Hello!

This page is going to cover the Impact analysis for Linear data metrics. How it is calculated, what we use to determine if a comment is positive or negative, and more.

Note: Lumoa calls any metric that goes on a numbered scale a Linear metric. This can be 1-5, 0-7, etc. Things like 5 stars or sad/neutral/happy smileys are also considered Linear metrics. Your overall score is calculated and displayed in Lumoa as an average of all of the scores your customers submit. NPS is NOT a Linear metric as it uses a different system to calculate overall score, which you can read about here.

Question: I don't know what the numbers next to my negative and positive Topics mean.

Answer: They are there to show you how much that Topic affects your overall score

Example Topics

Using a 0-7 Linear range as an example, the potential range is anything from 0.0 (if every customer responds a 0) and 7.0 (if every customer responds a 7).

In the above image, you can see the "Connectivity" Topic is bringing the overall score down by -0.51 points, and bringing the overall score up by 0.04 points. By taking (-0.51 + 0.04), it leaves us a total of -0.47 points contributing to the overall score. This means that your apps Connectivity is, as a whole, pulling your overall score down. If you were to turn all of the low scoring Connectivity comments into higher scoring ones, you would gain over half a point to your overall score. In something like a 0-7 Linear scale, that is a lot!

Question: How is it determined if a comment is put into the Negative or Positive section of a Topic?

Answer: The score, the sentiment, and associated topics

Assuming I am using a 0-7 scale: if I submit a comment saying "I could not log in and the customer rep was rude" and I leave a score of 1, Lumoa will put that comment into Categories like Login and Customer Service Attitude. It will add them to the Negative Impact of those categories because the customer left a Negative score and is speaking negatively in the open text response. However, its worth noting that resolving one of these issues will not turn this comment into a positive score. They are talking about two separate issues, and you will need to fix both in order to make this customer give the highest score possible.

Additionally, most Linear metrics have numbers that are treated as more "neutral". This is usually some thing like a 3 using a 1-5 scale or a 4 if using a 1-7 scale. These comments can be flagged as either positive or negative depending on the algorithm the AI uses when looking at customer sentiment in the open text response. Our system uses AI to analyze comments and determine whether or not the open text your customer submits is positive or negative, and then uses that sentiment to push a comment in either direction.

As an example, if I am using a 1-7 scale, and I submit a comment saying "great customer service" with a score of 4, that comment will be placed into the Positive Impact for the Customer Service Topic. This is because even though in a vacuum a score of 4 is neutral, our AI can determine that this comment is overall positive, and assign it to an Impact accordingly.

In the below image you can see this in action. Here is a comment with a score of 4, being placed in to the negative Impact Notifications and Alerts, because the comment sentiment is Negative.



Note: Comments that have a negative sentiment in their open text response, but leave the highest possible positive score will not show up in the negative Impact for a Topic. This is because while the customer is talking about the Topic negatively, it is not enough to actually affect the score. Additionally, resolving the issue wouldn't gain you any additional points, since they have already left you the highest possible score.

Question: How are the Impact numbers next to my Topics calculated?

Answer: The Impact is based off of the amount of associated feedback and the sentiment of those feedback

The exact numbers are determined by analyzing sentiment, scores, and amounts of feedback associated with a certain topic.

First, we split each comment into Statements. These are usually sentences, but you can read more about Statements here. Then for each of those Statements we use AI to determine sentiment and topics being talked about. We compare those Statements with the scores, and use predictive analysis to determine how fixing those issues would Impact your score (note that there can be multiple issues for each open text response as explained above). We then aggregate that data, factoring in the amount of feedback associated with a certain topic.

As an example, the Impact will be lower for a topic with 5 comments than one with 100 comments, because you have more to gain from fixing the one with 100 comments. A higher Impact score means that more people are talking about this topic in their open text response, and you can use the positive and negative impact to determine how your customers are feeling about a topic overall.
Was this article helpful?
Cancel
Thank you!