Customer complaint emails and other customer voices conveyed through emails often contain valuable, yet underutilized, textual data within organizations.
a) Initial Setup
Zurvey.io facilitates the analysis of such data by providing a dedicated connector for email traffic. To establish an Email Integration datastream, navigate to the relevant input card located at the top of the homepage.
There are two methods for directing email data to Zurvey.io. The first method involves setting up an auto-forward to a unique address. Multiple addresses can be forwarded to your redirect/alias, and all of them will undergo analysis. The alias is unique to a dataset and is always visible on the setup screen.
The second method involves populating your Email Integration datastream through an API. Authorization occurs through your API token (Authorization Scheme: Bearer), which can be accessed either directly from the Email Integration setup screen or via the settings page.
You can insert email records into your dataset by using either the datasetId or emailProcessorId. For more details, refer to our API documentation to learn about request and response samples.
b) Email Parsing
To uphold data quality standards, Zurvey.io conducts preprocessing on incoming emails before subjecting them to text analysis. In this parsing process, the system automatically extracts the most recent email from the thread and separates and removes the email signature from the email body. Zurvey.io employs a parsing method rooted in widely accepted posting style policies.
By default, parsing is enabled for all created Email Integration datasets. However, if the data structure demands it, parsing can be switched off to accommodate specific requirements.
c) Meta Fields
When utilizing auto-forwarding to send emails to Zurvey.io, you can include additional information, such as metadata indicating the direction of emails (incoming or outgoing). This metadata serves as a filterable category on the dashboard and is included in all exported data.
For accurate parsing of this information from email headers, it is essential to define these fields on the interface during the dataset setup process. This step ensures that the metadata is appropriately recognized and processed.
If you opt to use our API for sending MIME emails, it is crucial to set these fields in the HTTP Request header with the prefix X-Zurvey- (eg; if your metafield is direction it should be sent in the header as X-Zurvey-direction). This prefix is needed to differentiate Zurvey.io metafields from other settings in the header such as content type or subject.
d) Analysis Settings
After acquiring all the necessary information for forwarding emails to the dataset and, if needed, defining the relevant meta fields, the next step is to guide Zurvey.io on how to interpret your data effectively. This is where the Analysis Settings come into play.
Within this section, you have the flexibility to:
- Change the name of your API dataset: tailor the dataset name to enhance clarity and relevance on the Zurvey.io platform.
- Modify the analysis language: specify the language for text data analysis, ensuring accurate and meaningful insights.
- Adjust accented character handling: customize how Zurvey.io processes accented characters in your data.
- Manage lowercase characters: define preferences for handling lowercase characters in your dataset.
- Toggle automatic topic recognition: enable or disable the automatic recognition of topics based on your analytical needs.
- Modify text splitting style: tailor the style of text splitting to align with your data's characteristics.
e) Custom Label Analysis
To enhance the precision of text analysis, Zurvey.io offers the functionality to apply Custom Labels during the analysis of your email data. By default, both the email body and subject undergo text analysis.
To incorporate available Custom Labels into the analysis, navigate to the Custom Labels section and select the specific Custom Labels you wish to apply.
For a deeper understanding of Custom Labels and their integration into our text analysis methodology, we recommend referring to our dedicated documentation.
f) Editing & Reanalysis
If you need to make adjustments to your Email integration datastream, such as:
- Adding new meta fields
- Switching on or switching off parsing
- Changing Analysis Settings
- Modifying assigned Custom Labels
You can edit your previously saved and populated dataset and reanalyze the already processed data.
Upon making your changes and clicking the "Save Settings" button at the bottom, a popup will prompt you to decide whether you would like to apply your changes only to the newly incoming data or reanalyze the entire dataset.
If you've modified Parsing or Analysis Settings (including analysis language, lowercase character handling, accented character handling, or text splitting style) you will have the flexibility to choose whether to apply changes for new data only, reanalyze the entire dataset, or reanalyze data from a specific set date (with the default set date being the creation date of the dataset).
