Determining the website flow
Developing a content strategy
For this project, I worked to determine the content of the site.
I constructed a skeleton flow of the eligibility questions organized in a flowchart handed off to the frontend developer. I did this by studying existing CDC letter generators.
Besides asking the 7 questions required by the CDC, we needed to consider how we were going to ask the questions in consideration of what we learned about the user.
This flowchart helped to communicate the flow of the website to the developer. We reviewed the flow with lawyers to ensure legal sufficiency.
Knowing how anxious our user group can be in the face of authority, it was important to explain how any personal data collected (e.g. their name) would be used and how it wouldn’t be used.
For example, when asking for the tenant’s address and unit, we should explain this information will be used in the letter so the landlord knows which rental unit is being referred. This was intended to assuage any fears about their volunteered data being used against them.
I also worked to determine the appropriate response choices from the perspective of the user, where a simple yes/no would not be sufficient.
For instance, one requirements asks if tenant has used “best effort to obtain all available government assistance for rent.” If a tenant is not eligible for government assistance (in the case they are undocumented), there should be a third answer choice.
In the above screenshot, the tenant is assured that the information they provide will be used in no other manner except to populate the generated letter.
In addition, we found that answering statement with true/false was more intuitive than asking a respondent to answer Yes/No to a question.
We observed this during user testing and can only speculate why this may be. When responding to complex questions (Q: “Are you unable to pay rent due to loss of household income, extraordinary medical expenses...?” A: “Yes”), it may be unclear to what part of the question the user is responding to.
However, framing the same question as a statement and then having the respondent answer using logic operators was more clear.
(Q: “I am unable to pay rent due to AT LEAST ONE of the following: (a) loss of household income, (b) extraordinary medical expenses, (c)...” A: “One or more of these cases are true”).
We also did what we could to make the site reassuring.
The landing page includes a photo of community leaders in their familiar red shirts. We included language that was emboldening, reminding the tenant of their due rights.
The legal-ese of the original CDC requirements were simplified to be understandable, removing unnecessary legal-ese.
Seeing the community navigators in their familiar red shirts is intended to put a friendly face to the filing process.
Testing and deployment
Evaluating with proxy users
In line with our methods to conduct research using secondary sources, we tested a high-fidelity version of the website with community navigators: Spanish-speaking intermediaries from the community.
We observed how participants moved through the sequence of prompts, taking note of questions or points of confusion.
We presented the website at one of the Office’s staff meetings. We asked everyone go to the website, go through the eligibility determination process, and share ad-hoc feedback.
Findings from proxy user evaluations
- The purpose of the site was not clear. Some did not understand that the site was about the CDC Moratorium or that the end result would be a declaration letter, and not a form of rental assistance. Solution: More clearly describe this on the landing page.
- There was confusion about how to respond to prompts. As mentioned previously, using a Yes/No response framework proved to be difficult to interpret. Solution: Use a True/False response framework.
- There were questions surrounding what constituted as someone’s “best efforts” in trying to pay rent: We needed to make certain phrases more concrete, especially due to the fears of committing perjury. Solution: Include more examples that Atlanta residents could recognize (e.g UnitedWay).
Number of texts sent per month
At launch, the site saw a modest usage of the website, varying between ~10 to ~30 texts (to unique phone numbers).
However, in March, we saw a huge uptick in the number of texts sent. Welcoming Atlanta had just another wave of outreach because a round of promised rental assistance fell through, leaving a lot of folks in the community at major risk of eviction.
Therefore, the use of the website is directly linked to how the Office promoted it.
Update: As August 26, 2021, the Supreme Court has overruled the CDC's authority to enact the moratorium.