Understanding Correlation in the Contact Center Industry (Part 1)
Understanding correlation is crucial for making informed and effective strategy decisions.
It allows leaders to foresee how their decisions ripple across the business, influencing areas such as costs, efficiency, quality, and Customer Satisfaction.
In this article, I’ll explore correlation through the lens of the Contact Center industry, diving into practical scenarios that highlight its importance.
Many people find themselves working in the Customer domain – whether in Contact Centers, Customer Service, or Customer Experienc – without formal training in business strategy.
That’s why I wrote this article: to bridge that gap and help these professionals build their strategy expertise.
This post is Part 1 of a 2-part series on correlation – and Part 2 will be out soon.
Here is the table of contents for this post
Your decisions impact Business Outcomes
Let’s define the term Correlation
The 5 Question Approach to Correlation
Understanding Correlation Scenario #1: Two variables are positively correlated
Let’s look at a classic example of positive correlation in the Contact Center industry
Let’s look at another example of positive correlation in the Contact Center industry
Caution: Some correlations are scientific while others are not
Understanding Correlation Scenario #2: Two variables are negatively correlated
Let’s look at a classic example of negative correlation in the Contact Center industry
Here’s another example of negative correlation in the Contat Center industry
What it means when people talk about balance
Time horizons matter
What to expect in our Part 2 article on correlation
1. Your decisions impact Business Outcomes
When you’re a business leader at any level, you’re expected to accurately understand how the the decions you make and the actions you take impact business outcomes.
Some people call this seeing the bigger picture.
But no matter what you call it, when you connect your decisions and actions to how those impact desired business outcomes – you’re taking about an important aspect of business strategy.
One where we understand our Strategic Objectives – and how what we do relates to achieving those Strategic Objectives.
2. Let’s define the term Correlation
Before I go into the different business scenarios, let’s explain the term correlation.
Here’s a helpful definition –
Correlation shows how changes in one variable are linked to changes in another variable.
Let’s introduce the different correlation scenarios here – how one variable moves based on changes in another variable –
If when it rains, more umbrellas are sold, there’s a positive correlation between rain and umbrella sales (the variables of rain and umbrellas sold move in the same direction)
If the amount of exercise we do goes up and our body weight goes down, there’s a negative correlation between exercise and body weight (the variables of body weight and amount of exercise move in opposite)
The # of books I read in a year has no linkage to the # of ice cream cones I eat in a year – so there is no correlation between the # of books read and the # of ice cream cones consumed
We need to add on a bit more to our definition of correlation.
Correlation also expresses the degreeor strength of the linkage between the two variables –
Correlation also indicates the degree to which changes in one variable predict changes in another.
Where the ‘strength’ of the linkage is expressed as a coefficient denoted by r – which ranges from -1 (for perfect negative correlation) to 1 (perfect positive correlation)
Here’s how to interpret the coefficient number:
r = 1 Perfect positive correlation (as one variable increases, the other increases proportionally)
r = −1 Perfect negative correlation (as one variable increases, the other decreases proportionally)
r = 0 No correlation (no linear relationship between the variables)
So if someone tells you that the coefficient (or strength of the linkage) between the variables of Rain and Umbrellas Sold is .95 – there is a very strong positive correlation between these variables.
If someone tells you that the coefficient between the # of Books they read and the # of Ice Cream Cones they consumed last year was .03 that means there is little to no correlation between these variables.
We can look for correlation across any two variables
We can, in theory, look for a correlation – and the strength of that correlation – across any two variables.
Here are some examples of variables you might look at in the Contact Center industry:
Customer Satisfaction (usually people want to see this go up for example)
Sales
Costs
Quality
Customer Wait Time
Complaints (usually people want to see this go down for example)
Agent Retention
The list of variables that we can measure goes on and on
Because of interrelationships – the way variables relate to each other in any ecosystem – the decisions you make and the actions you take don’t operate in a vacuum.
When you work to intentionally drive the outcome of one variable – such as increasing Customer Satisfaction – there are likely to be impacts on other variables as well.
The key for Leaders is to understand and – as accurately as possible – predict the expected correlation and outcomes of key variables.
Before we dive into the topic of correlation it’s important to say that beyond correlation, variables can exhibit various types of relationships.
For this 2-part article, I am focused on the topic of correlation.
Though I may expand the lens to include other types of variable relationships in future articles.
Correlation plays a foundational role in data-driven decision-making, which is critical for modern business strategy.
While correlation itself doesn’t prove causation, it helps identify relationships and patterns that can guide further investigation and inform strategic decisions.
3. The 5 Question Approach to Understanding Correlation
Here is a 5 Question Approach to evaluating correlation assumptions and outcomes – before you make decisions or take actions.
The goal is to be sure that my assuptions have been thought through and that I’ve accurately – to the degree possible – predicted the outcomes of the decision(s) I’m about to make.
The 5 Question Approach to Understanding Correlation
For these 5 Questions I either write down, articulate out loud or work with a group (or better yet all three) to fill in the answers.
Our goal is to ______ (raise Customer Satisfaction, improve Agent Quality, reduce Customer Wait Times, etc.)
