welcomed me, she instructed me to identify myself using the 10-digit number that was printed in the top right corner of my last invoice.

And to top it off, it was up to me to explain my product portfolio… before we finally started talking about my problem. I was rerouted from the 1st line operator, via the 2nd line onto the expert who was finally able to help me out. I can’t imagine what would have happened if I got disconnected during this horrendous adventure.

I don’t know about you, but this isn’t my textbook definition of the ideal customer experience. In this article, we will review how the upcoming technology innovations can have a positive impact on the customer experience. The ultimate goal would be to introduce a customer-first strategy into the call center.


Synchronous vs. Asynchronous Communications

In order to avoid the waiting game, and allow each customer to immediately speak to an operator, a call center would need to be scaled to peak performance. This is economically simply not viable, you would basically need a lot of operators and the majority of them would be fiddling their thumbs most of the time.

The main issue with existing call centers, is the “synchronous” nature of voice communications. In “synchronous” mode, the communicating parties are sharing information at the same time. Examples of synchronous communications include: phone calls, live chats and conferencing (audio or video).

We should try to avoid the expectation that multiple parties should always be required to communicate at the same time, and opt for an “asynchronous” approach.  Examples of asynchronous communications include: traditional (snail)mail, email, texting or even online forums.

Note: The words “synchronous” and “asynchronous” are technical terms, and originate from both Latin and Greek:

  • “A” , means “negative”, or “not”.
  • “Syn”, means “equal”, or “same”.
  • “Chronos”, means “time”.

If combined, “synchronous” means, “same time”; and “asynchronous,” means, “not same time” .

Augmenting your Voice-Based Customer Support

The first option to improve the call center would be to introduce voicemail. If there are no operators available, an automated voice message would inform the customer of the limited capacity, and ask to leave name and number after the familiar beep (a short description of the problem would also be helpful).

This already makes the overflow calls “asynchronous”, but let’s be fair… this still sounds like a solution that dates back to the previous century. Did you know that “voicemail” was invented in 1973, and the “answering machine” even back in 1957?

An even better option, we implement a solution where an automated voice requests you to describe your problem at the start of the call. Your question could then be translated using speech-to-text technology, and analyzed in real time.

In this scenario there are multiple possibilities:

  • The translation of the speech-to-text fails, the customer intent can’t be clearly identified, or a negative sentiment is detected. In these cases it’s always advisable to route the call to a human assistant. These calls remain “synchronous”.
  • The translated speech-to-text indicates a high urgency intent. In this case, it’s again advisable to route the call immediately to a human assistant. These types of calls should also be handled in real-time (“synchronous”).
  • The translated speech-to-text indicates a low(er) urgency intent (for example the customer has a general inquiry). These calls can be handled with a delay (“asynchronous”).
    • The system can informs the customer that a callback is scheduled.
    • If a particular expertise would be required, the system can automatically schedule a callback appointment, and ensure that the appropriate expert contacts the customer.
    • If the intent is recognized (for example a frequent request), the automated voice could already make a few suggestions. The system might even automatically send a few pointers to the customer using an alternate communication channel.

In this setup, the voice-based customer support has been augmented: low(er) urgency calls are handled “asynchronous”, while the other calls are handled in real-time (“synchronous”). This is the desirable intelligent queueing that would already dramatically lift the customer experience in this day and age.


In the more distant future, as voice assistants become mature, it will become possible to handle most calls automatically (and “synchronously”). This is obviously a desired state, but let’s remain realistic… it’s not something companies can implement today.

I could have stopped this article at this point, but I’m personally convinced that “augmenting your voice-based customer support” is not the (r)evolution you should focus on today.

We need to take the changing customer behavior into consideration.

Voice Call Fatigue in the “Mobile-First” world

In the last few years people have started to use their mobile phones as their principal computing device. The smartphone has become so intelligent that it has taken the place of a tablet, a laptop, and even a desktop computer.

The functions that were once reserved for the desktop, have started to appear in mobile apps. The smartphone users now communicate via Facebook, Messenger, Skype, Snapchat, Twitter, or WhatsApp… and hardly make voice calls.

The days of using a phone just for voice calls are over.

In our personal lives, we haven’t abandoned phone calling… but calls are reserved for closer relationships. This translates in a decline of the number of voice calls.  Three years ago, nearly all smartphone users made at least one voice call per week. Last year, only seventy five percent of smartphone owners did.


With the exception of my family, it seems that I hardly make voice calls. I’ve actually never been the type to pick-up my phone, and “talk” to a stranger. It doesn’t matter if you are still a “talker”, but I’m definitely a “texter”.

If my Internet connection at home brakes down, I turn to Twitter. If the service in a restaurant is lousy, I Google for a complaint form (or an online culinary review site). If I have a generic product inquiry, I post a question to all my friends on Facebook (or to an online forum). And you probably guessed it… all of these actions are launched using my smartphone.

