Conversational marketing - combining the power of AI with humans

Barb Mosher Zinck Profile picture for user barb.mosher January 30, 2024
Summary:
AI vs humans? It's not that simple when it comes to conversational marketing.

AI

While AI is becoming an integral part of marketing technology, that doesn’t mean there still isn’t a human element that needs to be considered. Nowhere can this symbiotic relationship between humans and AI be seen more than with conversational marketing.

According to the Conversational Marketing Software Market report, the conversational marketing industry is expected to grow to US $4501.6 million by 2028 (compared to $840.13 million in 2022). Vendors like Drift, Intercom, HubSpot, Saleswhale, and Conversica are well-known leaders in this market, all of which have or are building an AI story. But there are plenty of other players, some pure-play and others part of a larger platform.

Drift set out to understand what conversational marketing means to buyers by analyzing 30 million conversations that took place on the Drift platform in 2023. There were several interesting trends, but I'll focus on two key ones. 

People still want to talk to humans

According to the data, the number of requests to speak with a human has increased by 2.5 times. You can infuse AI into your bots, but it's important to remember that a lot of people still want to talk with a human, so live agents are still going to be needed.

Also, the faster they respond, the better. Agents that respond within two minutes of a site visitor engaging with a bot have the highest chance of booking a meeting. The longer the buyer has to wait, the higher the risk of them leaving. Those waiting five minutes increase the risk of the buyer leaving by 10x; a 10-minute wait resulted in an increased risk of leaving by 100x.

An Intercom study had similar findings, noting that website visitors are 82% more likely to convert if they've chatted with a company first, whether it's short conversations or longer ones. It found: 

A simple conversation with six exchanged messages makes a visitor 250% more likely to become a customer.

This is not to say that people won't engage with a bot. Website visitors engage with conversational bots for a reason, including support, billing, meeting, demo, or pricing questions. 

In this study, high-intent messages like demo and email requests and "we're using a competitor" were five times more likely to convert into an opportunity than low-intent messages (just browsing, support requests).

Three tips to get conversational marketing right

If you want to succeed at conversational marketing this year, you need to have a strategy combining AI and humans. Drift’s Senior Content Marketing Manager, Elizabeth Hilfrank, offers three tips (to which I’m offering my thoughts on using them):

Get active

Hilfrank says to enable notifications and establish routing rules. Routing rules specify what to do when a visitor selects a certain option in your bot or asks a question.

Start with AI, offering relevant content based on where the website visitor is on the site. The visitor could also select from a set of common questions, including talking to a live agent. Your routing rules need to be clear on what to offer the visitor. If the request is to talk with someone, ensure you have people available to talk to them and build the processes necessary to get that conversation going as quickly as possible. If you leave them waiting too long, Drift's study says it all - they'll find someone else who can give them the answers they need.

Simplify and streamline

I thought this tip was smart. Make the conversation easy for the website visitor. If they want to have a meeting, have a calendar option ready. If there are common questions you can make available at the click of a button, show them first, along with the ability to ask an entirely different question.

Not every website visitor will convert, so the ability to ask qualifying questions is essential. Hilfrank suggests that the highest-converting qualifying flows in 2023 included 3-5 questions. Any more, and you might tick the visitor off. Work with your sales team to identify those questions, being careful not to ask questions that get too personal before the buyer is ready to share. 

Build credibility  

This tip is about allowing the visitor to find their own path. The common questions /buttons mentioned above apply here. It's also what comes after they click a particular button. What does that full path look like? 

You don't want a visitor to click a button with a certain expectation and have something completely different happen. What does the typical journey look like when a button is selected? Map that out in your playbook and implement it. But also remember that not every visitor is the same, so having the ability to go outside that typical path is important. 

Will live agents still be humans?

One of the big questions around conversational marketing is whether the live agents will be real humans or an evolved version of chat called conversational AI. A growing number of solutions, including Drift, allow the AI to talk to visitors as if they were talking to a real person. 

Some companies are trying it. It can work if they have trained their AI on the right data and have regular checks on the quality of that data and how the AI responds. But be sure you know it's working correctly. When buyers are researching without direct contact, you need to know you are giving them the information they need to get your company on their shortlist.  

Drift refers to it as its bionic chat. Matt Tippets, SVP of Product, explains that the firm’s new product offering is designed to engage with people but be ready to route them to a live person at the right time. 

The goal is still to drive business, not have never-ending conversations, so the bot will automatically identify if someone is expressing support behavior or purchase behavior. When it does, it will ask the visitor if they if they need specific help and then route them down that specific channel.

There are two important notes there. First, do you tell the visitor they are chatting with an AI (will the visitor care as long as they get the necessary information)? Second, how do you know the right time to route them to a human? The routing process requires careful planning and training of the AI to understand the parameters necessary to get a human involved.

My take

We're a little over a year into generative AI being out in the world. (It's been around a lot longer, but let's not get picky). Everyone is working to apply gen AI to their solutions, and conversational marketing is a prime use case. But that doesn't mean it should be all about AI.

If you take anything away from Drift's study, it's that there is still a role for real humans to play, and companies that employ chats  on their websites must think carefully about where and how to bring them into the journey. 

Loading
A grey colored placeholder image