Analytics skills for your AI agent
Conversion fell.
Rollback is the wrong call.
Both channels improved. Mix did the damage. One skill, four rows, two minutes — before someone acts on the average.
No warehouse. No Engine. Copy the prompt; the CSV is already on GitHub.
Conversion fell.
Both channels improved.
Tempting call: roll back signup. The weights moved, not the rates.
/diagnose-change Should we roll back signup? Separate mix from within-channel change.
The average changed.
The story needs context.
- −3.6pp Mix pulled conversion down.
- +1.0pp Each channel improved.
- −2.6pp Together they are the total. Not a product cause.
shape-narrative Lead with the decision: do not roll back on this aggregate.
“Both channels got worse.”
Does that claim hold?
- ↳ Direct: 10% → 11%.
- ↳ Paid: 4% → 5%.
- ! An aggregate drop is not a rollback brief.
analysis-review Kill the rollback claim before the meeting.
Four-row CSV on GitHub · prompt hits the live file · not a model transcript
diagnose-change is enough for this case. The rest of the set is one command.
npx skills add patrickjmorris/aftergrid --skill diagnose-changeThe average will
run the meeting.
These are the moments an agent without judgment ships the wrong story.
Vague ask
“Did onboarding work?” is four questions.
grill-question02Number moved
Reproduce it. Mix vs rate. Then talk.
diagnose-change03Draft too sure
60% vs 40% is not a mandate.
analysis-review04Same mistake twice
Keep the lesson. Retrieve it next time.
learn-from-analysisThen use the same skills on your CSV, notebook, or warehouse.
Choose a workflowPick the skill
the stuck point needs.
14 portable skills, plus optional Engine setup. Unsure which? ask-aftergrid names one and stops.
15 skills
ask-aftergrid
Choose the next skill for the work in front of you.
Framegrill-question
Turn a vague ask into a question worth answering.
Frameexplore-data
Learn the shape of the data before trusting it.
Framedefine-metric
Make the denominator, window and exclusions explicit.
Frameplan-analysis
Choose the comparison and the checks before the result.
Frameanalyze
Carry a question through to a reviewed answer.
Analyzechecked-analysis
Check the data and comparisons as you work.
Analyzediagnose-change
Reproduce a move, separate mix from within-group change, and test rival explanations.
Analyzewrite-finding
Turn the analysis into an answer someone can use.
Explainiterate-visual
Make the chart earn the space it takes.
Explainshape-narrative
Put the answer first. Make every section carry it.
Explainanalysis-review
Try to disprove the answer before someone acts.
Improverevise-finding
Revise the answer without losing what supported it.
Improvelearn-from-analysis
Keep the lessons that should survive the analysis.
Improvesetup-aftergrid
Add the optional Engine for evidence and reproducibility.
Engine · optionalNo skills match. Try a different word or stage.
See the miss,
then the catch.
A rollback on a mix-shift, growth that hid churn, a causal headline on self-selected retention, and a follow-up that has to re-test last week’s lesson.
Explore the examplesConversion fell. Both channels improved.
The rollback meeting. Total down, both channels up.
Follow the analysisSynthetic lab · revenueMRR grew. The starting customers shrank.
Headline +$10. Starting customers −$80. Reconcile before you celebrate.
Follow the analysisSynthetic lab · lesson reuseConversion recovered. Was it the same mechanism?
Last time it was mix. Check before repeating the story.
Follow the analysisSynthetic lab · reviewRetention was 50% higher. Did onboarding cause it?
A mandate and a forecast sitting on a self-selected 60% vs 40%.
Follow the analysisReal open data · NYCTwo questions about New York
Real taxi, weather, and bike data. One inconclusive answer, one descriptive one, dead ends kept.
Draft Findings · not publication-approvedFollow the analysisThe habits
behind the skills.
Ask first. Diagnose the move. Keep what you learned.
Built by an analyst.
Available for hire.
I’m Patrick Morris. I made these because the agent answers reaching product and growth decisions were fast, confident, and wrong in ways an analyst catches in a minute.
About the projectPut the judgment in your team’s agent.
Metric definitions a second analyst would apply the same way. A review step before numbers reach a decision. Lessons the next analysis retrieves. Pilots are open.
Work with meSponsorshipFund the public skills and examples.
A named line here and in the docs. Nothing commercial inside a Finding, and no sponsor touches a metric approval.
Sponsor aftergridNext time a metric moves,
don’t let the average run the meeting.
Install one skill. Paste the four-row prompt. See whether the agent still recommends the rollback.
Try the conversion trap