Analyze & Decide

A metrics review that keeps asking why until the number confesses

Dashboards show you what moved. Idam AI's Metrics Analyzer digs into why - comparing trends against targets, flagging anomalies, correlating drops with launches and incidents, and turning the whole review into a scorecard with recommended actions.

Weekly Metrics Review · Activationvs. target · last 12 weeks
targetAnomaly · Week 27Correlates with signup-form incident
Activation 58%+4.2 pts vs. target
Week-2 retention 31%-1.8 pts · investigate
NPS 41flat · on target
How a Review Runs

From raw numbers to a decision-ready scorecard

Weekly, monthly, or “the CEO just asked why retention dipped” - the same three-step review works at any cadence. You supply the numbers and the context; the agent supplies the analysis and the so-what.

01

Bring the numbers, any way you have them

Paste a table from your Monday dashboard screenshot ritual, upload a CSV, or connect PostHog and let the agent pull the period itself - along with comparison data and targets.

Paste or uploadPostHog connectionTargets & prior periods
02

Trends, anomalies, and correlations

Each metric gets a trend read: direction, rate of change, distance from target. Sudden moves get cross-checked against launches, incidents, campaigns, and seasonality you mention - and segment breakdowns reveal when one cohort is dragging the aggregate.

Anomaly detectionSegment drill-downEvent correlation
03

A scorecard your team can act on

The review ends as a structured scorecard - status per metric, the narrative behind each change, and recommended actions ranked by urgency. Share it as-is or hand it to the Team Translator for a leadership-ready update.

Metric scorecardRecommended actionsOne-click stakeholder update
More Than a Dashboard Narrator

Built on how working PMs actually read metrics

The agent structures every review around a metrics hierarchy - North Star on top, health indicators beneath, diagnostics below that. If you haven't defined yours yet, it helps you do that first instead of analyzing noise.

North Star first, noise last

Reviews start at the metric that matters and drill down only where something moved. No more treating forty equal-weight tiles as a strategy.

Metrics Hierarchy
North Star · Weekly active workspaces
Activation
Retention
Revenue
Satisfaction
L2 diagnostics surfaced only when an L1 moves

Aggregates lie. Segments don't.

A flat top-line can hide one cohort collapsing while another grows. The agent breaks trends down by segment and tells you which slice is actually driving the change.

Retention · By Plan
Trial users-6.4 pts
Standard plan+0.8 pts
Annual plan+1.1 pts
Aggregate drop fully explained by trial cohort

Anomalies, with their alibis checked

A spike is only interesting if you know what caused it. The agent lines anomalies up against the launches, incidents, and campaigns you provide before it lets a theory stand.

Anomaly Investigation
Signups -22% · June 3–5

Checked: pricing change (no), seasonality (no)

Match: signup-form incident, June 3 · 14:20

Recommended: exclude window from trend; verify form monitoring
The Monday Ritual, Upgraded

What changes when an agent runs the review with you

Metrics review without Idam
  • Screenshot the dashboard, paste it into Slack, hope someone asks a good question
  • Eyeball six charts and declare the week "basically flat"
  • Notice the retention dip two weeks after it started
  • Rebuild the same comparison table in a doc every Monday
Metrics review with the Metrics Analyzer
  • Every key metric read against target, prior period, and rate of change
  • Anomalies flagged the week they happen, with likely causes checked
  • Segment breakdowns that show which cohort is moving the number
  • A scorecard with recommended actions, ready to share in one click
Frequently Asked Questions

Common questions, straight answers

It combines three things: trend math (direction, rate of change, distance from target), segment breakdowns (which cohort is moving the aggregate), and the event context you provide - launches, incidents, campaigns, seasonality. It proposes the most plausible explanation and is explicit when the data cannot distinguish between causes.
Get Started

Next Monday, walk in already knowing why

Bring this week's numbers and run a real review - trends, anomalies, segments, and a scorecard with actions attached.

Pairs Well With

Related Agents

Analyze

AI Feedback Analysis

Numbers tell you what moved; feedback tells you how it felt. Pair the scorecard with themes mined from tickets, surveys, and interviews.

Learn more →
Research

AI Competitor Research

When a metric shifts because the market shifted, get a live-web competitive brief that explains what changed outside your product.

Learn more →
Communicate

AI Team Translator

Hand the finished review to the Team Translator and get a leadership update, an engineering deep-dive, or a board-ready summary.

Learn more →