The Focus Recommendations module uses predictive analytics to highlight the topics you should focus on - the topics that have the greatest impact on customer satisfaction/dissatisfaction. This can help you prioritize future initiatives.

Using data from sources like NPS or CSAT surveys (any source where a rating scale is paired with an open-ended response) the Focus Recommendations module will identify up to 12 topics and rank the impact of these topics on the associated score. The impact could be positive or negative.

You can add the option of leveraging GPT for personalized recommendations on how to improve your customer experience in the Focus Recommendations module.

You will have to set up Focus Recommendations individually for each dataset.

Go to the Sources icon on the left of the screen. From there select the source for which you would like to enable Focus Recommendations, click on the three dots, and select Edit.

Toggle the Focus Recommendations to on. Choose the Rating Scale from your source and then choose the associated open-ended responses. You can only choose one open-ended response variable so choose the one that most closely relates to the scale you have chosen.

On the right side of the panel, drag and drop the rating in order from most positive to most negative. Sometimes elements will appear in the scale that have no value, for example "N/A". It is essential that you remove these elements from the rating scale for the calculation to be accurate. You can do so by scrolling over the unwanted element and selecting the remove symbol at the right. Once all parameters are set, select "Save Changes".

To access Focus Recommendations, click on the Analysis Icon on the left of the screen. If you have more than one source, select the source you want to work with under the Filters panel. In our case we only have one source "Reviews".

There are two results options, standard and with the integration of GPT.

The size of the target represents how positive or negative the impact of the topic is on the rating. If a topic is highly positive or negative the target will be larger.

The percentage is the percentage of correlation. The higher the percentage, the greater the correlation between the topic and the rating.

Note: Topics with a higher impact will not necessarily have the highest volume of comments.

Scroll over the target to see details.

Beside each target are the Opinions associated with that Topic. Click on the link and you will be taken to a prefiltered Opinions page where you can dive right down to the original reviews of each opinion, if desired.

You can also add tags to your topics.

The view with the integration of GPT has the same layout as the standard version but replaces the list of associated Opinions with a personalized summary generated by using Keatext’s insights as a knowledge base for GPT. Please see our sales team if you would like to take advantage of this option.

In some cases, certain topics will be too general and provide little value to the analysis. These topics can be hidden to make room for more meaningful topics. For example, a company name. This will provide little value to the analysis as everything is related to the company. In this case, select the topic checkbox and then click "Hide".

You can also view the Focus Recommendation as a chart by clicking the chart view on the right of the screen. This view divides the Focus Recommendations into four quadrants showing which ones are the Key Strengths and Key Weaknesses and allowing you to easily spot topics for improvement which will most impact the rating.

In the example above we are looking at a Customer Satisfaction Survey for a Holiday Resort company. We can immediately see that the bed is the topic that needs the greatest improvement in order to raise the CSAT score, followed by the pool and parking. For more details look at the personalized recommendations furnished by GPT on the list page. The restaurant and the kitchen are two areas that lead to a higher CSAT score so keep doing what you are doing in this area!