A/B testing outline for UTM usage, GA4, and data storage requirements

[depietrodeanna]

See Michelle's comment below for finalized requirements.

/////

Documenting decisions made after conversation with Tony, Michelle, and Deanna around A/B testing and data storage/functionality of UTMs:

Main aim: Determine how many recruits went to MyConnect from each A/B testing campaign, and how many people from each campaign clicked the create account button. And develop reporting mechanism for sites. MVP requirement- GA4 needs to capture the site campaign information (from UTMs)

Hana looking into tracking clicks to "Create Account" on MyConnect using GA4.

To get at traffic to MyConnect, we will use UTMs on tokenized MyConnect URLs, which are individual-level de-identified URLs that we will store completely.

We will store the details of the UTM (3 components- site, source, and campaign) as a string called something like, “UTM Campaign." The way we will determine what the 3 components map to is by referencing the site spreadsheets. Sites will use the spreadsheet template developed by HFH/UC to organize their campaign information.

Once a recruit logs into MyConnect with their token / PIN, we pull in more de-identified information (demographics, etc.) that site pushes us from the EMR.

Sites already have the consent information and cross reference those data with the data available through the Get Participant Data API (which will contain the campaigns/UTMs in their nested structure).

It will take approx one release cycle for dev team to do their work to set this up on our end.

Decision needed: timeline/release cycle this work will fall into (December?)

@anthonypetersen and @brotzmanmj please fill in/correct any inaccuracies in these requirements.
 
 

[brotzmanmj]

Thanks Deanna. I would clarify a few things. We have two aims.

1. Capture the UTMs created and sent by the sites as part of active recruitment for recruit when recruit FIRST signs in to MyConnect with a valid token or PIN. The sites will grab these data via the getParticipantData API in order to analyze campaign effectiveness.
   The UTM will be stored in three parts so we will need three new variables:
   UTM_Source (which is equivalent to site; which site is the UTM from)
   UTM_Medium (what material was the UTM embedded in: recruitment letter, email, SMS, MyChart invitation, etc)
   UTM_Campaign (a brief descriptor of the campaign)
   Here is an example of data:
   UTM_Source = HenryFord, UTM_Medium = email, UTM_Campaign = 2024-example-campaign
   Variable properties for @mnataraj92: Each of these three variables will be a string variable. Not required. Not default variables. Not PII. Not retained as part of stub record upon data destruction.

Programming task for CCC: Upon first sign in (MyConnect account creation) with a valid token or PIN, take the UTM information and store it in these three variables in the Participants table associated with the token or PIN, and make available in the API.
Programming task for site: Test pulling these 3 variables via the API.

Because the above can only be stored and linked to the participant upon creating an account, we cannot use it to evaluate the step before that, namely what campaigns drive people to MyConnect, regardless of whether the recruit creates an account or not. This leads to need #2 not addressed by programming in this issue.
2) @HanaShiho evaluate if google analytics for MyConnect stores the UTM information we described above. And if so, is it possible to export it somehow (e.g. in CSV files we could put on Box, so the sites can access it and evaluate their campaign traffic with it).