Multi-dimensional analysis with Data Tables

When analyzing a rich dataset, analysts often need to compare multiple metrics at once, and slice and dice that data by different dimensions to generate a custom analysis. Amplitude’s Data Tables enable multi-metric, multi-dimensional analyses in a single view. It is an extremely flexible chart, allowing you to quickly analyze any combination of user behavior, user attributes, and metrics. 

Data Tables are useful for:

• Marketing attribution (total visits, page views, and conversion rate by UTM source)
• Market Segment analysis (comparing several actions broken down by country)
• Experiment analysis (multiple metrics by control vs variant groups)
• Trend investigation (quickly and easily breaking down a number by multiple properties)
• Comparing time periods across multiple metrics (metric A, metric B, and metric C, broken down by category, compared to last quarter)

You can easily sort columns in ascending or descending order (just click the metric header), drag and drop or resize columns, and highlight, copy and paste any number of cells from your Data Table.

TIP: To get a quick overview of how Data Tables work, check out our short Loom tutorial.

Set up a Data Table

To set up and use a Data Table, follow these steps:

1. Navigate to Create New > Analysis > Data Table.
   
2. In the empty Data Table panel, click Add an event or metric and select the event or metric you’re interested in. A new Data Table will open, with your chosen event or metric in the first column. Add more by clicking + Add Event or Metric in the rightmost column in the Data Table.
   
   You can create a new metric at this point, if you need to. 
   

    

3. To break out your events and metrics by property values—country, for example, or platform, or week—click Select property… in the leftmost column of the table and choose the property you’re interested in.
   
   
   
   This will run a group-by on your events and metrics, grouping by the property you selected. You can include up to five top-level group-bys in a single Data Table.
   
   
4. Once you’ve added a group-by property, you can run a secondary group-by on that row of your Data Table. For example, you can break your events and metrics out by the Week property nested within Brand.
   
   
   Click in the rightmost group-by column and select the property you’re interested in.
   
   
   
   
5. Next, add user segments, if desired. Saved segments are accessible. Multiple segments will show up in the table as separate columns within the same metric.

NOTE: Within any cell, click the Options icon to: 

• • • Open as chart, which will open a new tab with the chosen metric applied;
    • Copy the data so you can paste elsewhere as needed;
    • Export the data as CSV or Excel file.

Using metrics in Data Tables

With Data Tables, including a "-" character in any cells included in your formula's calculation will result in an error.

In some cases, using Uniques as a metric type in combination with group-bys can generate results that appear counterintuitive at first. For example, when a group-by is added to the event in the left column, the total sum for the event (as seen on the top row) is not a sum of each of the rows below. Because there is a group-by applied to the event, the same user can exist in multiple rows.

Filter your events and metrics for specific values within a group-by 

You can click to select which property values you want to keep or hide in the table.

You can also add an ad hoc filter for in-line events or metrics. To do so, click the three dots from the event or metric header and select “Add Filter.” This will let you apply filters on top of your events or metrics. Once applied, you can see what filters are applied by hovering over the icon.

Results limits and sorting logic

For more complex analyses, it's important to understand how Amplitude Analytics decides what results to display, as well as what happens when you sort on a given column.

• If your group-by includes a large number of different property values, Amplitude Analytics will only display the top 100 group-by results. With multiple top-level group-bys, the limit is 500.
• If you are nesting group-bys, Amplitude Analytics applies the limit to each one separately.
• For conversion metrics, the top property values are determined by the totals count of the first event in the funnel. For session metrics, they’re determined by the totals count for Any Active Event.

Once you have these results, any sorting you do will apply only to them, and will not bring in any new results. For example, imagine your group-by has enough different property values that Amplitude Analytics limits the results displayed to the top 100. By default, these results are sorted in descending order. If you opt to view your results in ascending order, you will not see the “bottom 100” results instead. You will still see only the same top 100 results—only their sorting order has changed.

When you are using multiple metrics, sorting by a particular column will display data for all columns based on the values in the sorted column. For a data table with multiple segments, multiple metrics, and a period over period comparison, sorting a period-over-period column within a metric will give you a dataset based on the first segment's current period.