Loading

Metrics deep-dive to narrative

What moved, why, and which ship caused it.

10 minProductClaude Cowork
Loading

Set up

Try a plugin

The Product Management plugin ships with /metrics-review and other analysis skills as a starting point, already structured to find the anomaly, segment to the driver, and write the story. If your admin manages plugins and it's not available yet, skip this; nothing below requires it.

Product ManagementWrite feature specs, plan roadmaps, and synthesize user research faster. Keep stakeholders updated and stay ahead of the competitive landscape.
Add
/metrics-reviewReview and analyze product metrics with trend analysis and actionable insights.
Run
/stakeholder-updateGenerate a stakeholder update tailored to audience and cadence.
Run

Connect your tools

Claude Cowork is more powerful when it works directly with your systems. You control permissions and access. Learn about tool access.

Navigate to Customize → Connectors in Cowork to set up.

Amplitude
Connect
Databricks
Query the metrics tables directly so the narrative cites the same numbers as the dashboards.
Connect
Browse all connectorsOpen in Cowork

Set your working folder

Drag the files you'll use (the Amplitude or Mixpanel CSV export, last week's narrative for comparison, the changelog if it's a flat file) into one folder on your machine, then point Cowork at it. Cowork reads from it and writes the narrative and chart images back to it. If you write this every week, create a Cowork project from that folder so your metric definitions, instructions, and memory stay attached.

Product / metrics / wk17
amplitude-activation-funnel-wk17.csvApr 24, 20262.4 MB
looker-retention-by-plan.csvApr 24, 2026880 KB
changelog-apr.mdApr 23, 20269 KB
narrative-wk16.mdOptionalApr 17, 20266 KB
In Cowork’s chat bar:Product / metrics / wk17

The prompt

Copy this into Claude Cowork

Find the anomaly in this week's activation funnel. Segment by plan, platform, and signup source until you've isolated the driver, check it against what shipped in the same window, and write the "what happened and why" narrative with the two charts that prove it.

Product / metrics / wk17

Why this works

Prompt

Ask for the evidence behind each claim. Starting with "find the anomaly" lets Cowork scan the whole funnel for the step that actually moved instead of anchoring on the metric you happened to notice first.

Prompt

Name the segments to try. "Plan, platform, and signup source" gives the isolation a search order, so the narrative points to the one segment that explains most of the move rather than a grid of every cut.

Source

Check it against what changed. Checking the changelog and Linear over the same window turns correlation into a causal hypothesis the team can act on, not just a chart that went down.

Prompt

Set a chart limit. "Two charts that prove it" keeps the output to the before/after and the segment split that carry the argument, instead of a dashboard dump.

Get a better draft

Practice

Add an example to match. Drop an example you like into the folder and Cowork matches your structure and voice.

Practice

Ask it to flag uncertainty. Add "flag anything you're not confident about" so you know where to look first when you review the draft.

Make Cowork work for you

A plugin skill is a starting point — customize it with your own practices and expertise. A few minutes of conversation and it runs with your standards from then on.

Make what we've done in this task so far into a skill, or edit the /metrics-review skill with my feedback.

Product / metrics

Make it repeatable

Run it every Monday morning

The "why did the number move" question comes every week whether or not you've had time to dig. Type /schedule in the prompt, or open Scheduled in the Cowork sidebar, and the customized skill writes the narrative before standup.

/schedule Every Monday at 7am, run /metrics-review for the prior week, write the narrative and charts to Product/metrics/<week>, and flag anything that moved more than 10% week over week.

Product / metrics
Scheduled taskActive
Weekly metrics narrative

Runs /metrics-review against the prior week's data, isolates the driver, and writes the what-happened-and-why with charts to the week's folder.

Every Monday at 7:00amOpen in Cowork

Share with your teammates

Your customized /metrics-review now carries your metric definitions, your segment order, and your narrative format. Share it so any PM can answer "why did this move" in your structure, and the weekly readout stays consistent no matter whose number it is.

Going forward