The metrics Customer Service Leaders should be looking at – our conversation with Ines Castro at On

5 May 2023

5 min read

Analytics on a laptop screen
Analytics on a laptop screen
Analytics on a laptop screen

When it comes to Customer Service in eCommerce, it can be hard to find the right KPI to measure. You could go for CSAT (Customer Satisfaction) but is it fair to measure agents on satisfaction when sometimes they’re unable to make customers satisfied? For example, if someone is asking to return an item outside of the returns window, they’re unlikely to be satisfied with being told they are not entitled to a refund or a return.

Instead you could go for something like Customer Effort Score (CES). This involves asking the customer how easy it was for them to have their query resolved. 

All of this is at the heart of our recent conversation with Ines Castro from our customer On. You can watch the full conversation here, but here are a few key points we took away. 

Having a lot of metrics can be overwhelming 

Ines likens metrics to health checkups. “Maybe you only do a blood test, but you could do 20 more things. You need to find a good balance.”

The point being that looking at all the possible metrics you could measure can be overwhelming and stop you focusing on the one or two that matter at that moment. It can seem like spinning plates, constantly trying to tweak to make each one improve a little a bit, rather than having a strategy to improve one or two.

For On in particular, the team can break down a lot of their metrics into vertical layers (region, country, team, and agents), and then on horizontal layers. Horizontal layers would be things like topic – for example Returns questions, Where is My Order questions – meaning that the combination could mean that you can measure how an individual agent performs responding to one topic vs another agent on that topic. 

Agent productivity is hard to measure, and it’s not about controlling the agents

At DigitalGenius we asked 200 Customer Service Leaders about the metrics they track, and agent productivity was one of them. But with so many different channels, can you really compare it. 

“When we are talking about phone calls or chats, it’s more or less easy because a phone call has a start and an end. The same with chat. But when we look at cases they handle by emails, maybe the agent had to contact the warehouse, or they had to call the carrier. So the time taken on the ticket, is not necessarily the time they were working on it,” Ines points out. 

Nevertheless, On try to boil it down based on tickets resolved in a certain time frame and can come up with a figure that works. 

Despite getting to a figure, the metric is not used as a way to control what agents do, but instead it’s to spot processes that are too manual and to see if there is a way that things can be improved. 

There are different metrics for the organisation vs. the individual

While On don’t use productivity specifically to measure individual agents, there are some metrics that do apply to individuals. Take CSAT for example: “For our happiness delivery teams they want to know what their own CSAT and how satisfied our customers are.” 

The other metrics that an individual might be measured on would include Quality Score, Phone pick up rate, and speed to answer chats. 

But for most of On’s KPIs the scope is to aggregate these metrics by Team, Region or Topic. Some are because the organisation wants to reach a target, for others it’s just to observe trends and comparisons. 

You can always add new KPIs when a challenge arises

Sometimes as part of looking at processes, you might notice that there is a black hole in what you’re measuring. Either you don’t have good data on something, the data is stored in the wrong place, or you just cannot see why a particular step is taking a long time. When that happens, you can get on top of it and add KPIs.

For example, On didn’t know how often agents were contacting the warehouse because they were doing outside the CRM. So Ines and her team moved that data within the CRM and now they can see how often the warehouse is being contacted. This became a new KPI.

When all your metrics look good, look at different ones 

Sometimes a particular metric is deemed to be important by the senior team for one reason or another, and when you start to track it you realise that improving it is not really moving the needle. 

“When all your metrics look good, and you’re not moving the business performance, it means you haven’t found the right metric yet!” Ines points out. 

One of the key things Ines points out is that the metrics are ultimately about measuring performance against a goal. When the goal changes, the metrics you focus on should change. 

Is there a #1 KPI?

Depending on your role in the business, the number one KPI may be different. If you are a team leader of a customer service team it might be the CSAT or productivity of your particular team vs others in the organisation. Or it might be the backlog of tickets and how quickly you and your team can get to each ticket. 

For a CEO, it might also be customer satisfaction because customer retention is so important to growth. But it might also be about the revenue that agents generate, because they can be some of the best sales people.

For someone like Ines, who looks at the operational side, the KPIs can change. One week it might be the automation rate – the number of tickets that are automated away from human agents. While other weeks it might be average handle time, and looking at ways that processes like talking to warehouses can be sped up. 

But if there is a #1 KPI for On, it’s Net Promoter Score. “If I have to mention one KPI that is well known across all departments it is the NPS. We monitor NPS on a weekly basis, and we break it down into specific regions. So everyone knows it.”