Graph view allows you to construct powerful visualizations of your data over time. In the graph view, you can analyze any event, segment, or property and look at multiple types of analysis on the same graph.
Making Your First Graph
To create a graph, all you need to do is select an event of interest from the dropdown and click run query! Let's look at the builtin Pageview event. By default, Heap presents results per day, over the past week and as a line graph.
- Query Your query is made up of your chosen analysis type and event, as well as any filters, groupings, or segment comparisons.
- Date Range This is where you choose the date range over which you analyze your data. By default, Heap will set the date range of the graph to the previous week. The graph view supports the
Past 3 Months,
Past Year, or a custom date range. A week, month, or year refers to 7, 30, or 365 days, rather than a calendar week, month, or year.
- Granularity This value determines the interval by which we view the analysis - either by
Month. By default, this value is set to
Day. Intervals refer to calendar days, weeks, and months, rather than a rolling 1, 7, or 30 day window.
- Chart Type A line chart is the default chart type, but Heap supports analyzing your data as a bar graph, pie chart or in table format as well.
- Trailing Average The trailing average checkbox changes your graph from individual day counts to the average of the past 7 days, 4 weeks, or 3 months depending on the chosen granularity.
As you can see from the resulting graph (click Show Results), the Pageview count varied over the week analyzed from about 12,000 to 27,000 per day. Hovering near one of the data points, you can see the specific value based upon the granularity defined. There were 21,800 pageviews on September 4, up 41% from September 3.
There are nine query types in Graph view, which you can find by clicking on the query dropdown.
Count, Count Unique
Count returns the number of times an event occurs each hour, day, week, or month. Count Unique returns the number of unique or individual users who have performed an event. If Sally clicks log in 10 times, Count Unique will return 1 because Sally is one user. (This replaces the
unique checkbox from the old graph view.)
Size of Segment
The size of a segment by day. The number you see will be all users who meet that criteria on the given day. For segments defined by actions that occurred at any time and/or user level properties, the segment will continuously grow as more and more users meet that criteria. For segments that use time constraints, such as has done 'Sign up' in the prior 30 days and has not done 'Purchase' in the prior seven days, the number on a given date will reflect the number of users who have not Purchased in the past seven days, and who have signed up in the prior 30 days.
Sum, Average, Max, Min
When used with numerical event properties (e.g. total order amount), these functions graph the total, average, or max/min per day/week/month, based on the chosen granularity.
Average Time Between
Calculates the average time it takes to complete event two after having performed event one. Granularity plays a large role in how the average is calculated. The user has a window of time to complete a set of two events in order to be included in the graph. If the granularity is a day the user has 24 hours to complete the second event. If the granularity is a week, the user has 7 days. If the granularity is a month, the user has 30 days. If a user completes both events, the graph calculates the time between the first event and the first time the second event is completed, then returns the average across all users who have completed event one and event two within the window specified.
Conversion Rate Between
Tracks the percentage of users who did event one and then did event two. As with Average Time Between, the amount of time a user has to complete event one and two is dependent on the granularity. If the granularity is a day, a user has 24 hours to complete both events; a week, 7 days; a month, 30 days.
Let's return to our Pageview graph example. Just like queries in other analysis views, we can add filter statements to our data. We'll run a query to graph the unique count of users who visited from Google.
Heap's graph view becomes even more powerful when using the Group by clause to compare activity across different groups. Building upon the previous example, we could use the Group by tool to see the distribution of pageviews from Google based on device type. To do so, all you need to do is add a group by
Device Type clause to the query.
In this example, we changed the
Chart Type to
Bar to better show the distribution of devices on which our users visited our website.
Heap's Group by clause can group graphs by a variety of parameters such as whether someone has or has not done an event, the count of how many times someone has completed an action, whether they are in a segment, the first date of an event, or any other user or event property. You can also perform a double Group by to add more complexity to your analysis.
If we want to know how membership in a segment affects a given analysis, we can group by a segment and see the data for users who are are and who are not in the segment. However, if we want to see how two or more segments behave against each other, we can use the Compare clause. In this example, we want to see whether users convert at a higher rate from Adwords or from our Adroll retargeting campaign.
As you can see, over the past 3 months, the trailing 7 day average shows that visitors who returned after seeing one of our retargeting campaigns convert at a significantly higher rate than from Adwords, although Adwords delivers a more constant stream of converted customers. With this kind of insight, we can make decisions on our marketing spend based on our current priorities. Adding data around acquisition costs and lifetime value (LTV) might further inform our decisions.
Heap's graph view also supports viewing multiple events at once and comparing statistics between them. Here, we've created a graph that compares the average and sum of the property
value from the
Upgrade to Paid event and the count of users who have downgraded to a free account over the past 3 months.
This graph gives us some great information, but it's a bit choppy. If we click the Trailing Average checkbox, it gives us a stronger sense of the trendline.
Go ahead and try graph view out yourself and feel free to reach out to our Customer Success Team with any questions or feedback that you may have!