Aligning the Customer Experience in an Outsourced Call Center Environment

Outsourced call center vendors normally see a disconnect between the service that they provide and the expectations of end customers who receive the service. This often happens because of the limited volume of end customer feedback and the gaps in measurement of customer feedback. The article provides some approaches and examples to overcome these difficulties.

Outsourced Call Center Environment

In a connected and competitive world, it is common to have companies outsource their call center customer service work to vendors across time zones, geographies and cultures. A typical outsourced call center vendor may address thousands of customers every day. Inherent to this choice are variations around the customer experience. Here are some issues and possible solutions to address them.

Issues with Customer Experience Measurement

The customer experience is normally measured on a daily basis using quality assessment forms. Quality experts select a reasonable sample of customer interactions and provide an expert judgment of the same. However, the frequency of sampling, the sample size, measurement errors and relevance of quality parameters are points of constant discussion.

The sheer effort in sampling introduces errors in judgment and incomplete coverage of assessment. It is normal to observe measurement errors of about 20 percent. Clients also find it difficult to maintain a sufficient number of experts on their side to help vendors in calibration.

Some studies indicate that individual parameters being measured on quality assessment forms tend to be generic in nature and may not always capture the nuances of a particular type of customer interaction. This may be because of the analytical effort required to decide what to measure and what not to.

The customer experience is also measured through customer surveys that are administered at a much lesser frequency than daily quality assessments, quite likely due to the costs involved. These surveys are designed to capture the overall customer experience. While they are generic in nature, they also tend to capture important perceptions around wait time before being serviced, time to service that transaction, number of interactions before issues were resolved and overall satisfaction levels.

[eventpdf]

Analysis of Customer Experience Measurement

Analysis performed by the author and his team indicate that there is little correlation between the parameters being measured on customer surveys and those being measured on quality assessment forms. The following table illustrates this point in one such study. Interestingly, knowledge level as measured on the quality form tends to have a better correlation with customer survey parameters in the same example. And this was seen as a consistent trend in many such studies. (Click on diagram to enlarge.)


On the other hand, in other such studies, it was found that customer experience measured through surveys had a good correlation with queue and service time parameters like wait time before being serviced, first call resolution, times contacted and resolution time. This is illustrated in the below figure. (Click on diagram to enlarge.)


The consistent trend was that such queue parameters received lesser attention than the quality form parameters. The tendency is to assume that parameters around communication are more important than these queue parameters.

Suggested Actions for Better Alignment of the Customer Experience

Automation of transaction assessments is the key to reduction of measurement errors. As an example, recent advances in voice analytics have made it possible to scan a large number of voice conversations and provide fast feedback. The volume of data generated through these mass measurements has the potential to identify hidden trends and gives better confidence in arriving at root causes.

Automation does, however, have its own difficulties. As most customer interactions are context dependent, it becomes difficult to define standard conditions for a good customer experience. This allowable variation in live customer interactions tends to create errors in measurement automation.

One can also record and keep control samples of good or defective customer interactions that can help in better calibration of manual assessors. Apart from reducing the inconsistency in manual assessments, this can also reduce the over-dependence on experts for calibrations.

Maintaining a large central repository of data with frequent analysis is another approach that seems to yield positive results. This also provides the opportunity not to be biased by historical trends and anecdotal knowledge. One can then afford to have a zero-based approach towards established correlations, and thus better respond to changing customer requirements.

These solutions normally demand upfront investments. It may finally come down to the risk appetite of an outsourced call center vendor or a client to implement these suggestions.