Spotlight On: Text Standards 

The more popular text-message-based outreach gets, the more important it becomes to collect text data in standardized ways, both for in-cycle reporting, and so tools like contactability models can be built and enhanced for future use in improving the targeting of texting universes. 


Best Practices 


  1. Use an appropriate Contact Type

    • SMS Text, Relational Text, and Distributed Text are all available. 

    • Do not use the type “Phone” for Text-based contact.

    • Check with your text vendor before attempting to use Relational or Distributed Text via API - some vendors may only support SMS Text


  1. Set Canvass Results Standards 

    • Train your team. Make sure the meanings of Results, Codes, Tags, and other text data jargon are clear, not left to individual interpretation.  

    • Set Up for Success. Make sure all critical codes and tags (see below) are pre-mapped, either for the API, or with reusable bulk upload templates


  1. Always Record Critical Canvass Results

    • Apply “Texted” for any text attempt. This data is valuable even if it did not result in a successful contact.

    • Apply Canvassed when: The individual you tried to text was successfully engaged (i.e., responded to your text message). You can then apply relevant Activist Codes or Survey Questions from any resulting conversation. 

    • Apply Opt Outs when: The individual you tried to text should not be attempted again at this number. Not all opt-outs are the same! The following reasons for opting out should always be recorded: 

      • Wrong Number: The individual you tried to text is not at this number. Always map this to VAN Canvass Result: Wrong Number

      • Moved: The individual you texted has moved, so these texts are no longer applicable. Always map this to VAN Canvass Result: Moved

      • Hostile / Do Not Text / GOP: These result codes vary across text platforms, so be sure to map them accordingly. We do not recommend mapping GOP as a canvass result alone, as this is not easily recorded in VAN’s data model. Instead, if the response is GOP but not hostile or a direct opt out, consider applying Canvassed and a Survey Response of Support GOP. 


Other Tips and Tricks


  1. Make Conflicting Results Mutually Exclusive

    • Texters should not be able to tag someone Hostile, then record that they want to Volunteer. Making these two results or tags mutually exclusive lowers the risk of recording incomprehensible data or two data points that cannot be applied to the same volunteer in one day. 


  1. Text Smarter, Not More Often

    • Send texts to targeted lists with a clear purpose. Minimize the risk of opt outs when texts become an irritation to voters and volunteers. 


  1. Use Voters’ First Names in Text Messages

    • Addressing voters by name helps provide confirmation that you reached the intended target. This makes collecting “wrong number” responses possible, and helps make our phone number data cleaner over time if the message in fact was received by someone other than the intended voter.


  1. Minimize Data Loss - Request an API Key

    • Several platforms like Hustle and ThruText offer VAN integrations that remove the need to manually upload text data. 


  1. Ask Your Vendor: How can I access and download my data? 

    • Some text vendors (like Hustle) will provide an S3 bucket with your texting data. This allows the data to be downloaded locally or, conveniently, set up as an automatically scheduled S3 transfer directly into BigQuery. Read more about S3 transfers here. 

    • While newer API integrations and well-trained bulk upload practices have improved the quality of text data stored in VAN, some types of information still don’t fit as well into the VAN data model. Taking the time to set up a recurring upload to BigQuery once will allow increased freedom working with and analysing your data in the future.