Monetizing Online Customer Care Feature

About two and half years ago, we were working on a project for a major telecommunications company with the sole purpose of shifting calls out of the call centers, and creating a self-support experience online that would resolve customers issues. This is the sort of industry created by the fourth industrial revolution. You can learn more about that right here. Of course, moving the service online did pose some problems, like what if the servers went down? Perhaps there were some frayed internet connections? However, although technology doesn’t run smoothly all of the time, for the majority of the time it should. Plus, in today’s digital age, many customers are already online and it’s much easier for them to chat with someone virtually rather than over the phone. If there are any issues, there are some denver managed it services in place to help get the site up and running as soon as possible.

One of the most interesting and challenging aspects of the project was coming up with a model that would allow us to quantify the financial benefit that online support has for the business. Having this data, and being able to project cost-savings for the business, allows companies to make business cases to create a better online support experience, and reduce call center costs, many find that could do this by using customer support software instead of call centers.

Here’s a step-by-step guide to creating your own self-support customer care monetization model.

What data do you need to track self-support customer care?

In order to track the benefit of online self-support customer care, you will need the following:

  • Track the numbers of visitors to your online support pages
  • Capture visitors who have a unique identifier that tracks their behavior across the business
  • Ask visitors via online voice of the customer surveys: “If you could not resolve your issue in your visit, where are you most likely to go for assistance?”
  • Gather call center data that gets down to call level information, you can do this by using customer data platforms or anything else that similarly collects customers details.
  • Create a cost per call value

Defining the Benefit Model

Daily Visitors to Online Support Content, multiplied

  • Propensity to Call Support rate (established via monthly survey results), used to calculate
  • Potential Care Call Volume, reduced by
  • NCP48 (Next Contact Prevention within 48hrs), provides
  • Deflected Call Totals, multiplied by
  • Cost Per Support Call, gives
  • Benefit $ Value of Online Support Content

Example Benefit Model:

Daily Visitors to Online Support Content1,000
Propensity to Call Customer Care (From Survey Results)30%
Potential Care Volume300
NCP48 Rate75%
Deflected Call Totals225
Cost per Call$5
$ Benefit$1,125

Sourcing the Data

Either Adobe Analytics or Google Analytics would work to track these variables. The important part is to implement a way to capture a unique visitor ID that can track the visitor’s behavior across the business. In our project we chose to use subscriber ID to track the visitor’s behavior across the business. If the visitor authenticated on the site, the ID would be captured and stored in a persistent cookie. If a customer called customer support, we would also capture the subscriber ID either from the IVR (Interactive Voice Response) or when an agent would look up the customer’s account, thus allowing us to track the customer’s interactions across the business.

The online voice of the customer survey is important because it addresses where the customer is most likely to go next for support if they could not resolve their issue online. It is not realistic, or accurate, to assume that most customers would call customer support if they couldn’t resolve their issues online. The data from the survey results allowed us to create a propensity-to-call rate, which was used to identify the percent of visitors that would call customer support as a next step in resolving their issues.

The most critical calculation in the model is the NCP48 rate (Next Contact Prevention within 48hrs) that provides us with the percentage of authenticated visitors that did not call customer support within 48 hours of visiting our online support content. This success measure allows us to understand how well our content helped visitors resolve their issue, or identifies opportunities to improve content. Within the model we will use the rate to reduce the Potential Care call volume (Daily Visitors to Online Support Content * Propensity-to-Call rate) to the total calls deflected.

Once we have identified how many calls we have deflected from calling into our call centers, we can multiply our results by our cost per call to understand how much money we have saved the business. Depending on how the call centers are structured, the cost per call may be a blended rate to account for differences in tier level support and cost per site.

Here is another look at how the model might look:

Daily Visitors to Online Support Content1,000
Propensity to Call Customer Care (From Survey Results)30%
Potential Care Volume300
NCP48 Rate75%
Deflected Call Totals225
Cost per Call$5
$ Benefit$1,125

Customizing the Model

Not all websites, call centers, and customer behavior are the same, therefore it is important to keep in the mind the following:

  • Some visitors may call support and never reach an agent due to IVR assistance, long wait times, or other reasons, so it is best to look at only calls that were handled by agents to get a more accurate NCP48 score.
  • Depending on your support section of the website, not all visitors may be current customers, and therefore should not be counted in your daily traffic to support content. If your support content is not restricted to customers only, you can apply an authentication rate to the daily traffic amounts to help reduce the traffic to current customers based on the how many visitors on your site have a customer ID at a given time.
  • In addition to measuring the benefit of reducing calls to customer support, the model can be adjusted to measure chat and email deflections as well. Just switch out the propensity to call with propensity to chat or email, and input the cost to chat or email.

If you have any thoughts or questions about the model, please reach out to us. We would be happy to assist.

Written By


This post is curated content from the Evolytics staff, bringing you the most interesting news in data and analysis from around the web. The Evolytics staff has proven experience and expertise in analytics strategy, tagging implementation, data engineering, and data visualization.