Skip to main content
All CollectionsPlanhat ModulesCustomer Intelligence Module
Trend charts: Cohort & Time-series charts
Trend charts: Cohort & Time-series charts

Deep dive into two trend charts: cohort charts and time-series charts

Christian Dreyer avatar
Written by Christian Dreyer
Updated over a week ago

Let's have a look at trend charts in Planhat! What’s unique about this category of charts is that trend charts display data over time, rather than a snapshot of current data.

Trend charts are therefore the ones you want if you’re assessing whether something is improving or worsening over time, or how proportions change over time (e.g. proportion of ARR in the Enterprise tier year over year).

It’s even possible to use trend charts for forecasting - with a Cohort Bar to view Licenses by status, plotted by future end dates, for example.

Let’s have a look at the differences between the trend chart types.

Cohort Bar and Cohort Line

Cohort charts are unique in that they can display current data over time. This means it can take data that’s not inherently time-series (i.e. field data), and convert it into a form of time-series data, which is really cool!

To do this, you choose a date field associated with the model on which the chosen field lives, which is then used to plot the time axis. For example, if it's a Company field, you can select from date fields such as “Date Created”, “Customer Since” and “Renewal (Date)”.

Note: Cohort charts are not the best suited for plotting data that’s already time-series, such as calculated metrics. We recommend you use any other trend charts for this.

So when would you use a Cohort chart?

Think of what you might show in a single bar or line chart, and then add a time component. The category becomes bar segments (rather than bars) or lines, as the x-axis is now time.


What do we mean by this? For example:

  • If the Single Vertical Bar chart is:

    • Sum of Company ARR (y-axis - number axis) / Industry (x-axis - category axis)

  • Then the Cohort Bar chart could be:

    • Sum of Company ARR (y-axis - number axis) / Industry (bar segments) / year created (x-axis - time axis)

As you can see, there are slight differences between the two Cohort charts, so depending on what you need, you could go one way or the other.

If you want to see the total for each time period - use a Cohort Bar.
If you prefer seeing relative sizes - use a Cohort line.


Setting up a Cohort Bar or Line

  1. Choose a model (e.g. Company)

  2. Set up the number axis (y-axis)

    1. Count the number of records (e.g. number of Companies); or

    2. Sum, average, max or min a field/property (e.g. sum of ARR)

  3. Choose what category property defines the bar segments or lines (e.g. Industry)

  4. Define the time axis (x-axis)

    1. Select a date field - the options here will depend on the data model your field chosen in (2) lives on (e.g. “Date Created”)

    2. Select what time period the bars should correspond to (week, month, quarter, fiscal quarter, year or fiscal year)

💅🏻 Design tip: If you end up with blank time periods (e.g. if it’s showing 7 years, but your data doesn’t go back that far), you can tidy this up:

  • Use “Limit results” to define the number of time periods shown in the chart.

  • Use “Offset (skip first x items)” to cut out specific bars, starting from the newest.

💅🏻 Design tip 2: Sometimes, you may want to simplify the view of your Cohorts, e.g. if you want to plot one thing over time and you don’t need to split it into categories.

Let's say for instance that your product is a training platform, and you want to plot course enrolments over time, which can be achieved in a Cohort chart by plotting count of End Users by “Enrolment date”. To do this, skip step (3), defining a category, to keep it simple.

👑 Pro tip: This one is really special: you can use “Offset” with negative numbers to look into the future! 🔮 This is a great trick to forecast your Licenses, for instance!

How would you set this up? Create a cohort chart that counts the number of Licenses, showing the Licenses per month, based on the default “End Date” field on the License model. Your bar segments are defined by the default “Status” field on the License model - bars are divided into are “ongoing”, “renewed”, and “lost”. By using a negative number in the offset, you can start showing Licenses with end dates in the future. The exact negative number will depend on how many bars you want to show; for the default of 12 bars for 12 months, having an offset of -11 means you show the current month as well as 11 months into the future.

Time Series Bar, Line and Stacked Bar

Next, let’s have a look at the “Time Series” charts. As you’d expect, their defining character trait is that they specifically display time-series data or metrics.

This could be a system metric (including those associated with conversations, tasks and revenue), or a calculated metric that you have created. You then use the charts to perform analysis on the time-series data, to gain further insight into changes over time. Remember, as they are trend charts, the x-axis is always time.

Time Series Bar and Line

The Time Series Bar and Line are typically used to display analysis of a single metric - although you can add others if you like (more on this later). This works best with a relatively “raw” form of time-series data, such as the system metrics “email”, “chat” or “number of users”.

In the chart settings, you choose how to visualise the data. There are three main elements to this:

  1. How to consider values over time. This affects the increments on the x-axis. Options are:

    • Simple daily values (plotting days on the x-axis)

    • Sum of daily values across a period (week, month, quarter, year etc.)

    • Average of daily values across a period (week, month, quarter, year etc.)

  2. How to consider multiple values (e.g. for multiple Companies) within the time period. Options are:

    • Sum

    • Average (all)

    • Average (values)}

  3. Choose your timespan - i.e. number of days/weeks/months on the x-axis - the number of days, weeks or months etc., depending on what you select in “aggregate by”. Let’s go through some examples, to show how settings (1) and (2) influence the chart you end up with. We’ll use the system metric “email”, and show different ways you can set it up, depending on which insights you want to gain.

Stacked Time Series Bar

The stacked Time Series Bar chart allows you even greater insights into trends over time. It's similar to the regular Time Series Bar chart in that it’s starting with metric data, and aggregating it over a time period, as defined by you. The key difference is that the bars are now also divided up into segments. Because of that, you can visualise how the composition changes over time, allowing you to analyse the relative contributions of the different groups in the bar segments.

They, therefore, look a bit like a Vertical Stacked Bar chart or a Cohort Bar chart. Whereas in those charts, the bar segments were qualitative category values, such as "Industry" (corresponding to the field); in a Stacked Time Series Bar chart the segments are based on numerical value ranges (corresponding to the metric), and the segment criteria are defined by you. This makes sense seeing as the source data is metrics rather than fields, which is why it is the only chart type with bar segments like that.

Setting up the Stacked Time Series chart

As for Time Series charts, you choose how to visualise the data in the chart settings. There are three main elements to consider:

  1. How to consider values over time. This affects the increments on the x-axis. Options are:

    • Simple daily values (plotting days on the x-axis)

    • Sum of daily values across a period (week, month, quarter, year etc.)

    • Average of daily values across a period (week, month, quarter, year etc.)

  2. How to consider multiple values (e.g. for multiple Companies) within the time period. Options are:

    • Sum

    • Average (all)

    • Average (values)}

    • Count - this is unique to stacked time series!

  3. Choose your timespan - i.e. number of days/weeks/months on the x-axis - the number of days, weeks or months etc., depending on what you select in “aggregate by”.

  4. Define your bar segments. You can have as many as you like.

    1. Define a numerical range for each segment that makes sense for your metric. E.g. For Health Score, set segments within the range of 0-10. Make sure the bar segments don’t overlap!

    2. Choose a colour for each segment. E.g. For Health Scores, it makes sense to apply the usual red/amber/green colouring.

Did this answer your question?