Table of Contents
About This Guide
It’s no secret that OpenAI’s generative AI product, ChatGPT, has made waves across the internet. And the recent launch of GPT-4 has only continued the tsunami of AI talk.
But the rise of AI is far from new.
For a while now, businesses have been exploring the potential that AI solutions have to offer. Back in 2022, our State of Marketing and Sales AI Report found that 51% of marketers are already either piloting or scaling AI technology to simplify their work lives and craft a more engaging customer experience.
Among our customers, this plunge into AI takes the form of Conversational AI — a type of artificial intelligence that allows consumers to have real-time conversations that resemble a human-to-human interaction. It’s the same idea as Amazon’s Alexa or Apple’s Siri: You ask a question via voice or text, and then it gives you the most relevant and actionable answer based on the data it has stored.
Already, Conversational AI has proven to provide a better experience for both customers and sellers alike. Just look at these stats:
- Drift customers who upgraded their accounts to include Conversational AI saw a 30% increase in pipeline with 30% less work for their chat teams.
- When reaching out for support, customers that chat into an AI solution are able to get their questions answered immediately, with no additional routing or ticket creation, 80% of the time.
- Overall, Conversational AI has proven to decrease cost-of-service by 30%.
Needless to say, if you want to win, you win with Conversational AI — and this book is here to help you do exactly that.
In this book, you’ll learn all about making Conversational AI work for your business. We’ll show you what Conversational AI looks like at Drift, explain how it differs (but can work with!) generative AI, and share several real-life use cases. We’ll also go over the crawl-walk-run approach to implementing Conversational AI so that, when you’re ready, you will have all the information you need to hit the ground running with AI.
Skip Ahead to Dive into Conversational AI
Every person’s journey with AI is different. This guide will cover all things Conversational AI — whether you’re just getting started or you’re looking to uplevel your efforts. We promise that there’s a little something for everyone.
Ready to learn? Skip ahead to the section that’s most relevant to you:
How Drift Defines Conversational AI
By giving site visitors the ability to express themselves in their own words, Conversational AI allows you to engage more people in real-time conversations that are personalized and relevant. But Drift’s Conversational AI takes things a step further by letting site visitors drive the conversation.
When a site visitor asks Drift’s Conversational AI a question, it will provide an answer based on the thousands of conversations it has analyzed and stored. The difference is that this data comes from conversations specific to your business. So, not only is the AI able to provide an accurate response, but it can also pull from various business use cases and guide the site visitor towards the right next step for them.
Drift’s AI also homes in on the outcome that the site visitor is looking for. This means that, no matter how the site visitor asks the question, our AI will be able to understand the intent behind the visitor’s input and provide an accurate outcome — be it routing to a sales rep, providing a help doc, or serving up content — 85% of the time.
Now, let’s take a look at what Drift’s Conversational AI looks like in practice. Here’s an example from our own home page 👇
- A prospect lands on www.drift.com and is immediately greeted by a chatbot.
- The prospect engages with the bot, which immediately acknowledges itself as an AI bot.
- The prospect types in the question, “What is Conversational AI?”
- The bot defines Conversational AI and offers up additional resources for the prospect to peruse.
As shown, Conversational AI is capable of accurately responding to input from site visitors and then laying out a conversational journey that is relevant to them. In this instance, the site visitor wasn’t ready to talk to a human, so instead the AI chatbot served up resources to nurture them. Not only does the site visitor get the information they need right away, but this also saves time for your sales team. A win-win situation!
Drift’s Conversational AI Cheat Sheet
Take a quick peek at all of the features that constitute Drift’s Conversational AI.
How Are Conversational AI and Generative AI Different?
By now, ChatGPT has practically become a household name. And you may be wondering: How does ChatGPT’s content generation differ from what Conversational AI does?
Put briefly, ChatGPT uses generative AI to create its content — and generative AI and Conversational AI are not the same thing. Here’s how they’re different 👇
Traditional Conversational AI uses natural language processing (NLP) and machine learning to understand the way people speak as well as the intent behind their words. Humans can be involved in choosing what exactly the AI is trained on and how it responds. Drift’s Conversational AI is specifically trained on B2B sales and marketing conversations — and even more importantly, on our customers’ businesses. Generative AI, on the other hand, uses machine learning to process a much larger data set — data that spans the entire internet — into patterns that it can use to generate new content (like text and images). It isn’t contextual to specific business needs or goals, and humans are often not looped into the training process.
Generative AI uses a transformer-based decoder that generates useful content like email copy, blog outlines, and even podcast clips for social media. Meanwhile, Conversational AI uses a model that guides site visitors towards the specific outcome or solution they are looking for — be it booking a meeting, solving a technical issue, or getting more content on a topic.
3. Response Accuracy
Generative AI is trained on general information that’s available on the internet with no human review, so it may respond to the same question differently depending on how it’s asked.
On the other hand, Conversational AI uses NLP and natural language understanding (NLU) to grasp the context and intent behind your site visitor’s words to generate a response. Drift’s Conversational AI makes use of a human-in-the-loop approach, which means that you always have the opportunity to review the AI’s conversations to ensure that the AI is providing accurate and safe answers in the right tone of voice, giving it an extra set of guardrails that isn’t included with GPT.
…But, Conversational AI and Generative AI Can Work Together
Although Conversational AI and generative AI differ, they both point to the massive impact that an AI solution can have on your business. Increased efficiency that provides a better customer experience while reducing cost-per-lead? Yes, please.
That’s why, at Drift, we are actively working to enhance and iterate on our Conversational AI with generative AI. Here’s how 👇
Quick Note! Our GPT integration still includes a human-in-the-loop component, so you always have the chance to review any content being shared with your customers or prospects.
