Problem Statements
Patients previously had two ways for requesting to opt out of messages from Artera: by sending a stop keyword (such as STOP or OPTOUT), or by asking a clinic staff member to opt them out manually. The first of these options was previously controlled by the third party vendor we used to deliver SMS messages, which meant that within Artera we had two separate opt out settings, which was confusing and lead to human errors. It also meant we were very limited in terms of the control and data we had related to keyword opt outs, so the patient-facing workflow was not specific to a healthcare setting and led to confusion and patients accidentally opting out of more than they intended.
We needed to bring the entire opt out workflow in-house in order to simplify our UX for our staff users, improve transparency and confidence for our patients, and to enable the development of more advanced preferences features.
Assessment
I spoke to dozens of clients and internal stakeholders to understand how patients are indicating their preferences, and how hospitals are honoring those requests.
The first thing that I noticed was that the opt out workflow lacked transparency between the patients and staff. Patients didn’t actually know what they were asking to opt out of; for example, they don’t realize that opting out of low-value messages like marketing messages will result in them also being opted out of high value messages like appointment reminders. And staff can’t tell what the patient is actually intending to do when they opted out, and based on conservative compliance requirements, many would automatically interpret an opt out request as a global opt out. In both cases, because the patient can’t tell what they’re doing and the staff can’t tell what they’re trying to do, the patient ends up opted out of more than they intended, meaning patients are missing important messages from their healthcare teams, including messages they may still be expecting.
My hypothesis is that if we can improve the transparency so that patients understand what they’re doing when they change their preferences, they will opt out at the least restrictive level naturally, and staff will no longer need to assume the most conservative interpretation. In both cases, more patients remain opted in to receive more messages.
The second thing I noticed was that hospital staff generally do not view patient opt outs as a valuable piece of data that they can use to improve their patient outreach. Some have an anecdotal understanding of this, for example saying things like “patients are more likely to opt out of generic campaigns than appointment reminders.” However none seem to be taking this to the next step, which is inferring the value patients get out of their messages and optimizing their patient outreach accordingly.
Our clients look to Artera to provide a consultative role in their patients outreach, and patient opt outs are currently an underutilized opportunity to use their data to make proactive recommendations for improving their outreach campaigns.
Actions
Based on the findings above, I implemented the following initiatives:
- Collaborated with data sciences team to build dashboards based on patient opt outs that makes recommendations for campaign optimization
- Spearheaded the work to bring the patient opt in workflow completely in house
- Managed the relationship with the third party vendor through the process of migrating opt outs away from their management
- Designed, iterated and launched a new in-house patient opt outs workflow
- Designed, iterated, launched new patient-facing opt out workflow
- Designed and iterated on new patient-facing preferences portal
Documents
Patient Preferences Domain Vision Strategy
A document for senior leadership to understand the vision and roadmap of the opt outs domain
This document has some sensitive info removed
Outcomes
- Effectively sunset the third party tool
- Decreased accidental patient opt outs
- Developed a roadmap for granular patient preferences, including category-level preferences
Alyson Gagne 