Skip to content
English
Contact us
  • There are no suggestions because the search field is empty.

Power BI: Cross Tables

The Cross Table page helps you understand the relationship between two different dimensions of your data. It answers questions like "when customers in different wards mention specific hazards, how does their sentiment compare?" or "which property types experience which maintenance issues most frequently?" This is your matrix analysis page for finding patterns and hotspots across your data.

How to use this page

Understanding the matrix

The main feature is a matrix in the lower half of the page. Think of it as a grid. Your rows are one dimension of your data, your columns are another, and the cells where they intersect show information about that specific combination.

For example, if your rows are health and safety hazard themes (asbestos, damp and mould, fire risk) and your columns are wards (North, South, East, West), each cell shows data about customers in a specific ward talking about a specific hazard.

Setting up rows and columns

On the upper left, use the row filters to choose what appears in your matrix rows. On the right, use the column filters to choose what appears in your columns. Using the hazards and wards example above, you'd select the hazard theme category for rows and Ward for columns. The matrix populates with each hazard type down the side and each ward across the top.

Choosing your metric

Above the matrix, select which metric fills the cells:

  • Volume: How many comments mention that row-column combination.
  • Sentiment: The sentiment score for that combination.
  • KPI: Your relevant business metric (for example, TSM score).
  • Row percentage: Of all comments from the theme in this row, what percentage also mention the theme in this column?
  • Column percentage: Of all comments about the theme in this column, what percentage come from the theme in this row?

Heat mapping is applied automatically, using colour coding to make patterns immediately visible. The scale adjusts based on what's actually in your filtered data, so it's always relative to your current view.

 

 

Understanding row vs column percentage

These two views answer different questions, so choose based on what you want to understand:

  • Row percentage answers: "For this particular issue, which customer groups are talking about it most?" For example, of all feedback about damp and mould (the row), what proportion comes from the West ward (the column)?
  • Column percentage answers: "For this particular customer group, what proportion of their feedback relates to each issue?" For example, of all comments from customers in the West (the column), what percentage mention damp and mould (the row)?

Note that column percentage totals don't always add up to 100%, because not every comment from a given ward will be about a hazard, customers may be discussing something entirely different.

Using the summary tables

Above the matrix, you'll see summary tables showing totals for your rows and columns. These update based on your selected metric and can be sorted by clicking on the column headers. This is useful for quickly identifying which rows or columns have the highest volume, lowest sentiment, or most significant figures.

These summary tables are also helpful for contextualising cells with very low volumes. A sentiment score of zero might look dramatic in the heatmap, but if the summary table shows only one comment behind it, the finding may not be statistically meaningful. Always check volume alongside sentiment before drawing conclusions.

Drilling up and down

If you want to switch between viewing whole categories and individual themes, right-click on the matrix and select Drill down or Drill up. This works independently for rows and columns, so you can start broad and get more specific as you identify areas of interest.

Drilling through to verbatim

Click on any cell in the matrix, right-click, and select Drill through to verbatim. You'll see all comments relating to that specific row-column combination, for example, all comments from customers in the West who mentioned damp and mould. The volume and sentiment figures at the top of the verbatim page update to reflect just that cell.

Filtering for cleaner visuals

If your matrix is getting busy, use Shift-click to select only the rows or columns you want. For example, if you're presenting on five specific hazards, Shift-click those five themes in the column table. The matrix updates to show only those columns, which is much cleaner for screenshots or discussion.

Key points

    • Purpose: Analyse relationships between two data dimensions to identify patterns and hotspots.
    • Matrix layout: Rows show one dimension (for example, themes), columns show another (for example, demographics), and cells show their intersection.
    • Row and column selection: Choose what to analyse using the left (rows) and right (columns) sidebars.
    • Metric options: Display volume, sentiment, KPIs, row percentage, or column percentage in matrix cells.
    • Heat mapping: Automatic colour coding highlights high or low values based on the selected metric.
    • Row percentage meaning: Of all comments from the theme in this row, what percentage also mention the theme in this column?
    • Column percentage meaning: Of all comments about the theme in this column, what percentage come from the theme in this row?
    • Summary tables: Row and column summary tables can be sorted by any metric for quick identification of extremes.
    • Drill up/down: Expand from categories to individual themes separately for rows and columns using the arrows in the top right of the matrix.
    • Drill-through: Click any cell to access verbatim for that specific row-column combination.

 

Use case

This page is particularly useful for:

  • Identifying geographic hotspots for specific issues (for example, which wards have the worst damp and mould experiences?).
  • Understanding demographic differences in experience (do older customers have different concerns than younger customers?).
  • Analysing service overlap (when customers raise one issue, what else do they mention?).
  • Building business cases (showing which customer segments are most affected by which issues).

Viewing findings for multiple segments at once (for example, the sentiment score for each stage in your customer journey, broken down by ward).

Zoe-1    ✍️ Article written by: Zoe, Customer Success Manager 

Still in need of some help? Give us an email on support@wordnerds.ai or reach out to your CSM directly.