The past shopping behavior data collected by Google Analytics is just one lens we can use to predict the likelihood of a product being sold in the future. Take advantage of the full breadth of the information you have at your disposal by importing additional product, user, or query data from third-party systems such as your PIM, CMS, ERP, or others, and incorporating it into your ranking strategy.

Below are instructions for importing and managing your account's custom data.

Table of Contents

• Repositories
• Datasets
• Importing Data
• Exporting Data
• Clearing Data
• Aborting an Action
• Using Custom Data

 

Repositories

Repositories are used to help organize the custom data in your account. You can think of repositories as folders; if you have lots of data from different sources, it may be helpful to set up different repositories to keep track of what data belongs where. An account can have an unlimited number of repositories.

 

Creating a Data Repository

1. Navigate to the Repositories page.
2. Click Add Repository.
3. Enter a unique, identifiable Name.
4. Click Save.

Once a repository has been created, you can change its name or delete it by navigating to the Repository Settings page by clicking on it in your Repositories table. 

Deleting a data repository will delete all data within the repository, so be careful.

 

Datasets

Datasets contain the product data imported to Convert. To import custom data, you must first create a dataset.

Datasets exist within a repository, and for organizational purposes, an unlimited number of datasets can be created within each repository. Once created, datasets cannot be moved between repositories, so make sure you've thought through your data hierarchy before creating a dataset.

Creating a Dataset

1. Navigate to the Datasets page.
2. Click Add Dataset.
   1. If you haven't yet created a repository, you'll need to do so first.
3. Enter a unique, identifiable name, select the repository you'd like the dataset to live in, and click Save.

   Dataset names cannot include spaces.

 

Importing Data

Before you import data, please be aware that for your import to be successful, the data you're uploading will need to be formatted properly. The guidelines are as follows:

• The import file must be a properly formatted .CSV
• At least one column of the spreadsheet must contain the product identifier used by Google Analytics
  • The column header for the product identifier must be labeled id
• An alphanumeric column header must be included for any data that should be imported
  • Column headers cannot begin with underscores or dollar signs
  • Spaces can be included in column headers, but Convert will automatically trim any leading or trailing spaces
  • Duplicate column headers will be rejected. Please note that column headers are not case sensitive, and so columns labeled "ID" and "id" will be treated as duplicates.
• If you're updating data in an existing dataset, column headers must mirror the existing headers
  • Please note that column headers are not case sensitive.
  • Columns can appear in any order, as long as the headers match
• All data with a valid column header and product identifier will be imported

Once your data is properly formatted, follow these steps:

1. Navigate to the Import/Export tab of the dataset you want to import into.
2. Click the Import Data button.
3. Select your desired Import Type.
   1. Merge will cause any new data to be added to the dataset, and replace any data where the product identifier and column header match an existing pair.
   2. Replace will erase all existing data and replace it with the import file.
4. Select the desired configuration for Notify on Completion and Notification Emails.
   1. Depending on the amount of data in your upload file, importing data can take anywhere from a couple of seconds to several hours. Enabling this feature will cause a notification email to be sent to any emails entered when the import is complete.
5. Click Choose File, select the upload file, and click Open.
6. Confirm the desired import settings, and click Import Data.
7. You can then track your import's status on the dataset's Import/Export page.

   If your import fails, click View Details to navigate to the Job Details page, then download the Error Log to see what went wrong.

 

If you plan on incorporating your custom data into rank models, it should be imported in a form that will allow it to be successfully digested by machine learning models. If you don't have a data science team that can perform this action for you, we can help!

 

Exporting Data

If you'd like to see the data currently in your dataset, you can export it. This is also an easy way to ensure that you're using the proper template when seeking to update existing data.

1. Navigate to the Import/Export tab of the dataset you want to export.
2. Click the Export Data button.
3. Select the desired configuration for Notify on Completion and Notification Emails.
   1. Depending on the amount of data in your dataset, exporting data can take anywhere from a couple of seconds to several hours. Enabling this feature will cause a notification email to be sent to any emails entered when the export is complete.
4. The export operation will appear in the dataset's Import/Export table. Once it has finished, click the date & time of the export job.
5. On the Job Details page, click Download next to Export File. The dataset will download in .CSV format. 

 

Clearing Data

If you'd like to prevent existing data from impacting ranking models without deleting the dataset or undoing any configuration work you may have done on existing model features, you can clear it. This will all delete all data in the dataset.

1. Navigate to the Import/Export tab of the dataset you want to clear.
2. Click the Clear Data button.
3. Select the desired configuration for Notify on Completion and Notification Emails.
   1. Depending on the amount of data in your dataset, clearing data can take anywhere from a couple of seconds to several hours. Enabling this feature will cause a notification email to be sent to any emails entered when the clear is complete.
4. You can then track your clear's status on the dataset's Import/Export page.

 

Aborting an Action

Once you've kicked off an import, export, or clear action on a dataset, you will not be able to complete any other actions on the dataset until the current job is complete. However, if you've kicked off an operation in error and need to take a different action, you can simply abort the current job.

1. Navigate to the Import/Export page for the dataset.
2. In the top row of the table, you should see the current operation along with its status.
3. Click the Abort button next to the current status to abort the job.

   If you abort an import operation after it has already begun, any data imported up to that point will remain in the dataset.

   Note: If you abort a clear operation after it has already begun, any data cleared up to that point will no longer be present in the dataset.

 

 

Using Custom Data

Once you've imported custom data, it can be incorporated into rank models, used to build sophisticated solution targeting and override rules, or both. First, you'll need to define the fields that you want to make available to your rank models and solutions. These attributes or metrics are called model features.

For more information on creating Model Features, click here.

For more information on incorporating Model Features into Rank Models, click here.

For more information on using custom Model Features for solution targeting and override rules, click here.