skip to Main Content

How do you analyse NPS comments to improve net promoter score?

NPS is the statistic consolidation of all the customer feedback scores collected with the famous “how likely are you to recommend” question (detailed formula to calculate NPS). Net promoter surveys are used by retailers to measure their customer loyalty (L’Occitane or Manutan for example) or by customer service to measure their customer satisfaction.

While the net promoter score is a good KPI of customer satisfaction, it cannot be properly understood without analysing the verbatim associated to the nps question.

Why do you need to analyse NPS comments?

In its original paper “the one number you need to grow” published in the prestigious Harvard Business Review, Frederick F. Reichheld concluded by these words:

The path to sustainable, profitable growth begins with creating more promoters and fewer detractors and making your net-promoter number transparent throughout your organization. This number is the one number you need to grow. It’s that simple and that profound.

Since its publication in 2003, the people in charge of the performance management of customer experience keep asking the same question: whats a good net promoter score? And keep looking for nps benchmark by industry.

To create more promoters and fewer detractors is to apply one particular customer feedback best practices: look for your net promoter score meaning! Stated otherwise: use the nps comments to build and drive nps action plan that will enhance your customers experience.

The voice of customer expressed in nps review is the best source of customer insight for good reasons:

  1. contrary to social media, only your customers answer these surveys,
  2. its linked to an actual and recent interaction in the customer journey (buying in store or online, calling the customer service…),
  3. its data can be enriched with customer segments and other data from your CRM,
  4. the nps comments can read in the context of a meaningful score that gives you an objective idea of the state of mind of your customer.

Therefore, if you really want to turn your NPS surveys into action plans, you must analyse verbatim from your customer feedbacks. How to use net promoter score when you have a lot of volume (more than a thousand per month)? have a lot of history or a lot of variables? You need to rely on some kind of natural language processing: text mining or sentiment analysis.

Dictanova API is a good solution to analyse verbatim from NPS surveys and easily correlate net promoter score and NPS comments.

How to analyse open ended survey responses like NPS comments in Dictanova API?

When you want to process data with the Dictanova API, you must create a dataset that defines how you will enrich the text to be analysed. In our terminology this enrichment is called metadata.

These metadata are additional fields that will act as filters or analytics dimensions. Each metadata has a type which tells Dictanova API:

  • what kind of values to expect
  • how it can be used for search and analytics

In particular there is a metadata type specifically designed for NPS scores. It will ensure that the NPS scores is valid (between 0 and 10) and will allow the use of specific actions like grouping in promoters, detractors and passive.

Here is a minimal example for creating a dataset enriched with NPS metadata:

{
    "name": "My dataset with NPS metadata",
    "lang": "en",
    "type": "SATISFACTION_SURVEY_RESPONSE",
    "industry": "RETAIL_B2C",
    "metadataDefinitions": [
        {
            "code": "my_nps_score",
            "type": "NPS",
            "mandatory": true,
            "personalData": false
        }
    ]
}

And here is how a document should be formated to be imported into this dataset:

{
    "externalId": "my-external-id-1",
    "content": "My very first verbatim with NPS to be analysed in Dictanova.",
    "metadata": [
        {
            "code": "nps",
            "value": 9
        }
    ]
}

To get the whole detailed process of creating a dataset and importing data, you can refer to the technical documentation : Importing data and get analysis results.

What customer insights can I get from NPS survey responses?

Once you have created a dataset with NPS metadata and you have imported open ended survey responses enriched with NPS scores, you can build net promoter score charts to get customer insight.

There is actually a lot of possibilities and it would be impossible to talk about them all in this blog article so we will focus on three common use cases which will be detailed in other blog posts.

Compute NPS subratings

Global NPS is not precise enough to guide decisions and measure the efficiency of action plans. Often you need a NPS for particular aspects of your business: services, prices, products… The most common way is to add NPS questions but it means longer surveys which is challenging when response rates to surveys keep dropping.

With Dictanova API you can compute a NPS score for each opinion expressed in customer feedbacks, for groups of opinions or for topics. You only need a global NPS score and an associated NPS comment. Then, you can compute subratings of the global NPS score for virtually everything that is discussed in the NPS comments. No need to add specific questions which is a good way to preserve the conversion rate of your NPS surveys.

Example of subratings by NPS groups

Identify opinions that impacts customer recommandation

Whether you have a good or a bad net promoter score, you certainly wondering : how to interpret my net promoter score? Or even how to increase my nps score? Once again the answers are most likely in the NPS comments.

With Dictanova API you can measure the impact of main opinions, of all topics, on the net promoter score. One way is to measure how much the global net promoter score evolves when you neutralize an opinion or a topic. If the NPS increases, the opinion or topic has a negative impact. If the NPS decreases, the opinion or topic has a positive impact. The more it increases or decreases, the more impact (positive or negative) it has.

Example of dataviz to display opinions with a positive impact on NPS
Example of dataviz to display opinions with a negative impact on NPS

Follow NPS evolution over time for particular topics or opinions

Relative values are often more relevant than absolute values. NPS is no exception. While you may be wondering whats a good net promoter score, you should instead certainly consider how your NPS evolves over time. Whether it increases or decreases, you should look for explanations of those variations… and they most likely are in the NPS comments.

As seen previously, with Dictanova API you can compute NPS subratings as a screenshot over a period but also as an evolution in time for each opinion expressed in customer feedbacks, for groups of opinions or for topics. To dig even deeper in the understanding of your net promoter score variations, you can visualize the evolution when it is discussed positively or negatively, which is one of the benefits of sentiment analysis.

Example of evolution of NPS associated to topic (positive and negative)

All the examples presented in this post have been computed using our API on real data.

Fabien Poulard

CEO et fondateur de Dictanova, je suis docteur en informatique spécialisé dans le traitement automatique du langage naturel.

Leave a Reply

Your email address will not be published. Required fields are marked *

Back To Top
×Close search
Search