What are Datasets?
Datasets let you reuse your data prep work across as many metrics as you want. In practice, this lets you save one or more reports with their transforms and joins, and reuse that data across multiple metrics.
Datasets help overcome several issues that arise as you create a data-driven culture in your company:
|Poor Data Organization||Using Grow's Datasets|
|Get conflicting numbers within your company due to different individuals or departments calculating the numbers in different ways.||Create a single source of truth for your company's data, with definitions for your data that everyone agrees with.|
|Spend a lot of time creating reports and adding the transforms for each metric, especially when some of them use much of the same data.||Separate data-prep from the metric-building by preparing the data once and then using it across multiple metrics.|
|Something changes in how you define your data and you have to add a new filter or other transform to a group of similar metrics ONE. AT. A. TIME. (taking a lot of your time and increasing the chance of user error).||Edit the dataset once, and all of the metrics built from it will now use the current data.|
|Each person in your company makes their own reports despite varying skills and knowledge of good data practices.||Leverage those in your company who are more skilled in data practices or more familiar with the data to create datasets that everyone else can use.|
For example, you advertise on Google Ads and Facebook Ads, and you track the spend from some local advertising in a spreadsheet in Dropbox. You can pull in all three reports, add any transforms, join them together in a master report, and save that as a dataset. Now each time you want to use your advertising spend or performance in a metric, you can use that dataset along with any other reports to create your new chart. What's even better is that if you decide to filter out a specific campaign from Google Ads, you can do that once in the dataset and it will apply to all of the metrics using that dataset.
You can add transforms at the dataset level to clean up your data to create a general purpose report, and then you can further transform the data from that dataset when you go to create a metric.
Datasets are not required when creating a metric, for example if you are just making one metric with that data, or if the data you're using is temporary.
How to Create Datasets
There are 2 ways to create datasets. You can create a dataset from the dashboard or from the Metric Builder. If you are planning on creating a dataset and then move right to building a metric with that dataset, start in the Metric Builder.
From the Dashboard
On the dashboard, click on the blue "Add New" button at the top right, then select "Dataset".
Then select one of your warehouse tables, data source connections, or existing datasets to add a report. This will open the Dataset Builder, which is just like the Metric Builder but without the Chart tab. Here you can add any transforms or joined reports to define your data and create a robust dataset.
From the Metric Builder
If you are working with in the Metric Builder and you want to save your data as a dataset, click the 'Create Dataset' button on the far right of the Data tab. This will save the reports and transforms currently being used.
NOTE: This will convert your reports into a Dataset. There is no undo. All your data and work is saved, but to make any changes you need to edit the dataset or add additional transforms for the metric.
A window will prompt you to give the new dataset a name, description, and select the Primary Report. Fill it out and hit 'Create'.
Using a Primary Report
When you save a dataset, you will select one of the reports used as a "Primary Report." This means that as you use that dataset in a new metric, it will only bring in that one report instead of all of the reports used in the dataset. If there is only one report used in a dataset, that one will be automatically designated as the a primary report.
It is most common to use a primary report when you have joined reports. For example, if you are tracking marketing spend from different data sources, you could bring in each report into a dataset, join them all together into one master report, and save that last report as the primary report since it contains all of the data and you don't need to use the other reports separately.
Selecting a Primary Report
- Create a new dataset or edit an existing one in the Dataset Builder.
- At the bottom left, under "Settings," select "Primary Report."
- Select the one report you want to use.
- Click "Save" at the top right.
How to create metrics using Datasets
It is simple to create a metric using a dataset. In the normal Add Metric flow, pick the dataset you are interested in. Click 'Get Data', and start to build your metric.
If your dataset has multiple reports in it, you can remove any of the reports after bringing it into the metric builder without affecting the other reports.
Dataset Refresh Rate
If you have a dataset that will be pulling in a lot of data and you want to build several metrics from it, you should set a refresh rate for your dataset. This will help avoid any API rate limit issues or slowing down the system if multiple metrics are trying to get the dataset to update all at once. Instead you can put the control of the data on the dataset so it gets the latest data as often as it needs, then feeds that to the metrics using that data.
You should set a specific dataset refresh rate when:
- Your data has more than 10K rows
- It takes over 30 seconds to pull your data
- Many metrics use that dataset
- Your data doesn't have to be up-to-the-minute. (For example, if your database updates once a day, you can set the refresh rate to every 12 or 24 hours.)
To set the refresh rate for a dataset, go to Data (at the top left) > Datasets > then click on the name of the dataset you want to edit. In the dataset settings page you can adjust how often Grow will refresh the data in that dataset.
To see all of the Dataset on your account, go to the Data tab at the top global navigation, and select "Datasets." Here you can also see what data sources were used, who created it, and how many metrics and dashboards use each dataset.
To see the specific metrics that use a certain dataset and who made them, click on the Actions menu at the far right and select "Details," or click on the dataset name.
The other options in the Actions menu, and on the specific Dataset Settings page, let you create a new metric, edit the dataset in the dataset builder, make a copy of the dataset, verify the dataset, share the dataset with other users in your account, and delete the dataset (which will also remove metrics that use that dataset).
Deleting a Dataset
If you delete a dataset, it will delete any metrics that are built off of that dataset as well. You will be asked to confirm before any datasets are deleted.
If you accidentally delete a dataset, you can restore it for up to two weeks after. Contact support to restore a dataset or a metric.