What would happen if you smushed an entire holiday season, with all its intensity and high stakes, into a period lasting less than two weeks? And just to make things even more complicated, what if your (perishable) product was a gift laden with significance because of the emotional ties between the giver and the recipient?
Those questions aren’t theoretical for UrbanStems, a DTC florist that must deal with two of these ultra-compressed holiday seasons — Valentine’s Day and Mother’s Day. Among other things, this requires UrbanStems to increase its customer service staff from approximately 16 people to 200 during these high-pressure periods.
It could all be a recipe for a customer service disaster, but UrbanStems’ deployment of AI-powered chatbot solutions from Zendesk has helped the retailer actually raise customer satisfaction scores by:
- Using the solution’s sentiment analysis functionality to triage the customer service “inbox,” helping identify angry, anxious or frustrated shoppers who will need a human touch to resolve their issues;
- Helping temporary agents reach maximum efficiency levels by their second shift, rather than their third or fourth shift as was the case previously; and
- Identifying and quantifying the biggest sources of customer frustration and using this data to prompt executives to make improvements.
Mother’s Day Brings ‘Dramatic Surge’ in Need for Customer Service
“Mother’s Day is our Super Bowl; it’s even larger than Valentine’s Day,” said Laura McDonald, Director of Customer Happiness at UrbanStems in an interview with Retail TouchPoints. “Also, it’s not the slow crescendo that starts on Black Friday; it’s a 10-day ramp-up, starting about seven days prior to the holiday and lasting three to four days after.”
Advertisem*nt
While UrbanStems has been using Zendesk solutions for nearly six years, this is the first time the retailer has deployed its AI tools to support customer service. “Prior to using Zendesk, customers had to tell us all the issues [they were experiencing], and our agents had to suss out the root cause of their issue,” said McDonald. “Zendesk gathers everything up front and asks [customers] all the questions we need answered to serve them more efficiently. It gives us, and the customers, time back.”
If the chatbot is able to solve the customer’s issue without requiring a human agent, that’s all the better. “Gen X, millennials and Gen Z often don’t want to talk to a human; they want a quick self-service option that they can do on their time,” said McDonald. “No one wants to sit on hold with customer service, and we’re still making a human available if they want.”
UrbanStems also has been able to shape these chatbot interactions to reflect the retailer’s customer-facing persona. “I’ve been able to upload all our internal documents so the [chatbot could] learn our brand tone,” said McDonald. “It now speaks to customers in the same way I’d speak to a customer.”
Dealing with the Emotions Involved in Gift-Giving
That tone is important given the nature of UrbanStems’ products. “Sending flowers is a deeply emotional thing,” McDonald noted. “Some people are constantly checking the tracking and anxious to make sure it arrives.”
The Zendesk solution’s sentiment analysis functionality is key to its ability to triage and serve customers whose emotions are trending negative. “Are they angry or anxious? The solution can tag it as such,” said McDonald. “This allows us to surface the customers that need a bit more TLC or a human touch sooner, someone who can say, ‘We’ve got this, let’s work through this.’”
AI also is instrumental in helping the nearly 200 temporary agents UrbanStems hires for these compressed peak periods to master their responsibilities more quickly. McDonald explained that the company monitors agent efficiency, and that prior to this the temps didn’t typically reach maximum efficiency until their third or fourth shift. Normally that would be a steep-enough learning curve, but not during these high-intensity crunch times.
“It’s very expensive to have [temp agents] operating at less than optimal efficiency for a third of the time” they’re working for us, said McDonald. “Now, with Zendesk, they’re hitting it by their second shift.”
AI has been improving all the agents’ work lives, she added: “Agents are finding that customers are not as angry because of not having to repeat themselves,” said McDonald. With AI, “agents have all the info they need to solve an issue and they can jump right into solutioning, which gives them more of a sense of satisfaction. We’ve also seen attrition decline fairly significantly, and we love to see that because training does have a cost.”
Quantifying and Addressing Customer Frustration
The solution’s ability to gather data around customers’ frustrations also has helped UrbanStems “identify certain parts of the business that aren’t hitting for customers,” in this case order tracking, said McDonald.
“We were flying blind about this before, because it’s difficult for a human to say exactly how angry or frustrated someone is — it’s subjective,” she noted. AI provides a “consistent, measurable way to find out how much frustration is being driven by a lack of transparency in order tracking.” These metrics were shown to upper-level executives, and the result is that UrbanStems is now “overhauling this entire system,” said McDonald.
She sees even more uses for these AI-powered solutions going forward, including improvements to internal operations that were recently introduced. “Flowers are perishable, and sometimes they’re just not quite right” when they’re delivered from a farm or fulfillment center, said McDonald. The solution “gathers all the info when, say, the dozen roses someone ordered aren’t right, in order to identify the next best option in our system, and it automatically sends an email to the customer [about the substitution],” she noted. “Prior to the use of AI there would have to be a human checking the inventory, ensuring that it’s the correct upgrade to make and reaching out manually to the customer, which is very time-consuming for the business.
“It’s been really fun to implement and also see how we can get creative with it,” McDonald added. “Every time I think we’ve exhausted it, we find a new use case.”
