Explore the future of customer service with insider insights from YoChats. Learn about the role of AI, the skills required for different support levels, and how to optimize your customer service team for success.
Today, we are entering a new era of fully automated customer service. Many companies are already developing AI customer service chatbot that will instantly respond to your customer's questions. These AI customer service managers have a complete understanding of your product.
YoChats is developing a platform that will allow you to create your own AI operator from your knowledge base in just 30 seconds. This AI customer service operator will provide support for your website or application 24/7. For only $15 and in just 30 seconds, you can fully automate your customer support process.
Obviously, when you're creating a product, it's necessary to conduct in-depth interviews with product owners or potential customers in order to fully understand all the pains and issues of business processes. Recently, YoChats conducted an extensive customer development study. Today, we want to share insights from the world of traditional customer support and publish the key important points from our interview.
To begin, let's describe our interviewee in order to better understand who we were talking to and what knowledge our interviewee might have about the customer support process. Due to a confidentiality agreement (NDA), we won't be disclosing the name and surname of this employee.
The main goal of any support team is to effectively solve any issue a user brings to us. We have two key quality criteria for solving problems:
In the companies where I've worked, we always prioritized quality. This means that solving a task with high quality was more important than how quickly it was solved. However, I've heard that there are companies where customer service managers prioritize speed.
Today, I will highlight two main approaches in supporting a system:
In my opinion, working in a company where customer service is provided through the Tickets system is much simpler. For instance, handling 50 chats per day for an average support service is practically an unrealistic burden, whereas processing 100-150 Tickets per day is a realistically manageable task.
We have two main levels of support:
Sometimes in some companies, they also have a Level 3, which involves product enhancements. I don't consider this as part of support, but rather as product changes.
At this level, the support department employees simply distribute incoming requests among different departments and products. Typically, Level 1 support employees respond using templates and actively utilize macros.
For instance, in the gaming industry, customer service managers could be addressed questions like:
The primary rule for Level 1 responses is that if the answer exists in the knowledge base, then they provide it. If the answer isn't in the knowledge base, the request needs to be escalated to Level 2. Speed is crucial at this stage, but delving deep into the issue is not encouraged.
Normally, all Level 1 customer service managers' responses are based on information available publicly within the company (such as FAQs, landing pages, and guidebooks). An important rule is that any question handled by a Level 1 customer service operator can theoretically be resolved by the user independently.
If a user's question can't be solved at Level 1, the customer support manager's task is to quickly escalate the request to Level 2. Usually, in this case, the manager identifies the relevant product and sends it to Level 2, to a specific department responsible for that product. In the companies I have worked for, there's a product division within Level 2. This means that inside Level 2 support, there are several departments, each responsible for different products. For example, in a gaming company, there was a separate Level 2 department for each game.
Sometimes, companies organize their support based on functional principles. For instance, there could be a Level 2 support division specifically for payments.
Sometimes, the Level 2 team can handle multiple products (several game projects). The product's architecture sets its own conditions. For instance, if each product has its own billing system, the corresponding product team supports it. However, if the company has a unified billing system for all products, a separate Level 2 support team is assigned.
Employees in the Level 2 department handle more complex technical issues. However, it's important to understand that at this level, employees are not involved in making changes to the product; they are focused solely on providing support. Here are some examples of tasks carried out by Level 2 customer service staff:
Employees in the Level 2 department have various levels of access for viewing and sometimes even editing databases. Some employees might even have access to payment terminals.
A customer service manager in the Level 2 department should possess deep knowledge about the product. Additionally, it's important for this level of support staff to have technical skills. Furthermore, employees at this support level are expected to provide high-quality and detailed responses to customers, which is why they need to have a strong command of the language.
– Typically, these are altruistic individuals who enjoy helping others. Otherwise, such a person might offer poor support in the long run.
– If the company provides support through live chats, the candidate must be exceptionally stress-resistant.
– The potential candidate should have impeccable writing skills, meaning they can compose correct and understandable texts.
– Additionally, the candidate must be highly trainable, as products often undergo constant changes, requiring continuous adaptation.
– A candidate should be able to independently find information, for example, know how to use Slack's search function or write effective search queries in Google.
– It's evident that a candidate should understand technical matters, meaning they should have some sort of technical background.
