Is Your Live Chat Disaster-Proof? How to Avoid These Real-Life Chat Fails

We all like to look at the positives within customer service – its ability to surprise, delight, and offer real help which makes people’s lives a little bit better.

But a lot can be learned from looking at the negatives too.

And it’s certainly better to learn from the mistakes of others than to go through that pain yourself.

Here are five real-life examples of live chats gone horribly wrong. Systemic errors within software setup, resourcing, and agent training in these examples have led to nightmare customer scenarios going viral, with companies scrambling to control the damage done after the event.

Don’t fall into the same traps these companies did – read on for our advice on how to disaster-proof your live chats, and guarantee the happiness, loyalty and respect of your customers.

Disaster #1 – Agents Divorced From the Customer Experience

This experience went public very quickly. There are a few issues within this chat transcript, so let’s break it down.

Agents are being assigned chats, then rapidly leaving them, with far too much transferring occurring. The canned responses they are using for this are not useful in helping the customer understand why this is happening. When the customer gets through to someone who sounds like they could help, he is again referred to using another canned response with no explanation why.

Agents also aren’t reading chat histories. The customer is forced to repeat his question three times, all the while getting no closer to having his issue understood.

Genesys Global Survey has shown that by far the most requested customer service improvement from customers is “Better Human Service”. Your agents need to be able to recognize the impact of their actions on the customer, consider how situations feel from their perspective, and treat them like a human being, not a ticket number.

You also need to make sure that your customer is getting through to the right agent to begin with, to avoid unnecessary transferring.

This issue can be a tricky one to resolve, but here are some tips to help:

  • Make sure that your chat routing rules are set up to allow customers to be connected to the right agent the first time.
  • Invest in quality agent training. Good training doesn’t have to be lengthy or expensive, and half of the battle is training agents to recognize that they can provide a great customer experience without Average Handle Time (AHT) slipping.
  • Get your agents thinking about, and practicing empathy. As well as running training activities around this, quick tips which get them building better rapport with customers from the moment a chat starts can be useful too.
  • Make the customer experience central to your company culture. If you’re not talking about it regularly, agents aren’t going to consider it regularly either.

Disaster #2 – Chat Response Delays

This complaint was posted on a forum at 19.00 precisely. In this case, it took the customer just 17 minutes of waiting for a reply to compose a complaint which ended up online at lightning speed.

When you offer live chat, customers expect agents to respond quickly to queries. Leaving a customer waiting a long time for a chat response isn’t any different to taking a telephone call from them, putting them on hold, and leaving them there. You’re still wasting their time.

There are several things you can do to prevent this from occurring:

  • Train your agents to keep customers updated. Even if they’re looking for a solution, they should be checking in with their customer periodically to let them know that they are still there.
  • Use load balancing to route chats to agents with the best availability. You can do this based on the total number of chats being handled, or round-robin style – in the same way an ACD system on the phones should do.
  • Use maximum chat limits to stop chats stacking up when things get busy. These can be tailored per agent – so you could assign fewer concurrent chats to a new hire, versus an experienced agent who can handle more at once.
  • Implement a backup team to handle overflow chats. Your software should be able to handle the routing of requests to a team who can relieve the pressure at peak times, minimizing the impact on the customer experience.
  • Prioritize chats based on query type. A customer who selects in a pre-chat survey that they have an issue with a product, or that they wish to complain, could be given a higher priority than customers with general questions. Customers who are already frustrated are likely to become even more agitated if subjected to delays, so routing control at this early stage is vital to minimize this.

Disaster #3 – Agents Out of Their Depth

This chat transcript went viral after a customer spent nearly an hour with an agent trying to explain his problem.

During the chat, the agent referred to the customer as “Maam” even though he is a man, used very poor English, and didn’t understand what he was saying or what the best course of action would be.

It’s vitally important for agents to demonstrate to customers that they understand their issue. Where they don’t understand, they need to be able to recognize when they no longer have the ability to help, and refer for further assistance in a positive and solution-focused way.

So what can be done to prevent this?

  • Make sure it’s crystal clear what best practice in your company looks like. Use case studies and group discussion regularly to set the standard for the appropriate tone of voice, the boundaries of agent knowledge, and the situations where escalation should occur.
  • Ensure inexperienced agents aren’t left to fend for themselves. Your training program should transition new hires into competent staff, with chats monitored closely throughout this period so that any issues like this are immediately picked up and dealt with.
  • Implement a qualitative Quality Assurance (QA) program to develop agent performance. Chats like this don’t happen in isolation, and this agent will likely have had a track record of misunderstanding and miscommunication which could have been picked up through robust QA processes, even if they were missed in initial training.

Disaster #4 – Automated Message Annoyances

Here, the customer was stuck in a long queue, with canned messages adding no value during the wait time. Once it seemed that the customer was nearly at the top of the queue, system issues occurred, closing the connection and losing the customer their space in the queue.

Although you can’t help system issues which occur on the customer’s end, you can minimize the annoyance of long queues in a few ways.

  • Allow your customer to leave an offline message. Your software should allow your customer to click to leave a message if they hit a queue once entering chat. The message can be logged within the chat console or set up to email to any number of email addresses, to be responded to by a different team or at a less busy time.
  • Optimize your queue experience to include some helpful resources for your customer to look at while they are waiting.
  • Ensure your chat software comes with 100% guaranteed uptime, to remove the risk of chats being disconnected. As well as guaranteed uptime, you also want to be sure that your provider ensures the stability of their chats.

Disaster #5 – Customer Data Breaches

In this situation, a customer was the victim of identity theft, with the impersonator managing to obtain the customer’s full address, then using this to impersonate the customer with other companies. The source of the leak? A live chat interaction.

This chat transcript, which the real customer was eventually sent, shows the impersonator passing security with an address which wasn’t registered to the customer’s account (it is the address of a hotel where the customer had once registered a domain name. The only thing this has in common with the customer’s real address is the zip code.)

The impersonator then goes on to obtain the customer’s real address.

There are a number of things here that went badly wrong, and a number of things which could have prevented this.

  • If your security process involves asking the customer security questions to verify their identity, make it very clear to your agents that answers given must match exactly. Where there is any potential for confusion, make sure that procedures are issued which address these areas directly, so that security processes are as black and white as possible.
  • Advise agents to check visitor data collected by your live chat system against customer records. If data doesn’t match, for example if the location the visitor is contacting you from is different from what is on the customer’s record, advise them to run extra checks.
  • Require visitors to log in in order to chat. You know and trust the security of your own website login process, so put access to chat behind a login. Your chat software can then pull the login details through to the chat console, giving you the assurance that you are speaking to the account holder.

Do any of these disaster scenarios ring true? Or do you have any ideas on what else could be done to prevent these chat fails? Leave us a message in the comments below.

Live Chat Benchmark Report 2019

Live Chat Benchmark Report 2019

Download our annual Live Chat Benchmark Report for access to data-informed insights into live chat best practices and optimization.

Download Now
Report

About Kaye Chapman

Kaye is Comm100’s Learning & Development Manager, an internationally-experienced writer and trainer, and an MA student at University College London, the world’s #1 center for Education and Social Science. Kaye has worked with Fortune 500, governmental and private firms across the world to advance customer service operations and embed leading learning and development strategy. As a specialist in Contact Centers, Kaye is passionate about using technology and training to improve experiences for customers and employees alike. Connect with her on LinkedIn.