@Erikahuer This is a little bit heavy on statistic side so let’s make it simple this way. The analysis find  which topics are most strongly correlated with higher or lower NPS scores. The % you see show the proportion of influence that each topic has to NPS. A higher % means the topic has a stronger correlation with changes in NPS compare to others  - hence the name “Relative Importance”

So in the picture it would be that paquetes influences a 43%
So if the comments has the topic paquetes and the comments is negative it means it is a 43% possibility that the NPS will be negative?

@Erikahuer No, the number 43% in here just to compares the influence power of the topic against the other one with 18%. It’s normalize the total power of 100% and your topics will share that 100% so you can compare them to each other. So in your case you can say that in all of the topics, when respondent talk about “paquetes”, their NPS will very likely follow what they said compare to when they talk about Sitiio.

In other words, Working on paquetes, compared to sitio, would likely have a higher impact on improving your NPS.

Also, not sure in which field you are exactly but be careful with one thing. For some topics, people won’t necessarily be satisfied if it works, as it’s expected behavior and they won’t talk about it but if it’s not working properly, you can expect negative comments. Example : the light switch in your hotel room. You won’t even notice it if it works and you will definitely not talk about that in an open text question but you’re expecting it to work properly and will mention it if the light is not working. Hope this helps! 

Guys thank you!
Just one more question how do we measure the importance of the topic?
I mean it uses the sentiment score? 
How do you know or mesure it?