Which quality indicator do I choose now?



Few activities today are more closely monitored than the operation of a contact center. The number of KPIs used is enormous, if not infinite. Every company has its own As well as every type of operation does. Some are very relevant, others less so. Some are really reliable, others need to be interpreted by those in charge.

We can highlight some of the most common ones:

Service levels


Customer Satisfaction Indicator


Response Time

And many more, of course. But this article is not about the KPIs used in contact centers but about the indicators for measuring customer satisfaction. We will start with the well-known and widely used NPS.

The NPS is the result of a simple calculation of the subtraction between the % of promoters (score of 9 and 10) and the % of detractors (score of 0 to 6). The NPS can be measured sometime after the customer/company exchange to know the degree of recommendation, where we would be speaking of relational NPS. In the heat of the moment and just after a specific and critical interaction to know the customer’s satisfaction in a specific process, it would be a case of transactional NPS.

A very good shopping experience in the APP may later coincide with a very bad experience in the delivery

In other words, we must ask ourselves either we want to measure an overall experience, or we want to measure a specific part of this experience. The results and their interpretation will be different, and so will the conclusions For example, in the second case, we could have several NPS indicators for the same company with different results depending on the service or the moment measured. A very good shopping experience in the APP may later coincide with a very bad delivery experience.

The first one offers a more established and faithful result to the combined reality of the company and the customer experience. Its purpose, which was to know the rate of promoters and detractors, is more realistic. A possible inter-company benchmark would also be more valuable with this data, whenever companies are willing to share their NPS…

Still, who hasn’t received a request from an agent at the end of a call: we are going to give you a quality survey and remember that a 9 or 10 is important to me…?

Then we can ask ourselves what is the difference between the transactional NPS and the CSAT.

Recall that the CSAT is a simple and effective system of measuring customer satisfaction. The customer rates the company in general with a score from 1 to 5 or with an Emoji icon. The sum of the scores divided by the number of responses gives the average score. The CSAT is often not very accurate and does not ultimately allow for decision making. The fact that dissatisfied customers tend to answer more often than satisfied ones can also make the calculation less realistic.

Speaking of indicators, we cannot forget the third one in the mix: the CES, also very commonly used. The CES changes somehow the concept, as we ask the user for their feelings regarding the “effort” made in their relationship with the company. The calculation is also simple, as the points obtained in the responses are added up and divided by the number of responses obtained.

The CES is interesting to know the customer’s opinion, especially in a complex part of a process. For example, it would be interesting to know the feeling of efforts made by a customer after having completed a registration process in our company or his feelings after having filled an incident report in an insurance company. At these important and complex moments, the customer will be able to give us feedback on how tedious or not the process was.

Collecting and analyzing data

But what about the contextual comments left by our customers? Where are they reflected and analyzed, what do we get from them and what corrective actions can we take?

At this point it is important to understand that the contact center is a strategic place to collect this information. Just after the interaction the customer can give a hot and well-founded opinion on the positive and/or negative. Analyzing this huge amount of unstructured information is not an easy task.

We can use the classical method of manually coding these texts. The company must define a coding table of one or two levels and the agents must code each one of the answers obtained in these tables.

The cost of this study can be high given the number of hours involved in this process, but the result in terms of accuracy can be very interesting.

But what about the contextual comments left by our customers? Where are they reflected and analyzed? Where are they reflected and analyzed?

Quickly, when faced with a high level of responses to be analyzed, the use of semantic analysis technologies will be necessary. The system will be able to extract keywords within expressions and contexts, and provide qualitative analysis based on high levels of data. These systems also allow us to cross-reference the data and to know, for example, whether those who have complained about the high price have also complained about other aspects. And if we have other data from our customers, we could find out the reasons for complaints from young people versus senior citizens. This is not a simple task and setting up an efficient process will require help from experts.

Over time, the temporal analysis will allow us to see the evolution of the different concepts defined and to check whether certain measures taken have had an effect. . For example, whether the platform improvements in delivery times on the southern area have led to a reduction in negative comments about delivery times in the south, or whether the new method of parcel return implemented on the platform has reduced the level of complaints.

The next step will be voice analysis, using similar systems, to understand and analyze customer feedback and sentiment. The first experiences are starting to take shape.

In conclusion, we can say that all the indicators discussed above are valid, and even complementary. Nowadays, knowing what customers think must be a driving force for analysis and improvement in companies. Customers have become accustomed to constantly rating the services they use, whether it is Cabify, a restaurant or their last purchase on Amazon. It is the companies that must know how to analyze and interpret these results in order to convert them into improvement action plans.