NEW Try Dasha in your browser with Dasha Playground!

An honest take: here is what voice AI apps can and can't do for your call center

Photo by Photos Hobby on Unsplash
Photo by Photos Hobby on Unsplash

The term voice AI is bandied about all the time. Many businesses that want to stay ahead of the competition seem to want a piece of it.

Some voice apps vendors market them as the ultimate solution for customer experience companies and their pain points.

The reality is a bit different. While you definitely can outsource some parts of your workflow to conversational AI, you can’t just sit back and let the AI do all the talking. At least, not yet. 

So here’s an honest take on what voice AI apps are capable (and incapable) of. 

General (non-functional) features

These are things voice AI apps can ensure:

  • Around the clock support. AI doesn’t need a break. In its interactions with customers, a voice app will solve basic queries, perform required tasks and put down important information as long as you need it, making sure your call center is available at all times and leaving no place for human error. 

  • Scalability and multitasking. There are only so many calls a human agent can take or make during their shift. The number of tasks they can perform at the same time is quite limited as well. As your business expands, you will most likely go through these growing pains. Voice AI is designed to handle hundreds, if not thousands of simultaneous conversations – scaling it would cost you a meager fraction of what you would pay to hire more workforce. 

  • No accent-related bother. Many businesses use off-shore call centers. Unfortunately, accented speech and cultural difference can create friction with customers. The communication gaps that stem from these calls can be irritating at best and damaging at worst, leading to wrong actions taken and customer trust betrayed. AI will speak with any accent you like or none at all.

And here are things voice AI apps can’t ensure:

  • Always a full understanding of customer emotions and sentiments. It’s a great challenge to identify how the customer feels. Some AI apps employ natural language recognition algorithms that mark words indicating emotions. But it doesn’t mean they run like clockwork: such tools fail to take in the context of the conversation and can’t respond to customer sentiments accordingly, which can ruin the relationship between the c2lient and the company. 

  • Being liked by customers due to lack of personality. AI apps run on scripts you feed into them, and these scripts serve specific purposes. That means, most apps can’t go on tangents or crack jokes to relieve the pressure if things go south. They treat all customers the same, which can be a bit problematic in such a customer-centered industry. 

  • Multiple languages support and recognizing all accents. Your AI vendor may not support all languages that your business requires. Accents on behalf of customers can create friction as well. I wish the famous ‘Scots in the elevator’ skit was just a joke. We’ve worked with some pretty complex accents (first generation immigrants) and have managed to successfully parse meaning out of the phrases. 

Functional performance

Let’s move on to specific tasks voice AI apps can handle: 

  • Call routing. Intelligent call routing, to be precise. It’s when AI uses the information your customers provide on the call along with the information stored in the database to make sure they get redirected to the proper service rep. To identify the best match, the system takes into account the agent’s field focus and level of expertise. As a result, you get improved AWT and happy agents who can now focus on helping their target audience. 

  • Voice of customer surveys. Email or text messages are the most common options to get customer feedback. Yet they get 10-30% and 20-40% response rates respectively. Over the phone, you can get a 90%+ response rate. Add in the fact that you often don’t get answers to open-ended questions in writing (and these tend to collect the most valuable insights). On the phone, the customer will elaborate, and the AI will put down everything they say so that you can export and analyze it later.

  • Lead qualification. Voice AI apps can talk to your lead base (no matter the size) and ask all the basic questions to determine interest. If the lead has shown interest in the product, AI will transfer the call to your agent. 

  • Simple customer upsale qualification and upsales. Same goes for upsales: the AI can call each of your customers and offer a service upgrade (or whatever) to them. Success can mean connecting the customer to an agent or updating customer details and/or forwarding a confirmation sign off form to the customer. This part is up to you. As for those customers who are not interested – well, AI will not ever get demotivated by rejection. 

  • Answering customer questions. To help agents who often have to deal with repetitive 1-2 minute calls regarding most basic questions, you can offload these queries to AI that will handle them with the same efficiency, allowing agents to focus on more valuable tasks. 

  • Addressing account-related queries and conducting simple processes. If you reinforce your call center with voice biometrics, your customers will be able to use AI apps to transfer funds or defer payments. If you’re running an insurance call center, voice AI can help your clients to file a claim. There are a lot of simple tasks that AI can automate so that your workforce would step in only when they’re needed.

And here are things that voice AI cannot handle – and these are the situations when agents have to step in:

  • Complex conversations. Today’s conversational voice AI apps can’t handle extremely multilayered conversations. For example – elaborate discussions of the product service, justifying in detail such and such actions being taken on the account, etc. Long story short - don’t expect to replace your tech support with AI. I will add to this the caveat that in theory you could make your conversations as deeply complex as you need on the Dasha Platform yet it will be a long and arduous process of testing, developing and training the AI app.  

  • Negotiations. Being a good negotiator is a skill that needs to be constantly mastered. Not every agent is able to settle disputes to everyone’s satisfaction, let alone an algorithm, no matter how smart and complex. 

  • Complex sales calls. No AI can outmatch a human when it comes to figuring out pain points or handling objections. While you can use AI for lead qualification, don’t even think of firing your superstar SDRs just yet! 

Some areas are already ripe for automation, in other aspects, it’s still early days for voice AI. I believe that you as a business leader should be realistic about what AI can or can’t do for you. If you like what you see – go automate and enjoy the results. 

Related Posts