The two key variables involved are X (name the first variable) and Y (name the second variable)
I expect X and Y to correlate _____ (positively, negatively, little or not at all)
The reason I believe that X and Y will correlate this way is because _____ (explain your reasoning including any assumptions you are making to support your prediction)
To ensure that X and Y correlate the way I predict I need to additionally do _____(implement the right processes, set the right parameters or conditions, design appropriate training, etc.)
I’ll show examples of The 5 Question Approach at work in the next section.
4. Understanding Correlation Scenario #1: Two variables are positively correlated
Here’s what it means when you say that two variables are positively correlated.
When one variable increases, the other variable also increases (or is likely to increase).
Conversely, when one variable decreases, the other variable also decreases (or is likely to decrease).
With a positive correlation, the variables move in the same direction.
We saw this earlier with the rain and umbrella example.
If when it rains, more umbrellas are sold, then there’s a positive correlation between rain and umbrella sales (the variables of rain and umbrellas sold move in the same direction)
The positive correlation relationship is displayed on the graph.
By the way, the word ‘perfect’ in the graph – Perfect Positive Correlation – indicates the exceptional strength of the linkage between the two variables, That’s why you see a straight line depicted.
You can still have a ‘less strong’ positive correlation between two variables – one which is not a Perfect Positive Correlation.
5. Let’s look at a classic example of positive correlation in the Contact Center industry
I’ll use The 5 Question Approach I shared earlier.
1. Our goal: To increase Customer Satisfaction
2. The two variables I am looking at are:
Customer Satisfaction: The degree to which the Customer is satisfied
First Contact Resolution (FCR): Fully resolving the Customer’s enquiry the first time they reach out
3. I expect a positive correlation between these two variables –
An increase in the rate of FCR
Results in an increase in Customer Satisfaction
Conversly –
A decrease in the rate of FCR
Results in a decrease in Customer Satisfaction
4. The reason I believe this correlation will happen
First Contact Resolution (FCR) is a well understood and researched metric in the Customer Service industry.
There is a proven positive correlation linkage between FCR and the resulting level of Customer Satisfaction.
(My intention here is not to teach FCR so I’ve kept this explanation very basic. In real-life you’ll probably need to present more data and findings about how FCR is positively correlated to Customer Satisfaction).
5. The additional work I need to do to ensure this outcome is:
I’ll need prioritize our Contact Types and work through a deep dive study of what is important for FCR for each individual Contact Type and intentionally design those Customer journeys.
(Again my intention here is not to teach FCR so I’ve kept this explanation pretty basic. In real-life you’ll want to layout the specific ‘additional’ work necessary to achieve the positive correlation that you seek.)
6. Let’s look at another example of positive correlation in the Contact Center industry
1. Our goal: To increase Agent Quality in the Contact Center
2. The two variables I am looking at are:
Agent Quality: The Agent’s measured achievement against pre-defined Quality standards as conducted by an internal Quality specialist or involving the use of automated tools.
# of Agent Coaching Sessions: The number of times measured across a month, that an Agent is coached on a Customer interaction (call, email, chat, face to face).
3. I expect a positive correlation between these two variables –
An increase in the # of Agent Coaching Sessions
Results in an increase in Agent Quality
Conversly –
A decrease in the # of Agent Coaching Sessions
Results in a decrease in Agent Quality
4. The reason I believe this correlation will happen
In this section you’ll probably share findings and case studies on the significant impact of Line Managers on Employee Engagement and improvements in Quality performance.
5. The additional work I need to do to ensure this outcome is:
In this section you’ll likely talk about the need to validate what your Center defines as ‘Quality’ – to be sure that it is clear and achievable.
And you’ll likely conduct a complete review of your entire Quality Assurance process to ensure that the conversation that takes place between the Line Manager and the Agent is effective – in all ways – in improving Engagement and Quality.
7. Caution: Some correlations are scientific while others are not
If we’re talking about a science – such as physics, mathematics or chemistry – there are certain ‘guaranteed’ positive correlations.
For example: Asthe temperature of water increases, the amount of sugar that can dissolve also increases.
This is a scientific correlation – it’s mathematical in nature. It’s always going to ‘happen’.
But with our Agent coaching example, we’re examining a human process in the workplace.
It’s much less scientific.
Which means that if we’re expecting to achieve a positive correlation between the # of Agent Coaching Sessions and Agent Quality, we are going to have to put thought and effort into our planning and execution to ensure that outcome.
Let me give an example of why more Coaching Sessions might not result in improved Agent Quality
If our Agent coaching sessions are poorly delivered, simply increasing the # of Agent Coaching sessions might not increase Agent Quality.
It could even have the opposite effect and actually decreaseAgent Quality.
Because the Agent might react negatively to the poor coaching conversations provided by their Team Leaders.
Agents sometimes complain that their Team Leaders or Quality Assurance people don’t add much value to their communication skills.
They may overly focus on nitty gritty behaviors, spend a lot of time criticizing or use their personal opinions to judge the Agent’s Quality performance.
None of which helps or inspires Agents to do better or be better.
Sometimes well meaning Leaders tell me that “based on their experience” there is no correlation between the # of Agent coaching sessions and Quality
When I hear someone say “Based on my experience” when describing a correlation outcome, I always listen carefully.
Because it might be a case where the correlation could have worked out the way that was intended.
But due to flawed planning or execution – such as the delivery of poor coaching conversations with Agents – the expected outcome was not achieved.
Learning point: Know which kind of correlation you’re looking at – scientific or not – and proceed accordingly. Non-scientific correlations require a level of thought and effort to achieve. Don’t skip or skim over that reality.
8. Understanding Correlation Scenario #2: Two variables are negatively correlated
Here’s what it means when you say that two variables are negatively correlated:
When one variable increases, the other variable decreases (or is likely to decrease).
Conversely, when one variable decreases, the other variable increases (or is likely to increase).
With a negative correlation, as one variable improves, the other variable tends to decline.
We saw this earlier with the body weight and amount of exercise example.
If as the amount of exercise we do goes up, our body weight goes down, there’s a negative correlation between the amount of exercise we do and our body weight (the variables of Amount of Exercise and Body Weight move in opposite)
9. Let’s look at an example of negative correlation in the Contact Center industry
When you increase the # of Agents in capacity (meaning they log in)
Customer wait times go down
Conversly –
When you reduce the # of Agents in capacity (meaning they log out)
Customer wait times go up
This negative correlation is mathematical – there’s no point to debate if it is right or wong. It will always happen.
10. Here’s another example of negative correlation in the Contact Center industry
When you raise your Service Level objective (and meet it)
Your Agent Occupancy (or how ‘busy’ Agents are) will go down
Conversly –
When you lower your Service Level objective (and meet it)
Your Agent Occupancy (or how ‘busy’ they are) will go up
This negative correlation is mathematical – there’s no point to debate if it is right or wrong. It will always happen.
You can’t answer Customers quickly and keep your Agents ‘busy’ (as defined by Occupancy) at the same time.
But now and then you will still hear someone say this –
“Can’t we just get the Agents to work harder? So that they can answer more Customers quickly and keep busy at the same time?”
Typically, someone who doesn’t understand Contact Center operations says this.
Or even if they do know their Operations they choose not to accept the mathematical reality (for whatever reasons I can’t guess).
It defies common sense to try and argue against a mathematical reality.
That’s like arguing that you don’t like gravity – or the fact that the sun rises in the East and sets in the West.
11. What it means when people talk about balance
It’s when people are talking about negative correlations that you’ll sometimes hear them use the word balance.
Because some business scenarios inherently involve trade-offs.
For example, inbound Contact Centers around the world ask this question –
“How can we balance between answering incoming chats or calls quickly enough to keep Customers satisfied while also keeping our labor costs in line?”
The word balance implies that this is a trade-off scenario.
There is a strong negative correlation between Speed of Answer and # of Agents Required – and thus labor costs.
The faster you choose to respond to Customers the more Customer Service Agents.
Knowing and accepting that you are facing a trade-off scenario is the first step.
From there you and the Team can sit down and decide how you’re going to identify the balance.
Let me give a scenario that shows you why time horizons matter.
You will definitely hear Contact Center Leaders say something like this –
“We can reduce our Customer Service staff and resulting labor costs after we implement our self-serve options.”
And that may well be true.
If the self-serve options work well, are able to handle a certain volume and Customers embrace them – then you could and should recalculate labor requirements.
But solutions like rolling out self service options or fixing Customer problems to reduce incoming demand – take more than a half hour, day or even a week to test and implement.
They’re mid to long(er) term solutions.
They may only be ready three, six or even 12 months into the future.
But when you’re staffing your Contact Center across half hour intervals – you’re looking at a short-term time horizon.
If one person in the room argues that labor costs can be reduced, they may be referring to the mid to longer-term time horizon.
And they could be right.
While another person in the room could be arguing that we have to have the right # of people in place from 9:00 – 9:30AM on Thursday morning to achieve Service Level and thus we can’t cut labor costs for that interval.
And they would be right too.
As you can see, the confusion is caused when different people are using different time horizons in their thinking.
Learning point: Before we debate and argue about the correlation outcomes we expect to happen, ensure that everyone aligns on the time horizon being discussed.
And I can’t close this topic without adding this caution
It’s clearly a bad decision if to succeed in the short term you end up damaging mid to long term outcomes.
If you cut your Customer Service costs now – in the short term – and have too few Agents in the right place at the right time- in the mid to long run you’re going to damage everything that matters.
Which means potential damage to the variables of Quality, Engagement, Customer Satisfaction and Agent Retention.
Success in the short term that ends up damaging outcomes in the mid to long term isn’t success at all.
13. In our Part 2 article on Understanding Correlation
In our Part 2 article, we will take a look at a common Contact Center scenario that confuses people.
Where they believe there is an inherent trade-off – or negative correlation – when in fact there isn’t.
Which scenario?
It’s the artificial balancing act that people believe exists between Agent Productivity and Agent Quality.
So we’ll look at that scenario and share more thoughts on correlation.