The one overlooked reason conversational AI strategies fail (And how you can get it right)

Despina Singh

August 8, 20237 min read

Share this Article

In all honesty, I have mixed feelings about conversational AI. Don’t you feel a bit anxious, too?

Sure, not a day goes by where I don’t think, “AI is so cool. Look at everything it can do!”

Buuuut….at the same time, I can't help but think, “Why can't brands using AI get it right the first time, or the second, or the Nth time?”

While conversational AI is mostly a meme machine for now, we can learn from experience that conversations with AI can cause all sorts of troubles.

But before we get to that, tell me this: Do you think every business has different needs for conversational AI or can we consider it a one-size-fits-all solution?

Here's my take on it: Trying out different conversational AI solutions in a never-ending cycle is no way to get your AI needs right. Why keep investing in inflexible solutions with limited capabilities and irrelevant use cases that strain your operations?

Instead of your conversational AI solution, YOU should be the one in the driver’s seat.

So how do you call the shots?

It’s simple: You create a conversational AI strategy based on your own business needs. Be a Conversational AI Chameleon. Master of adaptation. Changing your colors (strategic nuance) to adapt and thrive in its surroundings. Based on the feedback from customers.

Let’s live-create a conversational AI strategy together, right here and now, ok? I’ll make it an easy 3-step framework you can use and profit by. Let’s get to it.

[clinking glasses]

Table of Contents

3-step conversational AI strategic framework

First, as a little motivation in your day: Gartner projects the impact of Conversational AI to be a $80 billion reduction in contact center agent labor costs by 2026.

Here we go….

1. Start strong: clear goals and use cases

The motivation for the “Checklist to the conversational AI use cases and goals” (checklist below) might have been the crazy-scary viral story of The Rogue Bot and the Forbidden Pills.

You know the danger of bots giving inaccurate or incomplete information. Big red flag and turn-off for customers. Happens all the time, right? Ask a bad bot a simple question about a hotel, flight, or skincare product. Get back an irrelevant, unhelpful answer. Very frustrating. Worse when the stakes are higher, like with medical advice.

Unfortunately, once upon a time, there was a bad bot that went rogue. We’ll call it “Fay.” Things took a dark turn when Fay – designed to learn from people’s language and behavior, good and bad — suggested that a way to deal with a headache was to "pop a couple of pills and keep partying."

Not only was the advice random, but potentially dangerous.

So here’s how to NOT let loose that rogue bad bot.

Your checklist no. 1 to easily define clear AI bot goals and use cases:

Checklist 1 of what and why of conversational AI goals and use cases

I hear you. You need a real-life example.

Standard Chartered, a big multinational bank, used Conversational AI Bots to create a smooth and easy customer experience. (Yeah, it was Sprinklr’s. But it’s a valid story worth your time).

They digitized 70% of customer conversations — freeing up their human agents for the more complex transactions, such as mortgage applications and wealth management.

It worked like a charm. A 90% improvement in resolving customer queries on the first contact in the same channel.

Just defining use cases and clear goals isn’t enough. Here’s how to spark and fire up that community fan base. You’re all in, right? Let’s do it…

2. You're so close: put yourself in your customer’s shoes

Imagine after a great holiday in Hawaii, as you’re about to check out of the hotel, you ask the receptionist if you can settle your bill through your phone.

They tell you’ll receive a text with a payment link. But, it redirects you to a payment gateway you aren’t familiar with. What was supposed to take a minute or two, goes on forever.

So, what if you changed that scenario and outcome? Instead you receive the link from a chatbot. Upon clicking, a pre-filled form with all your transaction details pops up.. All you have to do is hit the “Pay Now” button. Two minutes tops. Done.

[wind chimes tinkling]

When it comes to designing a conversational AI bot, it's not just a nice thing to do, IT’S AN OPERATIONAL NECESSITY to factor in how you put yourself in your customer’s shoes.

Incorporate customer feedback into your bot's design. Listen to what customers have to say. Identify their pain points. Improve and implement your response. And continuously use feedback to refine your bot's flow and engagement.

Make your customers feel they're co-creating their experience with your bot. And getting the fast and right response every time. That's how you build trust and loyalty, and keep your customers coming back. And telling others about how great you are.

Your checklist no. 2 to win with a customer-centric design without a shred of confusion:

Checklist 2 of what and why of customer-centric conversational AI design

Want to learn how to benchmark NLU engines? (+ a comparison of market leaders)

3. Final checkpoint: future-proof your strategy with a different kind of growth plan

Many moons ago, there was a bot that failed to stand the test of time. Let’s call it "N."

N was designed to be a virtual assistant that could help people with a variety of tasks. You know, the ‘making reservations,’ ‘booking flights’ and ‘ordering food’ variety.

N required human operators to handle some of the requests, adding to operational costs. And it struggled to understand the more complex user requests, making it difficult to scale the service.

To no one’s surprise, N was shut down due to its high running costs and low adoption rate.

[sudden silence]

What can we learn from failure? When launching an AI bot, the cost and scalability of the solution are everything. You don't want to spend money on something that's not going to work for you in the long run.

Consider these:

  • How well your bot can handle different kinds of requests

  • A range of user needs you don't want to rely on humans to step in and fix all the time

  • Easy bot training to handle new requests, so you don't have to spend even more money on human intervention

Your final checklist to make your strategy persevere through time:

Final checklist of what and why of managing and scaling conversational AI

Another real-life example of a company’s growth plan

It's not the stuff of fairytales. Building a growth plan, and succeeding at it, was and is an imperative at Vi.

Vi, a leading telecom service provider, was able to scale their digital care services during a 2x surge in social interactions at the height of and through the pandemic.

Before the pandemic, most people who worked from offices had not considered having a stable and fast at-home internet connection as crucial. Understandably, the situation changed when working from home became the new norm during COVID-19.

The team was swamped with comments that needed troubleshooting assistance, and they couldn't get back to all the customers right away. They realized they needed a platform that could help them monitor their social accounts and engage with important stakeholders like enterprise customers. It was really the need of the hour.

With the intuitive power of AI they continue to sniff out the most important messages based on stuff like keywords, sentiment and issue types.

The result, you ask? A whopping 30% time AND HASSLE saved with automated responses. Pretty slick, huh?

Alexa, play Ain't No Stoppin' Us Now by McFadden and Whitehead

Because…we’re in this together. Thank you — for live-creating this strategy with me.

So many things that hold us down, but…

♫ Ain't no stoppin' us now

We're on the move

Ain't no stoppin' us now

….

Don't you let nothing, nothing

Nothing stand in your way of getting conversational AI right

Until we meet again…

Share this Article

Sprinklr Service
Conversational AI and Bots

Related Topics

Top 10 CCaaS Providers in 202411 Top Customer Engagement Platforms in 2024 [+User Reviews]Customer Retention Management & How to Do It