Adoption & Engagement Analysis

Accounts wanted to know whether their program was working. The honest answer required separating the numbers that sound like success from the numbers that are success, then saying which levers would move the second set.

Every account had access to the same dashboards and reached different conclusions. Adoption was being reported as installs, retention as a single blended percentage, and referral as a count of invites sent. None of those measures told anyone what to do differently on Monday.

• Separate registration, install, sign in, and active use. They are four different problems with four different owners.
• Read retention as weekly cohorts. An average hides the fact that one cohort collapsed and another held.
• Instrument growth as a loop with stages, not as a single number. Invites sent means nothing without activation and conversion behind it.
• Watch which screens people actually use. The event mix tells you what the product is for, which is often not what it was designed for.
• Compare against a known standard. Telling an account they are at a certain usage rate is information. Telling them the standard is higher, and here is the plan to close it, is a decision.
• Time-bound campaigns and challenges are the cleanest natural experiments available. Measure them properly and they answer questions no dashboard will.

1. 01

   An adoption funnel definition

   A shared set of stages from registered through weekly active, applied identically across accounts so results were comparable.

2. 02

   A cohort retention read

   Weekly joining cohorts tracked across their first several weeks, replacing a single blended retention figure.

3. 03

   A growth loop model

   Referral and trial pass activity measured as a staged loop from invite through activation, visit, and conversion, with an owner at each stage.

4. 04

   An engagement program read

   Challenges, campaigns, and notification activity evaluated on participation and follow-through rather than on send volume.

5. 05

   A recommendation set per account

   Each analysis closed with specific moves: signage, staff training, notification cadence, or program design, matched to whichever stage of the funnel was actually leaking.

1. 01

   Report the usage rate, always

   It was the least flattering number available and the only one that consistently produced action.

2. 02

   Attribute the gap to a stage

   Saying engagement is low is a complaint. Saying sign-in completion is where the drop happens is an assignment.

• Gave a portfolio of accounts a comparable read on performance rather than account-specific storytelling.
• Shifted conversations from install counts to habit formation.
• Directly shaped the later customer health work, which is the same idea applied to individual customers instead of populations.

Reading behavior at population scale taught me most of what I later applied to individual customer health. The mistake I made early was presenting more analysis when an account resisted a conclusion. More analysis rarely wins that argument. A single clear stage-level gap, with an owner, usually does.

• Customer Operations
  Customer Health Framework →

  A framework for reading customer signals, separating recoverable customers from finalized losses, and matching an intervention to the reason rather than the symptom.

• Customer Operations
  Advocacy & Sentiment Operations →

  Turning public reviews, ratings, and open-text feedback into coded reasons, counted themes, and assigned owners.