Skip to content

TAG🏛️COLUMNS

image-20221107172409769

The app also allows you to perform fast cohort, lost and like for like analysis.

Cohort analysis

Select a column for which you want to analyze cohorts (for instance to split customers according to when they were acquired). Select the dimension that you would like to analyze by "time since". In this example "Client". This will create a new "Client_Since" dimension that, for each Client will specify the first period in which it the Client is referenced - aka "active" - in the dataset

image-20221109153131602

Select a plot in the Plot 📊 Charts panel. Under Plot 📊 options select our new Client_Since dimension.

image-20220918144748205

Select a metric, for instance Sales.

image-20220918144839655

In this specific case, we want the integrated legends to show up on the right side of the chart.

image-20220918145315954

Submit. This is the result. In the last rolling 12 months, 300k revenue came from customers acquired two years ago, that grew with a CAGR of 28%. 32k came from customers acquired last year, that grew with a CAGR of 23%. The "*" after the CAGR value indicates that the percentage has been calculated on a lower number of periods (2 instead of 3). And of 67k came from customers acquired in the most recent 12 months.

stackedcolumn

In fact, however, what you want to know is how many customers you (still) have for each cohort and how much did they buy on average. In the Tag 🏛️ Columns panel, set the "Select column for which you want to count unique items" to Client.

image-20221109153351745

Back in the Plot 📊 Charts panel select Client_Since as Dimension to plot...

image-20220918181036844

...and select Nber of Client as metric to plot. We want to know the number of client by cohort class. Hit Submit.

image-20220918181142553

You get first a chart with the total number of clients in each year...

stackedcolumn

....and just below the same chart with the breakdown by client cohort. We added 76 clients this year! Out of the 119 customers of three years ago, only 59 stayed the second year. All are still active except three.

stackedcolumn (1)

Ok. Now we want to know how like for like customers (the customers that were active in all three years) did in terms of sales.

Under the Tag 🏛️ Columns panel click on Compare Like to Like.

If you are analyzing two periods (say previous year vs actual) select "Two periods" in the "Calculate like for like across" widget. If you want to study like for like clients across all the periods present in the dataset (aka the clients that were active in all periods), click on "All periods".

image-20221109153436758

If you set the dimension to plot to Client and the metric to Number of Client...

image-20220918182636893

...you get the number of clients that were active in all three years, 51.

stackedcolumn (2)

How much did they buy on average after discounts? Set the metric to "Sales net discount by Client" and hit Submit.

image-20220918182921857

Between 2.8 and 5.0k.

stackedcolumn (3)

What if we were only interested, to match our AC (actual) vs PY (previous year) analysis only in comparing the customers active this year and last year? Select "Two periods" in the Calculate like to like across widget and hit submit.

image-20220918184351340

Here is the result. It is better to be active across all three years than to be active only in the two most recent.

stackedcolumn (6)

To compare this to the company average (that also includes the like for like customers), you might want to filter out the customers acquired in the current year (they probably have less than 12 months of sales).

In the Filter 🚬 Data panel, exclude Client Since of the current year.

image-20220918183414618

Under Tag 🏛️ Columns unselect "Compare like for like".

image-20221109154136677

Go to Plot 📊 Charts and hit Submit. Average sales per client are lower than sales per customer of the most faithful clients.

stackedcolumn (5)

This is a case in which using the same scale across all three charts comes in handy.

Plot the "like to like all periods" chart first after setting the fix chart scale widget to True.

image-20220918184800841

The chart is plotted with a 5234 top scale, the same we will use for the other charts.

image-20220918184923334

Now plot the other two charts. Here are the three charts plotted with the same scale.

image-20220918185319301

You can do similar analysis with other metrics (average price per customer, average discount per customer, average units per customer,...).

Please note that if the metric is not stackable, the "per dimension" plots will only be shown as small multiple.

Take the plot we are already familiar with.

stackedcolumn

With "Plot small multiples" set on False the chart is plotted only as a total and not by dimension because this would not make sense (because you cannot stack the values).

image-20220919150937636

If you set Plot small multiples to True the get the same total chart plus the small multiples chart, in this case by Region. The plots are ranked by Sales descending.

stackedcolumn (1)

You can also do a similar calculation and analyze the "churning" (dropped/lost) customers.

In the "Select column for which you want to analyze lost & dropped" select customer.

image-20221109154241200

This will generate a "Client_Lost" dimension than you can select in the "Plot 📊 Charts" panel

image-20221109154555531

You end up with a chart like this.

stackedcolumn

This might look scary. Look at all those customers the company is loosing. Let's check sales.

image-20221109154810957

Hit the Submit button and you get this. In fact the "churning" customers almost brought no sales, and it is probably good that they have dropped out 🙂.

stackedcolumn (1)

Congratulations! You are now cleared to move on to the "Plot 📊 Charts" panel.