There are a lot of digital assistants designed for customer support. That has value. If you’re taking tasks off your live agent’s plate, then you’re helping them focus on more complex tasks.
However, that’s not all AI assistants can do.
Can’t they also generate revenue? Can’t they turn unhappy customers into smiling ones?
What is Conversational Commerce?
As Dawn says, “I would define conversational commerce as any way that a conversational AI assistant can generate revenue for a company.”
Dawn gave an example of a project she worked on. A company sold products that customers had seen on TV, similar to QVC. The customers would phone the call centre to order the stuff they’d seen, and first they’d be speak to a conversational IVR, which could route calls and occasionally upsell.
That’s it. That’s all you need to start generating revenue with AI.
Firstly, it’s an excellent use of conversational AI. Calls are routed before agents have to get involved which saves an awful lot of time (and therefore money). It’s the second part that deserves our attention though – the assistant can upsell.
AI is saving the call centre money and also generating it.
The important thing is to know where your bot will be most useful. You discover that by researching customer needs, and then considering where your AI assistant will be placed in the full customer journey.
Not always a quick sale
The assistant doesn’t have to act as a salesperson. It can grease the wheels for a sale to happen further down the line.
For example, when someone calls to book a restaurant table, an AI assistant might take their reservation. That call isn’t the sale –the sale will come later when they visit the restaurant. Here, though, the AI assistant has contributed to revenue production.
Or, if you’re experiencing a spike in calls, you can have the AI assistant field every call initially. When the caller wants to buy something, the assistnat sends them to a live agent. When the caller just wants to ask simple questions, the assistant can handle it. That worked very well for Landry’s Restaurants and Hospitality.
Or you can have an assistant that answers FAQs (frequently asked questions) to provide customers with information about products before a live agent takes their order.
It’s all about finding the best use case for AI to ensure that sales can happen, regardless of whether a digital assistant or live agent makes the final sale.
Dawn gave another example where an AI assistant identified customers and found out what they wanted. Then, it generated a link to send to the live agent so they can start their conversation understanding the context and needs of users. Next, the live agent talked to the customer about their needs. Finally, the agent sent links to the customer’s phone of various products that might fit their needs, and the final sale could be carried out via SMS or a phone call.
You could say that both the assistant and the live agent were vital for the sale to happen in that last scenario.
Using conversations to screen customers
Conversations aren’t great for everything. For example, it’s not easy to browse products through dialogue or via a standalone voice user interface, which makes product research challenging. Here, you need something multimodal, so that the user can ask questions like ‘show me all your widescreen TVs under $1000’ and see the results.
However, when the customer specifically knows what they want, such as “brake pads for a 1974 Chevy pickup truck,” then a voice assistant can be effective because there should only be a few (or one) options available to the customer. The assistant only needs to help them narrow it down.
Dawn says this is one area where conversational AI excels – when there are only a few options, it can screen the user with questions to define what they really need.
Perhaps someone is calling about travel insurance. The assistant can ask them questions like “is it a business trip?”, “where are you going?”, and “how long will you be gone?” to rapidly define which policy would fit best. Even if the bot doesn’t resolve the user’s need, it can collect enough information to make the live agent’s job a lot easier.
Another great use case is drive-thru restaurants. Fast food restaurants have a clearly defined menu with a finite amount of options. So long as the voicebot can deal with ambient noise at the roadside, it can take orders (this is still a complex implementation, requiring accurate identification of complex combinations of entities, but it’s doable, as HUEX is proving, along with SoundHound).
Use what data you already have
Dawn gave another great example of where conversational AI can fit into commerce. Consider this – a customer is using a brand’s webstore, they add something to their shopping basket, and then they open the chat window. Wouldn’t it be great if the first question they’re asked in the chat is “we see an item was just added to your shopping basket – do you want to talk about that?”
Data is used to help the user start their journey at a later point. They didn’t have to go through menus to say they want to discuss the item in their shopping basket.
We can do this when we have the data, and the data is accessible. You need a holistic approach to AI to make the most of your data. The data is already there and available, you just have to use it.
As Dawn says, “you know [the customers]. You know what they’re after. I’m always telling my clients to leverage that back-end information that they have. Let the system do all the heavy lifting, don’t make the user do it.”
And what about those resturning users? Some may have used your assistant many times already. Perhaps their car insurance needs renewal. As Dawn says, you could just send them a text message asking if they want to renew. That’s a very simple low-friction interaction. Most will hopefully say “yes.” The ones that say “no” should be instantly called by a live agent, and you should have the right data to tell you exactly how long that customer’s been with you, any issues they had, and what their needs are. Then you can make them an offer they’ll love.
This is all possible
All of this can be achieved. To get there you need to use AI holistically and consider the best use cases for your particular services that go beyond only providing customer support.
Then AI can generate revenue as well as saving costs.
Thanks to Dawn Harpster and Talkdesk for another excellent webinar with VUX World.