Dreamforce 2018 – why AI isn’t yet smart enough to deal with Wendy’s customers
- Summary:
- Three Salesforce Trailblazer customers open up on going digital - with some caution around AI's capabiltiies.
Artificial intelligence (AI) is being hailed as the foundation of the Fourth Industrial Revolution, and the entire tech industry is certainly excited about its prospects – case in point, the Apple/Salesforce tie-up announced this week at Dreamforce that should see Siri built into a lot more mobile business apps and saw a share price boost for both firms.
But outside of the tech sector, there are still concerns about putting our faith in technology over human interaction. While many are wringing their hands about potential job losses from AI, fast food chain Wendy's has a more immediate and practical worry: what if the technology offends its customers?
Speaking on a customer panel at Dreamforce, Oana Polanco, CRM Manager at Wendy's explained that AI posed a lot of challenges for her organization around how people phrase their complaints in the food service industry and the myriad things that could go wrong:
You could have a missing item but the cashier was really nice, so you’ve got multiple utterances. The customer wants to tell you that their item was missing, maybe you gave them too many ketchup packets, but the operator was super nice and they want to give her a compliment. Now you’ve got three separate utterances in one phrase. The AI, the state of technology today, can’t really parse out those individual utterances that might indicate one intent and that might be the wrong intent.
In cases that are very sensitive in nature, like ‘hey my child had a piece of metal in their food but the cashier was very nice’, and the bot says something back like ‘oh that’s great to hear’, now you’ve set a very different tone for that conversation. I think we’re a little bit further away from that bot AI where it can handle the complex conversations our customers are expecting.
Wendy’s, which has so far been slow to embrace trends like mobile ordering and online delivery, is keen to avoid any AI that is so conversational that the bot can’t understand what the customer is inferring or get confused, and instead the firm is looking at training AI in single, more simple utterances.
Opening the door
The Chamberlain Group, which manufactures garage door opening and gate entry systems, also has a cautious attitude to AI. The firm went live on Salesforce Service Cloud in March 2017, and only switched on the chat function this May.
Vicky Eimer, Manager, CRM Applications & Strategy at the company, said that from an AI standpoint, it’s still learning. The Chamberlain Group is using AI for call reporting, and mining that data. The next step is figuring out how to partner that with Salesforce to get the right information into its agents’ hands.
DoorDash is taking a more progressive approach to AI. The restaurant courier service, which also happens to be the delivery partner for Wendy’s, sees it as a way of speeding up the customer service process. Gina Wiley, DoorDash Program Manager, explained:
We’re really trying to dive headfirst into Einstein Analytics and do as many pilots as possible. Our main focus for now is empowering our agents to be able to answer customer questions really quickly.
Einstein is really great, especially as we’re on-boarding so many individuals per week and we can get them on-boarded really quickly. I can only tell you how many different issues can affect a delivery, you would be shocked. They need to know every singe edge case, so we have to be able to enable our agents to solve the problem really quickly and give the right solution.
DoorDash is also looking into how it can use an Einstein bot for external users, so if somebody wants to add an extra serving to their order they are able to do that straight from a chatbot.
Underpinning this rollout of new technology is an awareness of not wanting to overwhelm the firm’s customer service agents with too many new features. DoorDash has grown from 200 to 3,000 agents in a short period of time – the company was only founded in 2012 - and change management can be difficult with agents living all over the world.
At peak dinner times, just trying to get out new promotion information to agents takes a lot of time, and so the focus is foremost on ensuring delivery contacts can chat to the firm through the web or phone or social channels, rather than everything funneling through just one channel.
DoorDash is a very data-driven company, according to Wiley, and it has some main metrics that it’s always looking to assess, including how many cases agents can handle in an hour and the case deflection rate:
Moving to digital, we’ve been able to really see a smoothing, especially at the peak dinner and lunchtimes, of when people are contacting us and how they’re contacting us, and also the information they’re giving to us so the agents can work on that case immediately after getting it.
Despite its hesitance over AI, Wendy’s is still an advocate of going digital. Polanco explained that Wendy’s web-based customer service offerings have led to not only happier customers but also happier agents, as the conversations in online chat tend to be more positive in nature:
You can be really mad at someone and put things in really large font and make it red and bold with lots of exclamation marks, but it just does not take the same toll on a person’s feeling and emotions as yelling at someone over the phone.
We’re trying to blend it and offer all our advocates the experience of digital channels so everyone has a positive day.