The field segmentation chart shows important information about the various segments in your dataset and exists in its simplest form on your Kapiche dashboard overview.

Observed Frequency

The observed frequency shows the percentage of records that a particular segment matches in the data. This percentage will also be reflective of any filters that are applied to the dashboard.

The particular segments displayed are based on the field in your data that is selected, which can be changed via the drop down menu: 

You can also change the way segments are sorted with the SORT BY menu. If you choose to sort by Net Promoter Score (NPS), extra information will be added to the chart. Keep in mind that the blue bars will still show observed frequency.

When drilling down into a saved query on the dashboard, the field segmentation chart shows an additional value, Expected Frequency.

Expected Frequency

The expected frequency, depicted by the small vertical lines on each segment's row, is calculated as follows: "Economy Class" occurs in approximately 2/3 of all records in the whole dataset, therefore, we expect it to occur in about 2/3  of the records for our selected saved query. This is its expected frequency.

The real power of this chart comes to life when comparing the observed and expected frequencies. As they diverge from each other it indicates a potential relationship between a saved query and a segment.

In the above example we can see our saved query about "In Flight Entertainment" is mentioned significantly more by "Economy Class" respondents than expected. Conversely, we see the opposite with "Business Class" respondents. For more information, you can hover over a segment's row to show a contextual tooltip.

This tooltip shows:

  • The observed and expected frequencies as both absolute count of records and relative percentage.

  • NPS, if it is available.

  • Raw difference: calculated by subtracting the expected frequency from the observed.

  • Relative difference: divides the absolute difference of relative observed and expected frequencies by the arithmetic mean. This metric can highlight significant deviations in lower frequency segments. The exact formula we use is:

The approach outlined so far can give us some really solid indications of what customers in different segments value most.  From this point we could explore deeper and come to understand how the customers in a particular segment feel about a theme. Or we may have been building a case about a particular aspect of our product/service to share with the business.

We think this chart is a powerful tool for finding insights and unlocking value in your structured data!


Did this answer your question?