– An ideal candidate must possess excellent communication skills and demonstrate loyalty to customers in a courteous and respectful manner. All interactions should be polite and culturally appropriate. For instance, if a certain feature used to exist but no longer does, you should empathize with the user's disappointment and explain that you understand their dissatisfaction, but unfortunately, that feature is no longer available.
– Of course, a candidate should be creative. This is crucial for solving extraordinary questions that often arise at Level 2. Such a candidate seeks and is determined to find solutions. An ideal candidate will always say, "Let me try, and I will definitely find a solution." He always aims to solve the user's problem using any means necessary. A less suitable candidate typically just responds that they have forwarded the ticket to technical specialists.
– The ideal candidate should have a genuine interest in the product.
We use macros and template responses. But with this, we need to be careful – it's extremely important to always maintain personalized communication. That's why it's crucial to refine all our templates manually for personalization. For example, if a user complains about poor image quality and is unhappy with the picture, we look at their ping and the percentage of connection loss. Then, we find a template for situations when a user has a poor connection and provide a standard-like response:
"I can see your logs, and you have this type of connection. The image quality is significantly reduced due to network issues. Are you connected via cable or Wi-Fi?"
Then, you personalize the response based on the situation using response templates. Providing detailed and clear answers is important. That's why templates are necessary – to avoid typing a larger amount of text by hand.
In my opinion, Level 1 customer service can be fully automated using new AI technology. For instance, the same ChatGPT, with access to the same knowledge base as Level 1 customer service managers, can provide instant and highly accurate answers. There's no doubt about that. Moreover, such an AI customer service chatbot can work around the clock.
For Level 2 department managers, ChatGPT can genuinely assist and streamline work. For example, we often deal with logs and analyze them. ChatGPT can analyze logs within seconds, whereas we spend minutes or even hours on it.
Large companies prioritize security and often, according to their internal regulations, using external products for customer service automation is not allowed. Such companies frequently develop their own chat widgets and equivalents to Intercom. The adoption of modern technologies in these companies is limited by their internal resources.
For us, user ratings upon the completion of interactions are a highly important indicator of the support department's quality of work. At the end of each reporting period, we review all the negative feedback, focusing solely on the negative aspects. Understanding the support team's weaknesses is crucial to us.
Often, users are asked questions at the end of their interactions. This survey shouldn't be complex or extensive. It's best to provide the user with the option to rate from 1 to 100. We mustn't complicate the user's journey. Our main goal is to determine if the issue has been resolved. A great example is Binance, where users are offered a choice of 4 smiley faces. It's important to make this survey optional. Typically, feedback is left by either very satisfied or dissatisfied users.
It's also important to understand that negative feedback should be interpreted differently across various industries. For instance, in gaming, a negative experience can be a good indicator of support quality. However, in fintech projects, negative feedback doesn't always reflect support performance. For example, if someone lost $5000 due to theft, he would be unhappy, but it's not the support's fault. Hence, negative ratings don't always accurately assess the support team's actual performance.
The performance of a support team is usually assessed based on the following factors:
– First response resolution rate. This metric, known as first resolution measures how effectively support can provide a quick, concise, and clear response or guidance to the customer in the initial interaction. This approach benefits both the support team by saving time and the customer by swiftly addressing their query, without prolonging the conversation.
– Response speed. Responding promptly to incoming user inquiries is crucial for a customer service team.
– User satisfaction level. It's important to note that, for instance, negative ratings are calculated as a percentage of the total number of received ratings. For instance, if there were 500 chats during the reporting period, and only 100 of them received ratings, with 90 of them being positive, the support team's rating would be 90%.
– Number of support requests. This is a significant metric for the support team to monitor, as it determines how many more employees need to be hired in case of an increase in the number of requests.
– Quality assurance (QA) rating. QA also plays a vital role in the support process. QA team members randomly select, for example, 5 support cases. They evaluate how each user request was handled. The resulting ratings are then averaged into a single score.
Everything depends on the market and industry in which you provide support.
For instance,
Level 2 customer service managers often transition to product teams, but rarely to development. This rule especially applies to the gaming industry. In fintech companies, such specialists might move to compliance. Sometimes, some might transition to QA. It's important to note that Level 2 support managers, more than anyone else, understand all user pains and issues. That's precisely why they are ideal candidates for product teams.
