── SKILL
onboarding-audit✓ APPROVED

Score whether new customers feel taken care of.

Grades the first 90 days on the moments that make someone feel expected, not just whether they activated, and names the exact spot the welcome goes cold.

In plain English: “Is our onboarding actually any good?

Inside: The moment-by-moment rubric, the cold-spot finder, the silent-stall check, and the fixes.

01 / HOW TO USE

Install it in one line, or paste it in.

1
In Claude Code (one command)
Copy the install line, paste it into your terminal, and restart Claude Code. The skill installs itself to ~/.claude/skills/ and runs automatically when it is relevant.
2
In Claude, ChatGPT, or a Project (no terminal)
Open the file, then upload it to your chat or paste its contents in. A skill is just a markdown file of instructions, so any capable AI can follow it.
New to skills? A skill is a plain-text file that teaches your AI a workflow. Point any capable assistant at it and it follows the steps, on your data.
02 / MAKE IT YOURS

Connect your context. Set it to your motion.

CONNECT YOUR CONTEXT · AND WHY IT HELPS
a product-analytics tool

it sees where real customers stall versus where the playbook assumes they sail through.

a CRM or CS tool

it reads the human touches that actually happened, the calls, the check-ins, the silences.

support or survey data

it hears the friction in the customer's own words.

SET IT TO YOUR MOTION

This was built to be lifecycle-agnostic. Set these to your motion:

The rubric is yours. The point is to grade the experience your customers should feel, not a generic checklist someone else outgrew.

Set thisWhat it isDefault / Example
LIFECYCLEyour onboarding windowfirst 30 / 60 / 90 days
MOMENTSthe touchpoints you gradewelcomekickofffirst valuehabitcheck-in
VALUE_EVENTthe moment they first get valueyour product's aha moment
HUMAN_TOUCHESwhere a person shows upkickoff callcheck-inQBR
SIGNALShow you see a customer stallinglogin gapsetup abandonsupport spike
VOICEthe feeling you want to leaveyour brand's version of "taken care of"
03 / THE FULL SKILL

Everything the skill does, in full.

── WHAT THIS DOES

Grades the experience a new customer actually has, not just whether a checkbox got ticked. Most onboarding scoring measures activation: did they log in, did they finish setup. This scores the feeling: were they expected, were they guided, did anyone notice when they went quiet. It walks the first 90 days moment by moment, scores each one, and hands back the specific places the welcome went cold and what to do about it. In commoditized software the experience is the moat, so this measures the moat.

── THE METHOD
  1. 1
    Score the moments, not the checkboxes

    Walk the first 90 days as the customer lives it: the welcome, the kickoff, the first real value, the habit forming, the first check-in. Each moment gets a score on two axes: did it happen, and did it feel like someone was expecting them. A completed setup with no human warmth scores lower than it looks.

  2. 2
    Find where the welcome goes cold

    Name the exact moment the attention drops off. For most teams it is right after setup: the sale got all the energy, activation got a sequence, and then the customer is alone. That cold spot is where early churn is born, and it has a location you can point to.

  3. 3
    Grade the human touches

    Automation onboards; people make customers feel chosen. Score where a human actually shows up versus where the customer is handed to a drip. A high-value customer getting a templated welcome is a scored miss, not a nice-to-have.

  4. 4
    Catch the silent stall

    Score whether anyone notices when a new customer goes quiet. The worst onboarding failure is invisible: they stopped, and no one reached out, because no signal was wired to a human. If the stall is silent, that is the finding.

  5. 5
    Turn the score into fixes

    Every low-scoring moment ends in a specific move: the touch to add, the signal to wire, the human to insert. Not "improve onboarding." The exact cold spot and the exact fix.

── QUALITY GATES
  • Every graded moment scores both dimensions: did it happen, and did it feel intentional.
  • The cold spot is named with a location, never "onboarding could be better."
  • Human touches are scored separately from automation, never blended into one number.
  • The silent-stall check is explicit: is a quiet customer noticed, yes or no.
  • Every low score ends in a specific fix, never a generic recommendation.
── OUTPUT (EXAMPLE)
EXAMPLE OUTPUT
ONBOARDING AUDIT · [your product] · first 90 days

Moment              Happened   Felt intentional   Score
Welcome             yes        templated          2/5
Kickoff call        yes        warm, tailored     5/5
First value         partial    no guidance        2/5
Habit forming       no         -                  1/5
Day-30 check-in     no         -                  0/5

COLD SPOT: right after first value. The kickoff is great, then the
customer is alone. No habit nudge, no Day-30 human touch.

SILENT STALL: not caught. No signal wired to a human when a new
account goes quiet in week 3.

FIXES:
  1. Wire a week-3 inactivity signal to the CSM, not a drip.
  2. Add a habit nudge tied to the aha moment, day 7 to 14.
  3. Make the Day-30 check-in a human note, not an NPS survey.

Onboarding experience score: 2.0 / 5 (strong start, cold finish)
── EXAMPLE CHAT
EXAMPLE CHAT
You
Audit our onboarding. New customers get a welcome email, a setup wizard, a kickoff call, then an automated 5-email drip over 30 days. Renewal's at month 12.
CLAUDE
Scored it as the customer lives it, not as the flowchart draws it: Welcome email templated, no name of a human 2/5 Setup wizard works, but no one watching 3/5 Kickoff call the bright spot, warm and tailored 5/5 Day 1-30 drip automated, same for everyone 2/5 The cold spot is obvious: your kickoff call is excellent, and then the customer falls into a drip that treats a $50k account the same as a free trial. The feeling goes from "chosen" to "processed" in a week. And the silent stall is not caught: if they stop logging in during the drip, nothing routes to a human.
You
So what would actually move it?
CLAUDE
Three moves. Replace the day-14 drip email with a real check-in from the CSM who ran the kickoff, so the warmth carries. Wire an inactivity signal so a quiet account in week 3 pings a human, not an autoresponder. And make the day-30 touch a "here's what you've gotten to so far" note instead of a survey. One sharpener: connect your product-analytics tool and I'll show you exactly which day the average new account goes quiet, so the check-in lands the day before, not a week after.
── WHERE THE NUMBERS COME FROM

The moments, the window, and the scoring bar are yours to set. The examples here suited a 90-day mid-market onboarding. A self-serve motion compresses the window; an enterprise motion stretches it. The logic, score the feeling not the checkbox, find the cold spot, catch the silent stall, does not change.

04 / GO FURTHER

Where an operator takes this next.

The audit is step one. Here is where an operator takes it once the score is real.

You grade the welcome once, then keep it warm on purpose.

01
Wire the silent stall to a human

Have a scheduled Claude task read Amplitude or your warehouse nightly and DM the CSM in Slack the new accounts that went quiet this week, so no welcome ever goes cold unnoticed.

02
Score every new cohort, not just once

Re-run the audit each month against the latest cohort so a regression in the experience shows up as a dropping score, not as churn three months later.

03
Feed the cold spot into the offboarding side

Hand the findings to an offboarding audit so you can see the full arc: how you greet them and how you say goodbye, and whether the two match.

05 / PART OF A BIGGER PLAY

One skill is the on-ramp.

── GET NEW SKILLS IN YOUR INBOX

One operator-built skill a week. Free.