MixPanel Walkthrough

Prerequisites

• Mixpanel Connection

Mixpanel reports in Rivery

Rivery provides the reports, which are available in the Mixpanel reporting dashboard.

Using Rivery, it is possible to schedule the reports so your data will be always up to date and ready for your target tables for further analysis.

Start to build your Mixpanel river from selecting the mixpanel report in the report list:

Export Raw Data

The raw data export will provide all the events that happened during the required time range for the given event type.

1. Select the Start date to start pulling events from. Leave the end date empty in order to pull events until today.

   

Please note:
The Start Date won't be advanced if a River run is unsuccessful.
If you don't want this default setting, click More Options and check the box to advance the start date even if the River run is unsuccessful (Not recommended).

1. Mixpanel supports in filtering only by date level and not by the time. It means that it is not possible to take only a part of day or hour but only a full day.
2. Use last days back in order to pull data from a given number of days back before the start date. Usually it will be used in cases when there are retroactive changes.

2. Select the event types to pull the data for.

   

   1. When clicking the Events input, it will show up a list of all the top events in the account, which means it will be possible to select the events from that list. It is also possible to add additional events not from that list.
   2. Leave the events empty in order to pull all events. Please note that usually it will be a very large amount of data, and might cause an issue of multi-type columns (will be specified later in this document).
   3. It is not recommended to pull all events in the same River. Best practice is pulling business-related events under the same river.

3. Select the data intervals:

   

   1. Since Mixpanel might provide a large amount of data, the required time range might be too much for Mixpanel API to handle (the Mixpanel service might returns an error). Therefore, Rivery pulls the data from Mixpanel in time intervals. The default is daily (will send a request for every single day).
   2. In the case of pulling raw data for a long time range (in the case of migration etc), it is possible to enlarge the intervals in order to reduce the number of requests.
   3. In the case you know there are millions of events under your account for the given time range and the events types that were provided in the river, it is required to pull the data in 1-2 days intervals (otherwise there will be a risk to receive an error from Mixpanel).

4. Select the number of event to pull in one API call

   

   1. Again, in order to save the river from getting an error from Mixpanel, Rivery will split the request per number of events. That’s the way the amount of data in each API call will be smaller and easier for Mixpanel to handle.
   2. In the case of pulling more than one event, Rivery will split the according to this number. It does not affect the results by any mean.
   3. In the case you know there isn’t too much data under your Mixpanel account, pull the data in 10 events per API call.

5. Input Convert column to String:

   1. Those two inputs support a situation when the data source have some multi-type issue. Multi-type is a situation when some column appears in the data in type X and also in type Y (for example, field “grade” will be in some record 6 and in another record “A”, meaning sometimes it’s a String and sometimes it’s an Integer).
   2. Rivery will automatically detect those issues during the mapping process in step 3 in the river. Therefore, when initially creating the river you can leave those inputs empty until the mapping step is over.
   3. After clicking on Auto mapping in step 3:

   

   Rivery might present the following message regarding the multi-type issue:

   

   4. In this message, you will find all the columns that Rivery did find with this issue. You can either ignore those columns during the upload, and then those columns won’t be found in the target table, or convert those columns to String.

   5. Ignoring the columns is recommended in case those columns are not required in the target table.

   6. Converting the columns to String will change all the data in this column to String (so a number like 5 will become “5”). That’s the way it will be possible to load this column to the target table, no matter what its type.

   7. Any converted column will be automatically added to the input “convert columns to string” and to the mapping table with type String.

   8. When removing columns from the input “convert columns to a string”, Rivery will ask you if you want to remove the column from the mapping (and then it will be ignored during the loading to the target table) or leave it there with String type. Please notice that if you’ll leave it there and this column will contain some data which is not String, the river will fail during the upload.

   9. Important: There might be a situation that Rivery won’t find all the multi-type columns in the data source (since the mapping process works on a sample of the data source). Therefore, when trying to load the data into the target table Rivery will return an error saying it is not possible to convert some value in some column to some type.

      1. If this column is not required in the target table, you can look for it in the mapping and remove it. Then, go the Activities screen in Rivery and click on Retry on the failed activity of this river. This will only try to load the data again to the target table. This the best solution for that case.
      2. If this column is required, add it to the input of “convert columns to string”, and then run the river again -It is important to run the river from its screen , so it will pull the data again from Mixpanel, and not only retry to load the data.

