As a retail leader, if by now generative AI is not on your agenda, then it really should be. Amazon CEO Andy Jassy says the firm is investing heavily in generative AI. Fashion brands have already started to use ChatGPT to provide outfit recommendations (not wholly successfully so far), and a myriad of other uses are springing up.
One of the key uses that is coming up is in customer service. Old-school chatbots which have a limited number of pre-programmed conversational journeys are one area that is ripe for innovation with generative AI. Conversation Designers can already benefit from generative AI to create multiple variants on the same message, to give automated chat the feel of a real person.
This is just the beginning. Generative AI can be used in so many ways that will benefit the customer experience… but if you’re thinking of just integrating ChatGPT and letting it run wild, then you are running the risk of a terrible customer experience.
In brief, here are some risks you need to watch out for, but also the rewards that you could achieve if you embrace generative AI in customer service.
The Risks
Risk 1: providing the wrong answers
ChatGPT is known for giving incredibly convincing answers to questions. But it is also known for providing “hallucinations” – complete flights of fancy with details conjured out of thin air.
This is because ChatGPT is trained on a very wide set of data and doesn’t always “know” when things are true or false.
So by merely plugging in this technology, you would open yourself up to giving customers the wrong answers. However, if it is trained on a limited set of data (say all of your FAQ pages) it can give more accurate information. So all is not lost.
A question to ask any supplier: What data is your generative model trained on? How can the answers be limited to just my brand’s domain knowledge?
Risk 2: Irrelevant answers
Another problem with plugging ChatGPT in is that people may want to just have fun and ask a chatbot something random. If you are paying every time the technology answers a question (which you will under certain plans), then you don’t want to be wasting money answering silly questions like: “Is the moon made of cheese?”
Even if the question is relevant, if the answers are not limited to what is in your company information (FAQs, Product Descriptions, Shipping Policies, etc.) then you may be paying for an answer that doesn't actually help the customer find what they want.
A question to ask any supplier: How do you limit what questions the AI answers?
Risk 3: limited use cases
Generative AI is pretty great at answering general questions, but what happens when you get more specific? For example, how many questions do you get which are along the lines of “Can I have an update on my order?” compared to questions like “Does this product come in large?”. For most retailers the first group will be a much bigger group than the second. The second is what generative AI can be great at answering, but the former may be more useful.
However, if a generative AI engine can be connected to backend systems such as shipping carriers and warehouse systems, it's possible for the AI to answer these specific, personalised questions. The answers will be personalised based on that specific customer’s order history, which provides a far richer experience overall.
A question to ask any supplier: What other systems can you integrate with?
The Rewards
These risks are easily avoided or mitigated if you go in with your eyes open, consider your entire tech stack, and ask the right questions to any supplier. With those risks managed, the rewards can be significant. Here is a by no means an exhaustive list.
Reward 1: a more personalised experience
If Generative AI can be trained on order history, then it can know everything that a human agent would know when they initiate a chat. With this knowledge, responses can then be personalised, responding to VIP customers differently than new customers who need a different experience.
But more than that, the AI can offer product recommendations based on order history, product description and what a customer says they want.
And while all this is going on automatically, human agents have more time to spend on the cases that they handle – enabling them to provide a more measured and personalised experience.
Reward 2: round the clock sales support
It’s always a better experience if you can get an answer to a question faster. Generative AI doesn’t abide by office hours, meaning that your customers can get answers to their questions at any time of the day.
More than that, the generative AI can help customers find the right products to fit their needs, recommend products that can go along with them, and help smooth the journey to purchase by answering any further problems. This means you don’t lose customers who have to wait a day for a response to their issue, but can help them in the moment.
Reward 3: turn customer service into sales
As mentioned above, with more inbound requests being addressed by a generative AI, the human agents in your team have more time to add value to every interaction they have. This could include upselling, cross-selling, and even reaching out proactively to existing customers to see if they want to re-order.
Reward 4: tap into the ChatGPT userbase
The final reward we’ll cover for now is being able to integrate your product line into ChatGPT. Since launch, ChatGPT has reached one million users faster than almost any other technology. As people start to use it instead of Google to find answers, what if they are looking for information on products that you sell?
By integrating your product line with ChatGPT and feeding up to date information, you can ensure that when people are asking about your brand and product, they can make the right decisions.
As the userbase grows and grows, this could be a new channel to find customers.
If you want help on how you can integrate this technology into your business, get in touch with one of our experts today and book a demo.
