Skip to main content
Impact Analysis by Net Sentiment
Jack Miller avatar
Written by Jack Miller
Updated over a week ago


When we receive a new comment in your Chattermill Project, our powerful AI automatically applies one or more themes to it in order to help you understand how your customers feel about your product. Impact analysis allows you to see which themes contribute positively to your Net Sentiment and which have a negative effect, making it easier to decide where your attention needs to be focused to improve customer experience.

Reinterpreting your Net Sentiment

The theoretical minimum that your Net Sentiment can be is -100, so the positive part of your Net Sentiment is the difference between your Net Sentiment and the minimum. Using the example in the graph above, you can think of it as if you have 151 Net Sentiment points.

The theoretical maximum that your Net Sentiment can be is +100, so the negative part of your Net Sentiment is the difference between your Net Sentiment and the maximum. Using the example in the graph above, you can think of it as if you have lost out on 49 Net Sentiment points.

The positive and negative parts of your Net Sentiment are then split among the significant themes proportional to their individual total effects.

Interpreting the sums

The "contribution to Net Sentiment" for the positive coefficient themes should sum up to the positive part of the Net Sentiment and the same for the negative coefficient themes for the negative part of the Net Sentiment.

If you sum up the absolute values of the Net Sentiment points you have gained and the ones you have lost out on, you should get the theoretical maximum amount of Net Sentiment points you can obtain, which is 200 since there are 200 points on the Net Sentiment scale.

Our analysis can tell you which themes were responsible for the gains and losses on this scale.


Using contribution in comparisons

You can use Impact Analysis to compare two periods of time and the score contributions that different themes had to your Net Sentiment over time, as shown in the following model.


Can I add this to a chart or dashboard? What about exporting my data into a CSV?

Yes! You can add these charts to a dashboard just as you can for any other chart. Once you have made your changes, add a title to the report and click 'Save report.' You will then find this report in the list of reports when adding to a dashboard. Exports are not available at the moment but will be coming soon.

Can I share the Impact Analysis chart with my colleagues or externally?

Yes. Once your chart has been saved as a report, you can share this the same way as your other reports. You will find the share icon in the top right of the app, and you can select your teammates to share with internally or use the link to share externally (those without a Chattermill account will then have read-only access).

Do you use all responses (inc without a comment) when calculating Impact Analysis?

We calculate scores using all responses with responses when calculating Net Sentiment.

What is the minimum number of responses I need to run Impact Analysis?

There is no hard minimum threshold, but we would advise having at least a total of 500 responses. Additionally, themes with less than ten responses are filtered out.

Some themes have scores greater than 100, isn't that higher than the theoretical maximum Net Sentiment score?

Theme contribution scores can be greater than 100 or lower than -100 because of the way Impact Analysis is calculated. For example, when calculating the positive drivers of CX, we start at -100 (the largest negative number for Net Sentiment) to calculate a theme's impact on Net Sentiment and vice versa for negative drivers.

I am getting a 'Sample size exceeds maximum limit' error. What does this mean?

The maximum response limit for single Impact Analysis is 150,000 and 300,000 for multiple Impact Analysis. If you reach this limit we advise adding filters to reduce sample size or contacting [email protected] for a manual export.

I can select Impact Analysis by NPS or Net Sentiment. What is the difference between the two metrics?

The main difference between both metrics is that Net Sentiment reflects the positivity and negativity of comments without taking the score into account, whereas Impact Analysis by NPS also explains the score because we have both sentiment and score used within the calculation.

Did this answer your question?