AI Suggested Replies in Live Chat
When engaging with a chatbot, 29% of buyers expect a response immediately, and 43% expect it within five minutes. Needless to say, in a live chat conversation, there’s no time to waste.
AI Suggested Replies allow sales reps to answer complex prospect and customer questions without ever having to leave the conversation — no need for Google! Not only does this help customers get the answers they need faster, but it also accelerates new sales rep onboarding by providing on-the-job training.
Topic Example Generation
Tired of saying the same thing in a different way, five times over? AI Topic Example Generation takes that tedious task off your hands.
Train the AI on one question, and the GPT integration will suggest 10 different ways that question could be asked. This way, the AI will learn to answer 10 versions of the same question while you move on to your next activity. This not only saves you time, but it also improves the accuracy of your AI as it continues to learn and update its responses.
Today’s digital buying landscape means that your prospects and customers can visit your website from any country in the world. But that also means your site visitor’s native language may not be the same as your own. Even so, they still deserve to get the information they need quickly.
Drift’s GPT Language Translations provide translations directly in live chat so your go-to-market teams never have to leave the chat to answer the site visitor’s question. When a site visitor types in a language foreign to you, the GPT will translate both the question and your answer directly in the chat, allowing you to deliver a personalized one-to-one experience in just a matter of seconds.
Why Conversational AI Works
Conversational AI is the key to boosting your go-to-market team’s efficiency without sacrificing the authenticity of your customer experience. Because not only can you scale personalized experiences through tailored messaging, but you’re also able to qualify and nurture buyers while providing helpful insights to the sales team.
Let’s look at a few ways Drift customers are winning with Conversational AI 👇
24/7 Lead Capture
After a year-and-a-half with Drift, EAB — a trusted partner to educational institutions and corporations through their best practice research, technologies, and enrollment and marketing services — adopted Drift’s Conversational AI to try to capture leads outside of normal business hours.
By adopting the crawl-walk-run approach to Conversational AI (more on that later), EAB trained their bot to respond to their site visitors’ questions with 95% accuracy. The results? They doubled the amount of qualified leads coming in through Drift and saw a 120% increase in demo requests.
Deflecting Support Questions
Tenable, a vulnerability management platform, first adopted Drift to help its sales development representative (SDR) team drive more MQLs to the sales team. And it worked! By training its top performing SDRs on Drift chat, Drift soon became one of their marketing team’s highest converting programs.
To further increase their SDR team’s efficiency, Tenable worked with Drift to train their Conversational AI to handle the support questions that came through chat. In doing so, Tenable’s customers were able to get to the right resources, faster, which meant that the SDR team could stay focused on more revenue-generating activities. The results? The team was 3x faster in delivering SDR qualified leads to sales, and they saw a 30% improvement in conversation quality and length.
Boosting Operational Efficiency
Providing a seamless customer experience is critical to a positive website experience, and Wake Forest University’s School of Business knows this first-hand. Wanting to provide a digital experience that could live up to its world-class recruitment team without putting more work on the team itself, the School of Business adopted Drift’s Conversational AI to move prospective students through the funnel faster.
By enhancing their attribution and tracking capabilities on behavioral-based activities, the School of Business trained their Drift bots to serve up the appropriate solutions to their students, be it setting up a financial aid consultation, connecting with a dean, or otherwise. The results? Hours saved on the recruitment team’s bandwidth and 65% more applications from inquiries using Drift.
The Crawl-Walk-Run Approach to Conversational AI
If all of this Conversational AI talk is sounding good to you, then why not get started?
The good news: It doesn’t have to happen all at once. If you’ve already got Drift playbooks running on your site and want to start automating them, we suggest taking the crawl-walk-run approach 👇
Quick Note! Drift’s Conversational AI offerings are only available on our Enterprise plan.
Like anything, the first thing to remember is to “turn on” AI in your Drift playbooks.
From there, you can choose where you’d like to start based on the over 60 pre-trained AI conversation topics available out-of-the-box. These bot flows have been generated after analyzing over six billion conversations. So, all you need to do is add responses for your site visitors to choose from.
- Make sure your responses match your brand voice.
- Start your AI on a catch-all page, like your homepage.
Once you’ve enabled your Conversational AI, give it some time to gather information.
From there, you can use the “Review” tab to look at all of the conversations the AI chatbot has had with site visitors and rate it with a thumbs up or a thumbs down. This will help the AI understand what makes a good conversation.
In addition, the “Discovery” tab shows you all of the questions the AI chatbot was not confident in answering. Take a look at these to get inspiration for some new conversation topics or to iterate on your existing answers.
Once you’ve got your Conversational AI up and running in one playbook, it’s time to expand!
Start trying out the AI on other high-traffic pages, like the pricing page or careers page. Or, you can take a stab at AI-branching so that, when a site visitor types in their own answer to your chatbot’s question, your chatbot will either branch to another AI topic or down the Unrecognized topic path (which you can then use to train your AI on later).
Conversational AI isn’t one of those “set it and forget it” solutions. With every conversation the AI has, it collects more data and insights into the customer journey and buying experience. Use this information to continue iterating on your AI, as well as your overall website experience.
AI Is No Longer the Future. It’s the Present.
Although it may seem like the launch of ChatGPT is what brought AI into the mainstream, the fact of the matter is that it’s only accelerated what was already coming our way.
Soon, artificial intelligence will no longer be a nice-to-have but a mainstay in any digital strategy. Because, when implemented correctly, AI has the potential to transform the digital experience for both you and your customers — for the better.