If you have seasonal volume, it can be difficult to offer a consistent level of service at peak times. E-commerce customer service teams: we’re talking to you.
During peak season support tickets and wait times increase. High call volumes or real-time web chat requests become a strain on your center agents’ work lives. That was the case even before the Covid-19 pandemic.
Trying to optimize staffing is a tough job. It’s not possible to snap your fingers to grow and shrink your team at will. Hiring and training new customer support agents takes time. Even if you do manage to grow your team in time for the busy season – what happens to those valued team members once the quiet season starts?
Human power isn’t a workable option for managing big spikes in volume, so we need to look for other options. This is where artificial intelligence can lend a hand.
Working with AI solutions
When quality-focused customer service teams use a smart AI solution, they’re better equipped to deal with volume surges. AI tools don’t need breaks, they are available 24/7 and they don’t need overtime pay.
But heavy-handed automation doesn’t always make for the best customer experience. If you use AI to prevent customers from getting to a human without giving customers a helpful resolution, quality is going to suffer. Instead, you should use AI in a way that’s thoughtful, well-designed, and integrated into your customer journey.
Read on to learn how AI can help your team manage spikes in volume, while still maintaining a great customer experience.
Find low hanging fruit for automation
Customers want fast responses. According to Salesforce research, all generations of customers prioritise convenience. Beyond using self-service knowledge bases or frequently asked questions pages, you can use AI.
Image source: Salesforce
If the most convenient way for them to get help involves AI, they still think they are getting a great experience.
Customers are becoming used to using chatbots instead of calling a contact center, they don’t solve all customer problems. In fact, 75% of customers think simple chatbot messaging makes the interaction impersonal.
Finding the tasks that are the best for AI is the first step to implementing a solution that works for both you and your customers. Repetitive tasks are often the most appropriate conversations to automate. Some common questions that e-commerce companies can automate are:
- Refunds and Exchanges
- WisMO or “Where is My Order?” questions
- Sizing inquiries
- Delivery updates
Almost half of customer service representatives agree that 30% of inquiries are repetitive, yet easy to solve. And even more (80%) agree that at least 20% of all the questions they see are repetitive and easily solved. These inquiries are mind-numbing for your agents.
When volume spikes, most of that new volume comprises basic, repetitive questions.
You’re the expert in your team’s workload: how many incoming questions do you think a conversational AI platform could manage?
Why are you making your customers wait for an answer from a human when automation would give them a better experience?
Once you’ve identified the types of questions that automation can resolve, it’s time to build out your workflows. Most AI platforms are limited to responding to customer questions with the best possible answer.
These answers develop from previous conversations or from knowledgebase articles. Automating conversation is a good first step. It looks something like this:
“Hi! Can I exchange this jacket that I bought? It’s too small.”
“Hello! I’m happy to help. We accept exchanges for 30 days after purchase as long as the tags are still attached. Did that answer your question?”
While the automation technically gave a correct answer, it’s not actually resolving the customer’s inquiry. To do that, the AI needs to automate processes, not conversation.
Process automation uses an AI platform with backend systems and APIs — creating an omnichannel system. This makes it so that automation can take action.
With access to the order system, process automation can retrieve delivery information, and confirm refunds. It can even personalise suggestions based on previous customer behaviour.
Compare the example above with this process automation:
“Hi! Can I exchange this jacket that I bought? It’s too small.”
“Hello! I’m happy to help. Can I have your order number please?”
“Thanks! It’s #3458286.”
“Great. We have the same jacket that you bought available in a medium. Do you want us to ship it to the same address?”
“Okay, we’ve placed that order for you. Please return your previous order to 123 Brand Street, New York, NY so that we can confirm the exchange.”
The process automation workflow understands the customers’ intent. It asks clarifying questions, takes the necessary action, and confirms with the customer. If the issue was not resolved for whatever reason, the customer can transfer to a live human for further troubleshooting.
Agent Guided Automation
Automation can’t (and shouldn’t) answer every question. For conversations that need human touch, nothing beats a well-trained customer service agent.
Unfortunately, customer service teams are stretched thin already. The average customer service agent uses between 5 and 8 different systems to resolve customer questions. From billing platform providers to CRMs, agents are flipping between screens and tabs trying to find information. We call this “swivel chair” and it’s a recipe for slow response times and missing information.
For these cases, an integrated AI platform can become an agent’s best friend. On each incoming customer question, the platform can retrieve all the information needed. It will then suggest the best answer and even perform the necessary backend tasks that the agent requires.
Agent Guided Automation might look something like this:
“My order #459872 is missing and I ordered it three weeks ago! This is unacceptable.”
[[AI detects frustrated sentiment and routes question to an agent]]
[[AI platform retrieves shipping information, customer order ID and customer history]]
“Hello! I’m so sorry to hear that, that’s not normal. I’ve had a look into our system and it shows as delivered and signed for on April 27th. Is there anyone else in your business that could have possibly received the order?”
The agent can review the information gathered by the AI. Then select pre-written response templates composed by the AI, personalise if needed, and then send it to the customer. The agent has full oversight of what the customer receives, but their job is much easier.
Agent Guided Automation reduces average handle time and prevents agent burnout from “swivel chair”. In essence, it can help new agents get up to speed faster!
While the above scenarios talk about a more reactive response to customer inquiries, AI can do more — proactive automation.
Instead of the customer contacting you about an order, the AI finds these issues and proactively sends email or SMS notifications to them. Customers are then updated along the way when issues resolve.
This proactive system helps to lower “call spikes”, webchat tickets, or social media messages. Customers in the loop at every stage.
Use AI to Handle Volume Spikes with Confidence
Dealing with seasonal spikes in volume is one of the most challenging parts of running an e-commerce customer service team. Customers still expect an immediate response. So stretching your customer service agents thin can result in a drop in quality standards.
Instead of relying on human agents, consider implementing an AI solution. It can help keep responses times lightning fast, even throughout the busy season.
Teams can create workflows that deliver resolutions to customers without human intervention. By providing agents with help from AI, teams can become more productive and avoid swivel-chair burnout.
These solutions result in incredible customer satisfaction metrics and operational savings. All with the use of AI, machine learning, and natural language processing. You can check out results from DigitalGenius customers in our customer case studies.
See how professional services by DigitalGenius can help your team manage seasonal volumes.