Salesforce’s new State of Service Report reveals AI is expected to handle 40% of all customer service cases by 2027 – up from the 25% already being handled by AI today.
Based on a global survey of 6,500 service professionals, including 300 in Canada, the data finds AI has leapt from the margins to the mainstream, becoming the #2 priority for service leaders, up from #11 a year ago and second only to improving the customer experience.
Key Canadian findings from the Salesforce report include:
- Service professionals project AI agents will drive significant business value, including a 20% decrease in service costs and a 20% increase in customer satisfaction scores, alongside a 20% lift in upsell revenue.
- 83% say customer expectations are higher than they used to be
- 81% of Canadian service professionals using AI say it’s making them more productive.
Michelle Grant, Director, Strategy and Insights, Retail and Consumer Goods at Salesforce, said a previous Salesforce report found that the number one opportunity that Canadian retailers see for their business is improving customer service.
“So I think that we don’t necessarily talk about the importance of customer service in the retail industry as much as we should. Retailers have identified it as an opportunity, and then when we surveyed consumers in that same report, the third worst experience that a consumer can have in Canadian retail is poor customer service. Waiting in line at the checkout is the worst experience, Canadians said. And the second one was out of stock,” explained Grant.
“And so this report comes along, and it’s really fascinating because it sort of echoes the importance of customer service and how consumer expectations are much higher than they have before, and how to meet that need. One of the key themes, though, that I really took away is just how quickly AI is being adopted in the customer service centre.
“ChatGPT has only been around since November of 2022, but as of today, this year, customer service professionals in Canada see that 25% of their customer service cases are being helped by AI, and they expect that penetration to increase to 40% in the next two years. So it’s really important to have customer service professionals using AI to deliver great, best in class service. But also resolve things very quickly and really enhance that self service option to deflect cases.”
Consumers these days want efficiency and fast service, says Salesforce.
“People don’t want to spend time trying to fix a problem that they shouldn’t have had or don’t think they should have had. With the generative AI technology in particular, because the models work so fast, for the most part, people are getting self serve responses at a much faster rate, and then they also have, if it is delayed in thinking, you can see that it’s thinking, and there’s usually an icon denoting that it’s thinking and going through the steps, whereas in a traditional decision tree chat bot, you would not get that type of notification, and it could take a bit longer,” explained Grant.
“AI is only limited by how fast the model is processing things. And at this point the latency is really, really low, and they really have worked on that, even in that consumer experience with like a ChatGPT or a Gemini.
Grant said the biggest barrier for companies in adopting AI is security concerns.
“The models do ingest a lot of very sensitive information about your end user, so making sure that those models and your process are protecting that data and are not using that data to train the model, or having that data leak into the model and show up in someone else’s search results is the biggest barrier to overcoming this technology. And the second thing is just the ability to trust the results,” she said.
“You really need to make sure that what that model is generating and telling that consumer is extremely accurate. So that means you have to do a lot of quality assurance, a lot of testing, to make sure that it is consumer facing. A human agent that might get a result may know from institutional knowledge and experience that, oh, this is not accurate. I need to override this. And so oftentimes, our companies will test with the human agent first to have them provide that quality assurance, have them tested, know things that are incorrect before they deploy in a consumer facing instance.”
More from Retail Insider:
