Skip to main content

Project Stabilization Narrative - September 2025

The Current Situation​

We onboarded a doctor/influencer today who has been eager to use the product for months. She's exactly the kind of person we've held off until the app was more stable. Unfortunately, the experience was extremely painful - she couldn't create an account, I had to give her an Eli lab login, and she still ran into bug after bug. It took a full hour to get her through, and even then the glitches made her results unclear.

As an MD, she said: "That's definitely not the normal reference range" when she saw the glitched visualization. This eroded trust in our scientific accuracy. She also made it clear she couldn't promote the product until stable - otherwise she'd get flooded with DMs acting as unpaid customer support.

The Goal​

September = Month of Stability

By October 7 launch, the app must be:

  • Press-ready
  • Influencer-ready
  • Ready to scale to many more users

Key Problem Areas​

1. User Experience & Onboarding​

The onboarding flow is currently broken at multiple points:

  • Authentication is painful: Forced auto-generated passwords, password manager conflicts, broken reset flows
  • Integrations fail silently: Clue won't sync, Oura gives expired tokens, no way to skip
  • Results don't display: Screen freezes showing "complete" but no results
  • Data visualization is wrong: Wake-up times default incorrectly, points appear in wrong places, reference ranges distort

Why we need to own authentication: We need to really nail this one. We want to make it as simple as possible. That means:

  • Social login (Google and Apple) - eliminate passwords entirely for most users
  • SMS verification - it's quicker than email
  • Browser-based password reset - not forcing users into the app
  • Smooth email verification that opens the app directly
  • Complete control by moving off Firebase to our own system

2. Lack of Visibility​

We're flying blind when issues occur:

  • No distributed tracing from backend to HAE
  • No mobile logs in the cloud
  • No user journey tracking
  • No real-time health checks
  • No automated alerting

3. DevOps Gaps​

Our deployment and operational practices need formalization:

  • No governance structure or standardized processes
  • Dev/staging/production environments aren't identical (some setup was manual, some by code)
  • No progressive rollouts to limit blast radius
  • No automated rollback mechanism
  • No defined release procedures or smoke tests
  • No maintenance windows or user communication channels

4. Integration Reliability​

Third-party integrations are fragile:

  • No monitoring when integrations stop sending data
  • No QA platform to test integrations
  • Users aren't notified when connections are lost
  • No way to know if data stopped coming

5. Quality & Testing​

Insufficient test coverage leaves us vulnerable:

  • Low code coverage on mobile and backend
  • No comprehensive smoke tests
  • Missing integration tests
  • No load testing

6. Customer Support Process Gap​

The customer support process is disconnected from development and it takes time for us to get back to the customer. Normally, we should be able to get very quickly:

  • Disconnected processes: Support process exists but is disconnected from development
  • Slow response times: Takes too long to get back to customers
  • No investigation tools for QA/Support: Support can't see what's happening
  • Developers become support: Every issue requires developer time to debug
  • No visibility tools: Support needs dashboards to see user journeys, errors, integration status

7. QA Platform & Device Simulation Gap​

We need to be able to simulate third-party devices and see how the app behaves:

  • No device simulation: Can't simulate Oura, Fitbit, Apple Watch environments
  • Limited testing devices: Need proper testing devices for real device testing
  • No test environment: Can't test integrations without real devices
  • Manual testing only: No automated way to test various device scenarios

8. Proactive Support & Churn Prevention Gap​

We're reactive, not proactive - we don't know when customers are struggling:

  • No visibility into user struggles: Don't know when tests fail or users give up
  • Can't identify churning users: Someone stops after a failed test - we need to know
  • No proactive outreach: Can't reach out to help struggling users
  • No success tracking: Need to know if users succeeded after issues
  • Missing user journey insights: Support doesn't know exactly what problems customers experienced

The Solution Categories​

Usability & Smoothness​

Fix the painful onboarding experience. Make authentication work. Ensure results display properly. Fix visualization bugs. Remove Firebase dependency with our own login mechanism. Add social login, SMS verification, smooth email flows.

Backend Integration​

Implement OpenTelemetry tracing all the way from backend to HAE. Build queuing system for handling load and failures. Create health checks for all components (database, pub/sub, email, logging).

Alerting & Monitoring​

Push mobile logs to cloud asynchronously. Implement GCP log-based alerting. Build user journey tracing. Create performance dashboards showing system speed. Know immediately when something breaks.

Third-Party Integration & QA Platform​

Build QA platform to simulate third-party devices like Oura, Fitbit, Apple Watch - we need to see how the app behaves with simulated environments. Acquire proper testing devices for real device testing. Monitor when integrations fail or stop sending data. Alert customers when we lose connection with their wearables. Give full visibility into what the customer is doing.

Maintenance & Communication​

Define maintenance windows based on usage patterns. Implement Intercom for real-time user notifications. Set up dev/staging/production Intercom environments. Communicate proactively during issues.

QA System & Testing​

Increase code coverage on mobile and backend. Define proper smoke tests covering all must-haves. Create release procedures defining what tests to run and who tests. Build integration test suites.

DevOps Governance​

Establish governance committee (Thomas, Chip, Lena, Iannick). Define best practices and standards. Ensure dev/staging/production are EXACTLY identical. Implement progressive rollouts. Automate rollbacks.

Customer Support Process​

Define processes for when customers have issues. Establish response time SLAs. Build investigation tools for QA/Support teams. Create dashboards for support visibility. Remove developer dependency for basic troubleshooting.

Proactive Support & Churn Prevention​

Having proper logging will help significantly to know exactly what the user experience is - full visibility into logging and know for a user what errors they reached. When somebody reaches out, support knows exactly what problem the customer experienced. Proactively reach out to struggling users. Identify customers about to churn because of issues - know when someone stops after a failed test vs succeeding. Create platform for customer support to see all of this.

The Risk​

The doctor's warning was clear: broken login flows and password bugs could be enough to "tank your launch." She can't promote until stable - the alternative is becoming unpaid customer support for frustrated users.

The Path Forward​

These aren't nice-to-haves. These are must-haves for October 7. The team is already working tirelessly, including evenings and weekends. We need to execute systematically on these tickets to hit that bar with full confidence.