A recent study on call center statistics found that 91% of consumers reported poor customer service in 2021. Providing high-quality service is essential, especially today, to retain customers and drive more business. 

Quality service is only one important metric in running a profitable call center. No matter your goal, the first step is understanding what's going on in your call center. 

Since call centers produce so much data every day, it becomes difficult to easily and quickly analyze and take action on important data points.

We’ve partnered with Bill Inmon, best known as the “Father of Data Warehousing,” who guest-wrote this blog post detailing what drill-down processing is and how it can help you better understand your call center's performance.

thumbnail image


Drill-down processing is a way to analyze and better understand the data created. The first step to understanding your call center's performance is to capture conversations in a recorded form - usually a .wav format. Then the .wav records are converted into a text format, such as a .txt format. After the conversion into a text format is completed, the records can be analyzed using the drill-down method.

Using the drill-down process in your call center will allow you to analyze your call center data in an easy-to-consume, exploratory manner.



In other words, when you examine your call center conversations at first, it’ll seem that you’re dealing with an extensive, unstructured collection of data. Making sense of these call center conversations will require you to explore the data quickly. Drill-down processing allows you to quickly “dabble” with the data so you can find out what is there.

Said another way, when you are presented with call center output, you don’t know what you don’t know. You need to quickly and easily look at the call center data to find out what is there.


Drill-down processing for analyzing call center data is essential because call center conversations are typically structured much differently than written text. People do not talk like they write, meaning that call center conversations are much more casual and informal than written text. Because of this, the informal language style needs to be investigated and analyzed differently than written text.

Call center processing is exploratory processing, where you don’t know what you will find. 

One of the best ways to analyze the data is by using a tool that allows you to easily extract, transform, and load (ETL) your call center data to analyze it in a structured manner. Integrate.io provides a new ETL platform (with reverse ETL capability), so you can easily get your data to where it needs to be. Try it yourself today.


Here’s what the process of creating call center records looks like. 


  1. The call center conversation occurs. 

  2. The conversation is recorded electronically, such as on a .wav file. 

  3. The .wav file is then converted into an electronic text format, such as into a .txt file format. 

  4. The text file is then read and analyzed by textual ETL (like Integrate.io) into a standard database format. Much editing, classification, and other activities occur in textual ETL. The output of textual ETL is a standard database format. There is both text and context that appear in the database

  5. Once the database file is produced, drill-down processing is done. After drill-down processing is done,  then visualization can be done.



As you can see from the above processing flow, drill-down processing is an exploratory process. Nearly everything that happens in drill-down processing is on a one-time-only basis. Once the analyst has looked at the call center data through drill-down processing, it is not necessary or even desirable to look at the data the same way again. 

If you need to review already analyzed data, then data can be incorporated into some form of visualization. 

Visualization processing – unlike drill-down processing – is done on a repetitive scheduled basis.


So, what sort of things does the analyst look for in drill-down processing?

Three of the most common data points that are analyzed are:

  • Have the call center exchanges been conducted in a polite and/or impolite manner?

  • Has anyone mentioned a lawyer, an attorney, or a lawsuit?

  • “Red flag” items which could cover various topics like emergencies, injuries, harm, etc.

Additionally, analysts usually will want to gain an understanding of the regular topics that have been discussed. What does a “normal” discussion look like? Are there conversations that can lead to future opportunities? Is the customer asking for a new service or product?



A great starting point for drill-down processing is to look at the taxonomies used to classify the data. You can see the elements of the taxonomies that have been used to classify the data below.


Step 1: The taxonomy relating to politeness is selected because you want to review and surface instances where politeness is used within your call center conversations.  


Step 2: After selecting your taxonomy, all the documents that have an element of politeness detected will surface. 


Step 3: After reviewing which documents have surfaced, you will want to select the document you want to review.


Step 4: The selected document is displayed. The words that have indicated politeness are highlighted. The words “thanks” and “thank you” are shown in this case. Whoever is analyzing the data can thoroughly read the entire document for additional context.

This is one example of drill-down processing for call center conversations.

Let’s say our analyst would like to analyze other types of taxonomies. For this next example, the analyst would like to look at lawsuits since businesses want to know if a potential lawsuit is arising. 

Step 1: The analyst selects the taxonomy containing “lawsuit.”


Step 2: The documents where a lawsuit is mentioned are located and displayed for selection. 


Step 3: The analyst selects the document they would like to review. 

Step 4: The document is then opened for review. In this case, the conversation mentions the word “lawyer.” By reading the entire document, the analyst can determine whether the mention of a lawyer is a casual comment or whether the caller is threatening to contact a lawyer to initiate a potential lawsuit.

The sooner the corporation can detect a potential lawsuit, the better the chances are the lawsuit can be settled outside the courtroom. 

Our analyst will review “red flag” items within the call center conversations for the following example. "Red Flag" items are those that could cause additional concern for the business.

The same four-step process is followed to surface documents containing “red flag” items.


Here are some of the many topics found in “red flag” items. Once again, our analyst will select the topic in question, surface the recommended documents, and review its contents.



In this case, the program has detected the words “sick” and “difficult.” The analyst can now read the full document to determine what is happening.


On occasion, it is desirable to examine a document that contains words that do not appear in a taxonomy. The analyst can create a custom list of words to examine records containing the custom words. In the example shown, the analyst is looking at the words – “claim”, “email”, “essence”, “injury”, and “payment”.


A search is done on documents containing the words that have been specified. The system will then surface the relevant documents.



Once the analyst has examined the call center conversations through drill-down processing, the analyst is then prepared to formalize the analysis through a visualization.


As you can see, analyzing your call center data with drill-down processing can significantly impact your business. With the large amounts of data created, you’ll need an easy and quick way to extract, load, and transform the data to be analyzed.

Integrate.io’s powerful on-platform transformation tools allow its customers to transform, normalize and clean their data while also adhering to compliance best practices. Schedule a demo today to try it yourself.

This article was written by Bill Inmon, the “father of the data warehouse,” who has authored 65 books and was named by Computerworld as one of the ten most influential people in the history of computing. Bill’s company – Forest Rim technology, is a Castle Rock, Colorado company. Bill Inmon and Forest Rim Technology provide a service to companies in helping companies hear the voice of their customer. See more at www.forestrimtech.com.