How might we increase accuracy perception and compliance (...and trust)?

Worked on quick, simple ways to improve the algorithm's accuracy to get users more engaged with the bracelet and app. Focused on solutions that made the biggest impact with the least effort, keeping in mind the project's limited resources.


Company
Ava Women
https://www.avawomen.com/ 

My role
Product Designer  

 When
June - July 2019


Context




The company
Ava is a women’s health company based in Zurich, San Francisco, Makati and Belgrade. The company's initial product consists of a sophisticated sensor bracelet, an app and a powerful backend with self-learning algorithms for interpretation of hormonal changes, accurate ovulation tracking and screening of women's health issues.
This is a semplified explanation on how Ava works 👇🏽
 

User Problem
Thanks to an NPS survey we found out that Ava users were complaining about accuracy (13 % after 1 month and 18.8% after 3 months).




  1. How might we increase accuracy perception and compliance?
  2. How might we improve the signal quality that reach the algorithm therefore increase trust?



Understanding and brainstorming

Together with product managers, we took a step back and looked at the entire user journey, asking key questions at every stage:

User Expectations & Understanding
  • Is the bracelet’s functionality explained clearly?
  • Does the user have the right expectations?

Wearing the Bracelet
  • Is it worn consistently (same hand, same tightness) every night?
  • Is it charged and ready to record?

Data Recording
  • Are the sensors capturing accurate data?
  • Is any important information lost during storage?

Data Transfer
  • Are the correct results being transferred?
  • Can the user sync their data seamlessly?

App Display & Performance
  • Are the correct results displayed?
  • How fast is the system’s response?

User Interpretation
  • Are the results easy to understand?
  • Do users interpret them correctly?
  • Are there complaints or confusion?

Armed with these questions, we engaged with users directly and sent out a survey to gather more insights, ensuring our next steps were data-driven.

One of the biggest finding we had was that users were forgetting to wear the bracelet:





Solutions



To address the problem, we implemented a mix of hardware and software solutions. 
Here’s a breakdown of the key changes:

1. Hardware Improvements
Strap: Replaced the loop with the “nose” with a regular loop to make it easier to put on the bracelet.

2. Printed Guide (IFU) Updates
Added clear instructions on how to wear the bracelet (i.e. below the wrist bone, at the same strap notch every night).

3. Software Enhancements
Small Tweaks:
Added an LED explanation in the app to help users understand hardware signals more easily.

Bigger changes:

  • Compliance messages on the dashboard during the cycle
    Depending on the cycle day (CD) and on how much the user syncs, she will get a different message on the dashboard to increase engagment, sync more and therefore improve the algo, which then leads to a better accuracy.


  • Reminder / Push notification to wear Ava at selected time:
    We introduced the possibility for our users to receive push notification as a reminder to put on the bracelet before going to sleep.







Results



Change is a constant in a startup, and unfortunately, after the release, all available resources were redirected to other projects, making it difficult to properly track the impact of our work. 

However, we did see positive signs—our users actively engaged with and shared their excitement about the new features in our Facebook community. While we couldn’t monitor performance as closely as we wanted, the enthusiastic feedback reassured us that the improvements were well received.