   10. Click on “Stringify all columns” in order to convert all the columns in the data source to String. That way, the river will never encounter the data type issue.

       1. Please note that selecting this option will update all the mapping to String.
       2. This action is irreversible. In order to roll it back, it will be necessary to re-run the mapping again.

Retention report

This report will provide the retention data in 3 different types: Recurring, First time and Addiction.

1. Select the retention report
   

2. Select the Start date to start pulling events from. Leave the end date empty in order to pull events until today.

   1. Use “days back” in order to pull data from a given number of days back before the start date. Can be useful in order to have retention report of some given last days.
   2. Hard start date - checking it will cause the river to always run from the given start date. Otherwise, it will be updated after the next run of the river.

3. Select the events for the Retention report. In the case of First time Retention report, it is possible to select an additional event.

   
   When clicking on the event input it will present a list of all available events.

4. Click on ‘add condition’ in order to filter the results by events or user properties.

   

5. Select a segment for the Retention report in the “On” input (this input will be visible only after selecting an event).

   1. When clicking on the input, a list with all the available properties of the event will be shown.
   2. Please notice that when pulling a Retention report with a segment, the results won’t contain a date column. (However, it is possible to add a date column in the mapping which will contain the current date).

6. Select an interval count for the retention report - how many days back the report will contain. Leaving it empty will result in 2 days only.

   

7. Select a unit for the Retention report - decides the level of granularity. Days will return the Retention report in days and weeks/months will return the Retention report in weeks/months intervals.

   

Segmentation report

This report will provide a segmentation report with zero, one or two segments.

1. Select the Start date to start pulling events from. Leave the end date empty in order to pull events until today.

   

   1. Use “days back” in order to pull data from a given number of days back before the start date. Can be useful in order to have retention report of some given last days.
   2. Hard start date - checking will cause the river to always run from the given start date. Otherwise, it will be updated after the next run of the river.

2. Select the event for the Segmentation report.

   

   When clicking on the event input it will present a list of all available events.

3. Click on ‘add condition’ in order to filter the results by events or user properties.

   

4. Select segments for the Segmentation report in the “On” input (this input will be visible only after selecting an event).

   

   When clicking on the input, a list of all the available properties of the event will be shown.

5. Select the level of granularity in the Segmentation report. Each level will change the granularity of the time in the results (for example week will result as every row is a week and month will result as every row is a month).

6. Select the analysis type of the Segmentation report. This input determines the nature of the count in the report. General is the total count.

   

Funnels report

1. Select the Start date to start pulling events from. Leave the end date empty in order to pull events until today.

   

   a. Use “days back” in order to pull data from a given number of days back before the start date. Can be useful in order to have retention report of some given last days.
   b. Hard start date - checking it will cause the river to always run from the given start date. Otherwise, it will be updated after the next run of the river.

2. Select the Funnel to have the results for. When clicking on the input a list with all the Funnels in the account will be shown.

   

3. Click on “add overview condition” in order to filter the results by some properties. Please note that those conditions will be effective in an overview look of the funnel (and not for a specific event that composes the funnel).

4. Select segments for the Funnel report in the “On” input.

   

   When clicking on the input, a list of all the available properties of the event will be  shown.

5. Select the length unit and the unit of the funnel report. The default is the settings of the original funnel in Mixpanel.

   

6. Time resolution - Determines the resolution of the results.

   

   1. Results of the funnels report is a record for each event that composes the funnel and time range (columns date and to_date in the results).
   2. The summary will result in a record for each event in the funnel for all the time range. Any other resolution will result in a record for an event and the selected time resolution.
   3. For example, for a funnel with two events A & B, the results of a funnel report for 14 days will result in 2 rows in case of summary resolution (2 rows for each event that contains the all-time range), and in case of daily resolution the results will contain 28 rows (1 row for each event and day). Please note that in order to refer to unique metrics it is important to pull the funnels report in a Summary resolution. T he results in the Mixpanel UI are only in Summary resolution.

Profiles report

This report provides all the users in the mixpanel account.

The results will contain all the users in the Mixpanel account at the time of running the report.

Input Stringify lists - The results contains columns which are lists. In order to have those lists in String types in the target table, select this checkbox.

Reports: Events, Events Properties

General reports that provide a summary analysis of the selected events and properties.
The results contain the count of the event/property per each day in the time range.

Reports: Events top, Events Properties top, Event property values, User properties, User property values.

Those reports provide look-up tables that contain the data in the Mixpanel account.