Skip to main content

Engagement Metrics 2026

Status: Implementation Complete - Pending Stakeholder Review Stakeholder: Lena Implementation: eli-kpi

Assumptions to Verify with Lena​

Critical Assumptions

These assumptions directly affect the accuracy of all metrics. Please confirm or correct.

1. Tests Per Pack​

AssumptionValueSource
Tests per pack4 testsShopify product descriptions

Evidence:

  • "12 Cortisol tests" product description: "3 packs of 4 Cortisol tests"
  • "Cortisol Pack - One-time Purchase" description: "4 tests"
  • "Cortisol Pack - Subscription" description: "4 tests"

Question for Lena: Is 4 tests per pack correct for all product variants?

2. Subscription Quantity Data​

The Loop subscription data includes a quantity field that varies by subscription:

Plan TypeMost Common QuantityTotal TestsTests/Month (Amortized)
Monthly (1mo)1 pack44.0
Monthly (1mo)3 packs1212.0
6-month prepaid3 packs122.0
6-month prepaid6 packs244.0
12-month prepaid3 packs121.0
12-month prepaid9 packs363.0

Current Implementation: Uses actual quantity from each subscription, calculates:

tests_per_month = (quantity × 4) / interval_months

Question for Lena: Does this formula accurately represent how many tests a user should be using per month?

3. One-Hour Deduplication Window​

When a user scans a test:

  1. The test takes ~20 minutes to develop
  2. Users may scan multiple times during this window
  3. All scans within the same hour are counted as ONE test attempt

Question for Lena: Is a 1-hour window appropriate, or should it be different (30 min, 2 hours)?

4. Eligibility Window for WAU/MAU​

Currently using 3-month eligibility window (not 6 months):

  • A user is "eligible" if they made a purchase within the last 90 days
  • This affects the denominator for WAU and MAU calculations

Question for Lena: Confirm 3-month window is correct. Should we also track 6-month for comparison?

5. Subscription Types Included​

Currently counting:

  • Loop subscriptions (all types: monthly, 6-month prepaid, 12-month prepaid)

NOT currently counting:

  • Shopify one-time purchases
  • Partner samples
  • Replacement orders

Question for Lena: Should one-time Shopify purchases be included in "tests delivered"?

6. Internal Account Exclusion​

Excluding all accounts with @eli.health email addresses.

Question for Lena: Are there other internal/test accounts to exclude?

Metric Definitions​

1. Monthly Test Completion Rate​

Formula:

Completion Rate = Successful Tests / Tests Delivered × 100

Current Results (as of Jan 24, 2026):

MonthSuccessful TestsTests DeliveredCompletion Rate
January 20263,94817,81022.2%
December 20253,34015,27621.9%

Interpretation: About 22% of the tests we're "delivering" (amortized) result in successful completions.

2. Weekly Active Users (WAU)​

Formula:

WAU = Users with ≥1 completed test this week / Eligible users (purchased in last 90 days)

3. MAU Standard​

Formula:

MAU Standard = Users with ≥1 completed test this month / Eligible users

4. MAU Power Users​

Formula:

MAU Power = Users with ≥4 completed tests this month / Eligible users

Why 4 tests? One pack = 4 tests = expected monthly usage for fully engaged users.

Data Sources​

Key Joins​

  1. Loop → Firebase: Match by email address
  2. Firebase → App User: Match by uid = accountId
  3. App User → Readings: Match by userId

SQL Implementation​

Location: eli-kpi/src/sql/bigquery/loopwork/

MetricSQL File
Test Completion Ratemonthly-test-completion-rate-delivered.sql
WAUweekly-engagement-rate-2026.sql
MAU Standardmau-standard-2026.sql
MAU Power Usersmau-power-users-2026.sql

API Endpoints​

EndpointDescription
GET /api/engagement-2026/test-completion-rateMonthly completion rate
GET /api/engagement-2026/wauWeekly active users
GET /api/engagement-2026/mau-standardMonthly active users
GET /api/engagement-2026/mau-power-usersPower users

Questions Pending Resolution​

  1. Quantity verification: Does every subscription have the correct quantity in Loop? Are there data quality issues?

  2. Prepaid delivery timing: For prepaid plans, all tests ship at once but we amortize for calculation. Is this the right approach?

  3. Cancelled subscriptions: Should we count tests as "delivered" for the months before cancellation?

  4. Subscription renewals: For monthly subscriptions, do they get new tests each month (which we should count)?

  5. Historical backfill: Should we calculate these metrics historically, or only going forward?

Revision History​

DateAuthorChange
2026-01-24ChipInitial implementation with quantity-based amortization