Most call centres collect the same metrics, which is necessary because everyone wants to serve customers faster and more effectively. But the impact of average handle time (AHT), for example, on your overall business goals isn’t always clear. You need to consider this when making staffing decisions, revenue changes and other assessments. Once you better understand what the data is telling you, you can apply those insights and have a greater impact on your business.

To do that, let’s look at forecasting.

Forecasts Impact Almost Everything

Even with a variety of algorithms to deliver the best and most accurate forecast, situations can change in an instant. That perfect schedule you created for the current week comes crashing down when a minor weather event turns major; a power outage hits; or a myriad of business situations derail your plans. When this happens, agents overwork or you can’t respond to customers quickly enough. Calls might get dropped. Interactions aren’t handled quickly enough — and customers wait longer. These all negatively affect customer experience and quite possibly your bottom line. And longer wait times affect more than customers.

Managers and agents become stressed. Stress increases attrition rates – creating a costly charge to the business. One change in forecasting variables can not only render your forecast inaccurate, but it has the potential to cascade through your business.

This scenario doesn’t have to repeat on a loop. You can learn from it by comparing your proposed forecast and schedule against what actually happened — and why agents and managers handled situations in certain ways. To do this, use quality management, evaluations, analytics data and insights. Together, these inform your corrective actions, reveal areas of improvement and generate strategies to adapt when new scenarios arise.

Forecasts Are Always Dynamic

Even without a crisis, quality managers should provide feedback to improve forecasts by digging into metrics. AHT is a good place to start. If customer interaction calls are too long, you need to find out why. First, review the interaction transcript.

The agent might be struggling to answer the customer’s question. Answering unexpected questions and delivering empathetic service are even harder on new agents. Suppose you have inadequate staffing in your consumer financing division, so your agent Andres has a call routed to him. The customer asked about new qualification requirements for leasing. But Andres wasn’t trained on these. So, he searched multiple applications — Oracle, the web, a CRM system — and couldn’t find the answer. It took him more time than average, and he still had to hand off to another agent.

When a quality manager reviews the call, she sees that Andres didn’t have the proper training. This is the initial reason why first-call resolution (FCR) SLAs weren’t met. Improved forecasting and better routing would prevent newer agents from receiving calls they aren’t trained for.

Another interaction might be shorter than expected. Initially, that seems positive as it’s adhering to the expected time. But is the agent providing the empathetic service you want to deliver? Look at the attitude the customer expresses as well as the overall sentiment of the interaction. This can lead to a collision of different metrics evaluated as AHT, FCR, Customer Satisfaction Score (CSAT) and Net Promote Scores (NPS).

Get to the Source of Good and Bad Results

Ideally, when Andres received that call, he’d know where the customer is calling from, and how to access the information needed based on the topic — to fulfill expectations. However, sentiment analysis reveals even more.

Let’s say you set up criteria for positive sentiment, with 25 and above being a good indicator of positive sentiment. If your analytics indicate the overall sentiment in an interaction is below 25, it can trigger an evaluation. On the opposite end, if 100% of sentiment is positive, you’ll want to review that one to learn from it and replicate whatever led to such a positive customer experience.

And you can go even deeper for insights. For example, one of Andres’ recent interactions shows a sentiment score of 65. But it started at a very low 5. What did Andres do to go from 5 to 65? This is an important interaction to understand so others can emulate how he turned around a tough call. Sometimes your metrics are buried under a lot of other valuable data.

Using Filters for Better Interaction Sampling

Quality managers used to review random sample interactions, perhaps one from each agent. But random sampling doesn’t give you a true picture of interactions. Whether you have 200 or 2 million, it isn’t possible to review a representative sample using that method.

Most quality assurance managers focus their process on specific lines of business, such as leasing, insurance and so on, as well as interaction criteria like duration, channel or wrap-up codes. For example, you might segment for AHT and review calls that are under 2 minutes or over 15 minutes.

Many contact centres use speech and text analytics topic filters as well. When a specific topic is mentioned during an interaction, it can trigger an evaluation and an interaction review. This could include a customer who asks to speak to a supervisor or one who mentions a competitor’s name.

Reviewing interactions by topic can improve your understanding; you can evaluate the differences between the wrap-up code, queues and topics discussed during the interaction. When a simple request on one topic is resolved, customers extend the conversation with another request. Knowing why the call was extended and the different topics addressed reveal connections between different customer inquiries. And this allows you to make smarter business decisions before acting.

Building a quality monitoring process into your organisation requires more than tracking expected metrics. You need to understand exactly what happened during a wide range of interactions to help employees work more efficiently.

Metrics and data tell you a lot. But you’ll find deeper insights when you get to the root cause of inefficiencies and likewise, reasons certain things work well. Put the human touch on data to improve forecasting accuracy, employee engagement and your customer experience.