Your social customer service agents are the public face of your brand, more so than in any other area of your customer service operation, because much of what they write is not only visible to the customer it is intended for, but to all of your customers. As such it is critical that you are able to properly measure the performance of your social agents, not only to ensure you are able to offer them the appropriate training and support but also to ensure you are able to properly resource your operation with the right number of agents at the right times.
In my role at Conversocial I’ve spent a lot of time talking to people about the many things you can measure your social agents on, below are the metrics that I think the best service organisations should be looking to measure and improve upon.
For many of these metrics you will want to take measurements over a reasonable period of time and develop upper and lower thresholds that you want your agents to fall within.
For me the most important measurement - are the customers being served actually leaving happy or satisfied with the service the agent is providing? You should be measuring customer satisfaction against a variety of different issues the agent has dealt with. A great agent with the right tools at their fingertips will be adept at turning unhappy customers into happy ones. The result of this is often the sum of many of the other metrics below, which if an agent performs well at then they should inevitably lead to satisfied customers.
Simply responding to customers is no longer good enough, your agents should be aiming for far more, and should be resolving customer issues in channel. As such you should be measuring whether or not your agents are actually performing at this.
Are they deflecting to other channels (phone, email, or even letter…)? Are they deflecting to other agents or teams? Is this happening where desired by business process, or to frequently? Ultimately the customer has come to you on social media expecting service, nothing irks them more than being deflected straight into another channel unnecessarily!
You should be regularly reviewing the content of a percentage of your agents messages and ensuring you are analysing all of the following:
Tone of voice - does the way they write come across the way you want your brand represented, do they convey some of their own personality whilst sticking within acceptable boundaries?
Grammar, spelling, mistakes/typos - are they taking the time to construct well written replies, or are they rushing and making mistakes?
Empathy - do they display a good sense of empathy with the customers, they might not always be able to provide the solution the customer wants but can still show empathy with them.
Repetition - one of the toughest parts of being a social agent is trying to avoid repetition of the same words and phrases across long periods of time, measuring the spread of words used by an agent can help to spot repetition and ensure you keep your operation sounding fresh.
Good use of social platforms - are they making the most of the social platforms they are writing responses for, are they using the full functionality available in the platform and getting the best possible responses out.
The percentage of messages the agent is responding to compared to the number they are deeming as not needing a response. The percentage levels will depend on your brand objectives on social and if you want your agents to engage purely in providing service or if you want them to build brand engagement and get involved in conversational type messages as well. The types of messages you want them responding to may also change depending on the volume of service issues and whether you are meeting your targeted SLA as responding to service queries should always be your primary goal.
It’s important to look at the types of messages an agent is dealing with, some issues will be far more complex to resolve than others, so you can’t just judge an agent based purely on volume. Understanding the volume and types of messages an agent is dealing with will give you a reasonable measure of efficiency for that agent.
Average Responses Per Hour
How many responses is an agent making per hour? This is one to be very careful with, set this to high and quality will drop, to low and agents will waste time. It is also critical to tie this back to the types of issues an agent is dealing with.
Average Handling Time
How long on average does an agent spend dealing with each conversation (analysing the messages, marking sentiment, categorising, researching the customer, composing a response, resolving the issue, etc…)? Again this will change depending on the types of issue an agent is dealing with.
Average Response Time
How long does this agent take to get back to the customer on average? Does this fall within your targeted SLA. You should use this in conjunction with other items such as types of issue being dealt with and average handling times to see if there is an issue.
Total Handling Time
Look at the total handling time for the agent, if this is a long way from the amount of time they spend on shift then there may be a problem, what else are they doing if not responding to customers? Team leaders or somebody who is present on a day to day basis should know if they are being asked to perform other activities, e.g. covering other channels during peak hours, attending training, etc...
One of the hardest to measure, but also one of the most important. For me this is arguably a step beyond customer satisfaction, the point where a customer is so happy with your service that they actually start to advocate your brand and recommend or defend you publicly on social channels.
Using the above metrics you should be able to build profiles of what good and bad looks like for your agents. This will enable you to see where agents are excelling and where they are struggling. Where they are excelling you can praise them accordingly and where they are struggling you can design training and exercises that gives them the opportunity to improve. It is also critically important to regularly review and update what good and bad looks like.
So that’s my opinion, what’s yours? I’d love to hear your thoughts on these metrics and how you go about measuring them? Is there anything you think I’ve missed that you are using to measure your agents?