It’s not always young people that prefer texting over a phone call; although recent statistics indicate that the younger you are, the more likely you are to prefer texting.

Adopting a Messaging-First Customer Support Strategy

The real (r)evolution to improve the call center would be to adopt messaging as the primary support channel. It’s as simple as that, stop endorsing the telephone number of your call center… and start promoting your messaging support.

I’m not talking about adding a little live “How can I help you?” support window on your website, but integrating your customer support in a (or multiple) messaging platform(s); with WhatsApp and Facebook Messenger being the leading players.

In January WhatsApp launched its “WhatsApp for Business” service; empowering small business to interact with their customers more easily using messaging. The mobile app (currently only available on Android) includes features to automate, sort, and quickly respond to messages. The enterprise version (presumably “WhatsApp for Enterprise”) is currently being tested by an exclusive set of innovative global enterprises.

One of these frontrunners is KLM. The Royal Dutch Airlines already offers proactive flight information, booking confirmations, check-in notifications, boarding passes, and flight status updates via a verified KLM WhatsApp account.

If you require a higher level of authentication (for example in financial services), you should create a messaging interface – similar to that of the leading messaging platforms – inside your app. A customer should be able to start a conversation on a public messaging service, and for more sensitive questions be seamlessly transferred into the messaging interface of your app.

The big advantage of this messaging-first strategy is to be connected to many of your customers, yet be able to engage with them on an individual basis. In addition, the messaging support can be real-time (“synchronous”), but its actually an “asynchronous” support model by nature. In other words, your call center does not need to be scaled to peak performance.

On top of this, you can augmented your messaging-first strategy and implement content based routing for all your requests. The support questions can all be analyzed in real-time using Artificial Intelligence (AI), more specifically Natural Language Processing (NLP).

In this strategy the support is no longer handled on a “first-come, first-served” queuing basis, but can be intelligently routed – based on the content – to ensure the best customer experience.

The “Automated Support Agent” handles the following request…

  • If the intent of the question is clearly understood, and the answer can be provided automatically… the automated agent provides the answer. I sometimes hear people say that customer prefer talking to humans, but if it’s 3AM in the morning … you will be glad if an automated system can help you out.
  • If the question cannot be answered automatically and there are no support agents available, the automated support agent could suggests the timeframe when the request will be handled. The system would obviously remain responsible in managing this expectation: sending updates to customers, and escalating the request to your support agents.
    • If the intent of the question requires a specific expertise, the question would immediately be routed to the most appropriate support expert.

The “Human Support Agent” handles the remaining request…

  • If the intent of the question is clearly understood but more sensitive in nature. Or if the request is a little more complex to handle programmatically, the human support agent takes over.
    • The urgency of the issue (determined using the content in the request of the customer) would define how quickly a customer would be in contact with a human support agent.
  • If a negative sentiment is detected, the request should always be handled by a support agent… trying to automatically deal with these situations is a recipe for disaster.

In this setup, you’re probably noticing that the boundaries between “synchronous” and “asynchronous” support are fading. A large set of requests would be handled in real-time, and the others would be intelligently scheduled to be addressed as quickly as possible.

A “messaging-first” strategy is a big improvement for customers, but also a revolution for the support agents:

  • As frequent requests are automated, support agents no longer need to provide “parrot” support… and repeat the same answers every day. In essence the 1st line requests will be automated, and operators can focus on the more challenging questions.
  • In the messaging interface, the conversation history remains accessible. A support agent would be able to scroll through the last conversations, and quickly get a full understanding of previous or related questions.
  • An integration with the Customer Relationship Management (CRM) system ensures that an incoming support messages are immediately linked to the customer (using the mobile phone number).
  • The automation will ensure that key performance metrics in the call center move from quantity (number of requests) to quality (customer satisfaction). The human operators will be able to spend more time on their requests, and be truly able to focus on the best possible experience for the customer.
  • In a messaging-first strategy, the implementation doesn’t restrict a human operator to a single call. Your operators will be able to handle multiple request at the same time.

In addition, you would be able to send your customers proactive servicing messages using the same platform; based on the context of the customer, suggest timely relevant services… entirely personalized for the customer.

The customer will get the feeling that they are genuinely communicating with your brand.


In order to improve your call center, you could focus on a strategy to augment the intelligence of your voice-based customer support. This is however ignoring the “voice call fatigue” trend; statistics indicate that people are abandoning phone calling, and using this type of communication almost exclusively for family and friends.

Instead try and see the advantages of immediately adopting a messaging-first customer support strategy; promote your messaging support, and de-emphasizing your telephone support.  An implementation using content based routing; where support questions are analyzed in real-time using technology based on Artificial Intelligence.

The shift to an “AI-First” era, will enable you to introduce a customer-first strategy into the call center; providing the best experience whenever your customers are in need of your support.


Categories: Blog

Tags: , , ,

Leave a Reply

%d bloggers like this: