Maybe you’re drowning in data but can’t answer questions from your leadership team. Or maybe your teams delay giving you an answer when you ask what seems like a simple question. If either of these scenarios sounds familiar, you might need to follow some best practices for creating and using data in a modern contact center.

I recently started capturing and organizing a family history from all the materials my mom acquired over the years. I thought it would be a simple type-and-scan exercise. But it turns out, people keep changing their names, typos occur — even in newspapers — and dates are fuzzy. The contact center is no different.

The contact center is one of the great generators of data, but names keep changing, an understanding of what data means gets lost and numbers often are fuzzy. Having the basics is foundational to modernizing your contact center. But in this case, the basics aren’t necessarily simple. And artificial intelligence (AI) creates an exciting factor of and to the reporting story.

Following are the five best practices for creating and using contact center data.

  1. Completeness

The foundation of all reporting is having complete information. You can’t understand drivers or relationships without it. And managing work requires a set of basic measures that likely won’t ever change. They are the basic interaction-level metrics (Figure 1).

 

 

 

 

 

 

 

The first four columns are long-standing metrics in a well-run contact center. If you’re missing any of these, I challenge you to work on adding them to your reports. However, “Results” is an area that’s sometimes ignored or shunted off to a system that’s separate from the contact center-generated measures. To lose the relationship of results to activity is to lose the ability to focus and improve the other measures for the greatest value.

Remember: The purpose of interacting with customers is not just to talk, but to achieve some goal for them. The result measures represent the outcome of that goal — that’s where you need to maintain focus without losing sight of the core metrics. A complete view across all of these metric areas drives success.

  1. Consistency

The “Completeness” stage includes measures that we’re used to seeing. In this “Consistency” stage, these measures apply across a contact center that handles multiple types of communication media and work types. You need the same measures across all channels because what you’re really measuring is work. In fact, COPC, one of the leading sources of contact center standards, lists consistent measures across channels.

Here are some reasons why having consistency, and a shared place to store customer engagement across channels, is a good goal.

  • Consistent measures allow you to blend your agents across channels and support them with coaching from data that’s believable and understandable because it adds up.
  • First contact resolution is more accurate because it includes customers who come back on any channel, not just the one that they started with.
  • You can do some interesting things regarding channel preferences. You need to understand if you get better results on email over phone because it has a record that customers can use as confirmation. Or if your multichannel outbound campaigns do best with SMS. Also, ask yourself if certain types of customers, or particular customers, prefer certain channels.
  • The ability to attribute costs to marketing campaigns, regardless of how a customer engages, provides your marketing teams with information about what works. It also provides contact centers with data that supports cost requirements. Most companies have several phone numbers — or a different link per campaign — that they measure by. But consider if a customer contacts you differently, and the agent determines it’s in response to a marketing activity. Tying this to the results as noted in the “Completeness” section reinforces that you want consistent results measured across all channels.

Consistent measures — for activities and results — across the channels, enables you to add up factors and answer broader questions than you can if you had measured media in a silo. Keep in mind: There are some variations in channel-specific measures.

For example, abandons only apply to real-time channels; average response time between contacts only applies to messaging channels. So, be rational but avoid the mistake of creating different measures per channel.

Consistency allows agent fairness and meaningful views across your customers’ journey. It also reduces the management efforts needed to tie data together. Avoid the trap of setting up independent data silos per media channel or work type.

  1. Detail

When I introduce this concept, I asked the following question: How many Fs are in the following sentence?

Finished files are the result of years of scientific study combined with the experience of years.

Audience answers spanned three to seven. The actual answer is six. The point here is that even having data available requires looking at the details.

This area is obvious to any data geek or to those of us who are just addicted to data. Using detailed data is the core of all analysis because it allows us to find truth that’s hidden by averages. It also allows us to find relationships that we didn’t expect by using dimensions to bundle the data in different ways. Here are examples to trigger your thought processes on this topic.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

The bottom line is to have details available to use and then use details to answer “why” questions. Finally, balance your use of averages and details to effectively manage and improve your operations.

  1. Targets

Contact centers have a purpose other than just existing. Results provide the business context for contact center activities. Setting targets to achieve, or thresholds to avoid, is the method for giving good/bad context to the numbers we see. Here are considerations for reasonable targets and techniques you can use to identify them.

Study current behavior to understand customer expectations and agent capabilities.

  • Service level: Look at abandons, complaints and results to find the customer-patience level.
  • Results or handle time: Bucket agents by skill level to find reasonable targets versus using an average across many skill levels.
  • Set targets per intent: Not every type of work has the same goal.

Remember the rules of math so you set reasonable goals.

  • Schedule adherence: Set a target to avoid “noisy” exceptions. This allows your staff to be a few minutes late from a meeting without you needing to touch your workforce management tool.
  • Competing priorities: Avoid setting un-makeable goals. For example, high occupancy and high service-level adherence on low volume work almost never works out mathematically. You’ll always manage both but pick the one you’ll plan around.
  • Averages: Calculate numbers from absolute values, not from other averages. This happens more often than you’d think; make your 4th grade math teacher proud by avoiding this mistake.

Keep tuning so that the targets continue to drive the business positively.

  • Re-evaluate regularly: Catch changes in expectations and performance. Do this at least annually.
  • Multichannel expansion: As you set goals for new media channels, result targets are less likely to vary, but operational metric targets might be different.

Targets exist to drive the behavior you want. Use the data you have to set reasonable goals that keep the focus where it needs to be — and continue to review the validity and effectiveness of the targets.

  1. Artificial intelligence

Artificial intelligence (AI) is less about the characteristics of data and more about how it’s used. Because this series is about basics, I’m not going to go into a tutorial on AI. But this series is about basics as a foundation for innovation. So, let’s talk about AI a bit, and particularly about machine learning.

Machine learning is a type of data analysis technology that extracts knowledge without being explicitly programmed to do so.

Here’s how the basics on reporting and data translate when looking at advanced capabilities like AI and machine learning.

  1. When using machine learning to inform bots or predictive solutions, you’re training the system to accomplish something à the result. Therefore, feeding the learning system with complete data that ties results with activity allows the system to determine what levers and activities will best optimize the goal.
  2. Consistent, usable information about the activities in a customer journey that spans media channels provides context that informs predictions of what action a customer will take next. For example, the customer has spent 30 minutes on chats in the last two days, the prediction is that they will call tomorrow. Try to anticipate their need by reaching out proactively to reduce the customer effort.
  3. Analytics are always grounded in detail. In the case of machine learning, detail information on interactions and journeys — details that are complete and consistent — teach the learning engine what the real data relationships are and whether tuning one area (for example, selecting an agent that has a lower average handle time on billing questions) optimizes the goal (customer satisfaction). The beautiful thing with machine learning is that these details show shifts in behavior and outcome over time, for example, result improvements on one area might change focus to a new area.
  4. This is more philosophical than tactical. Like results, targets might represent what you’re training your machine learning engine to achieve. As you build and refine your AI technologies, remember the human dimension of data. Targets drive human behavior. If you change your targets, your people will follow. And that will shift what you see in the data. Continue to review your targets.

AI is becoming a legitimate technology — and it builds on the good work you’ve done to generate complete, consistent and detailed data.

Data and Visibility
A customer once told me, “If I don’t see it in a report, it didn’t happen.” Data and the visibility of it through reports is the lifeblood of a business and definitely of the contact center.

As the scope of the contact center expands to more channels and to work that isn’t front office, the volume of data increases and the temptation to silo it across an organizational or technology boundary is strong. Strive to create complete datasets that are consistent across media and work types so you can add up the details in different ways, as needed. The payoff is a well-run contact center with understandable and believable results — and the basis for any AI tool you choose.

Watch this analyst webinar to learn more about how to evaluate a cloud contact center.