When is it best to use chatbots vs. humans for customer service?

A common misconception is that chatbots are intended to solve 100% of problems and completely replace human agents. This is not the case. At this time, chatbots are most useful for solving the repetitive interactions that humans don’t need to spend time on—which allows humans to focus on the high priority case drivers where they’re needed most.

There’s a time and place to offer support with a chatbot versus a human.

Chatbots and the 80/20 rule

Support follows the 80/20 rule to a certain extent. When it comes to support, there are high frequency, common questions that tend to represent about 80% of your total support volume. The other 20% are usually outlier questions or issues. While the ratio may not always match, it is fairly safe to say that the majority of companies have a small variation of questions that represent a large portion of total support volume, and therefore cost. These high-frequency questions tend to be low in value and simple to solve without human intervention. Common examples include account balance lookups and password resets or login failure—actions a bot can do instantly.

Low value, repetitive questions are typically solved with a scripted reply or email macro, and the solution is the same for every customer. There is little value in the human component of the interaction, as the primary human elements are stripped out in favor of an efficient and repeatable script. According to Harvard Business Review, “84% of customers would prefer a straightforward solution to their problem” anyways.

Companies with higher ratios of these repetitive tickets will derive the most value from bots. Bots excel at reducing the volume of these highly repetitive inquiries while providing a superior user experience since they are always available and always instant.

A customer should not have to wait in a long queue to get a simple answer on how to reset a password, and likewise, an agent should not have to spend their time explaining how to reset a password 20 times a day. There is substantial value extracted on both sides of the interaction— a customer saves time, can solve the problem on their own, and gets back to engaging with your site or application sooner. The agent is able to avoid a low value-repetitive inquiry, and spend more time focused on higher value, more rewarding interactions.

Let’s break it down. Here’s how the business wins across the board:

• 1. The customer has a better experience and a better user experience means they’re more likely to return as a customer and more likely to make a purchase or keep using a service.
  2. Support volume goes down, which unlocks efficiencies in your support team and drives more high-value interactions.
  3. Agent attrition can decrease as the job becomes less repetitive and more valuable. The higher the frequency of these interactions (the closer they follow the 80/20 rule) the more instant impact a bot can have.

However, if a company has no common questions and support is more of a one-on-one relationship solving unique problems, that company is not as well suited to use a bot. There are also times when bots simply don’t cut it.

In general, we see three buckets of case drivers that have had more success with human support:

Win backs: When a customer is dissatisfied with a product or service to the point where they cancel the service, agents are trained to talk customers down and offer empathy. If we want empathetic bots, we have to remember that our technology is only as empathetic as we are.

Upselling: When a customer indicates purchasing intent, a human agent may be able to offer other products the customer might also be interested in, whereas a bot will typically just process the transaction requested. Furthermore, us humans know when to use discretion to offer discounts and upgrades when appropriate.

Complex troubleshooting: While bots do a good job offering basic troubleshooting, the number of variables and the complexity of deeply technical troubleshooting makes it better suited to a well-trained agent.

Beyond these three buckets, there are also clear times when a transfer to a human agent is necessary.

For example, most customers that contact support are already frustrated to some degree. If a customer continues to express negative satisfaction throughout their bot experience, it’s best to get them to a person as soon as possible—especially if a bot isn’t understanding the customer’s question. If the bot is designed to answer common repetitive questions, there will be some gaps in what it can help with. In the case it can’t recognize a user’s question, you don’t want to keep the customer stuck in a loop of asking questions and being told the bot doesn’t understand. After the customer has attempted to rephrase their question, it’s best to offer an escape path to an agent to avoid a poor customer experience.

Lastly, if you do need to make a transition from bot to human, it’s crucial that transition is seamless. Nobody wants to be duped into thinking they’re talking to a human when they’re talking to a bot—be honest and transparent with your chatbot practice. At Zendesk, we’re able to provide our customers with the flexibility to do this using our Conversations API. You can learn more about our chatbot solution here.

Better together

If you’re thinking of using a bot, make sure to first identify your company’s problem area. Are you inundated with requests and inquiries? Could your agents be spending their time more efficiently? It’s equally as important to understand the scope of the bot you might be implementing. Does it help with repetitive tasks? Or is it an expert on a certain product? A value-adding bot is one that is capable of collecting customer information with an intent to arm agents with the info they need to save a customer, solve complex problems, or drive a new sale.

There are times when it’s best to use a bot versus a human, but when it comes to tackling a range of tasks—from simple and repetitive to sensitive and complex—bots and humans work better together.

You can learn more about chatbots and the future of automation here. Don’t have a chatbot yet? Learn more about our solution here.