Metrics Analyst
Tells you what the numbers show — and what they cannot tell you
4 of 8 metrics need attention.
8 metrics checked. 4 flagged. The largest gap is 92 days.
Start with Churn score. Kim has the most flagged.
What a metric should be is for its owner; the list shows which ones nobody stands behind or that have gone quiet.
Why rows were flagged
- No written definition30%
- Not refreshed for 30 days30%
- No owner20%
- More than 20% below target20%
Value flagged by Owner
- Kim1
- Dana1
- -2
Items to act on, worst first
| Item | Owner | Value | Days since the refresh | Why |
|---|---|---|---|---|
| Churn score | - | 0 | 81 | No owner, No written definition, Not refreshed for 30 days |
| Gross margin | - | 0 | 0 | No owner |
| Trial conversion | Kim | 40 | 92 | No written definition, Not refreshed for 30 days, More than 20% below target |
| Pipeline coverage | Dana | 40 | 59 | No written definition, Not refreshed for 30 days, More than 20% below target |
What to do this week
- Resolve Churn score first: it is the most urgent on the list.
- Fix the data or the status on every flagged row, then re-run.
- Re-run this check every time the list changes, before it reaches a decision.
Do
- Fix the highest-risk item first, then run the check again on fresh data.
Don't
- Don't act on one row alone: open the source record first.
How to improve
- Add more detail per row (owner, date, source) so the next report can trace each item.
- Add a second period of data so the next report can show whether things are improving.
Demo
Overview
Ask most data chatbots why revenue fell and they will tell you why. The data almost never knows. It holds a correlation, a change in mix, or a coincidence, and a confidently stated wrong cause is worse than no answer — the owner acts on it.
This agent answers in three parts. What the numbers show, stated as arithmetic anyone can check. What they suggest, labelled as a hypothesis with what would confirm it. And what this data cannot answer, with what you would need to collect.
It checks the shape of a change before explaining it: a fall that is really one large customer leaving is a different business problem from a fall spread across everyone, and averages hide the difference. It looks for the composition change behind a moving average, flags periods too short to conclude from, and refuses to attribute a change to a cause the data does not contain.
Benefits
Decisions on real findings
You can tell a fact from a hypothesis at a glance.
Faster than a spreadsheet
Composition and cohort checks nobody has time to run.
Honest about limits
The most valuable output is often 'this data cannot tell you that'.
What it does
Shows, suggests, cannot say
Three separate sections. Mixing them is how a correlation becomes a decision.
Checks the shape of a change
One large customer leaving is not the same problem as a fall across everyone. Averages hide it.
Finds the composition change
The average moved because the mix moved, not because anything got better or worse.
Arithmetic you can check
Every figure shows how it was computed, from which rows.
Says when the period is too short
Rather than reading a trend into three weeks of noise.
Names what to collect
The unanswerable questions come with what data would answer them.
What this changes
What this changes about your week
- It takes over
- Working out what actually changed underneath a number that moved.
- Done today by
- An analyst, or an argument in a meeting.
- How often
- Every review, and every surprise.
- Typical manual effort
- Composition and cohort checks are a day in a spreadsheet.
Why Data & Business Intelligence in particular. It separates what the data shows from what it suggests. Acting on a confidently stated wrong cause is worse than having no answer.
Runs unattended
Started by you or by an event, and it finishes on its own. Nothing waits for someone to be at a desk.
The same standard every time
The two-hundredth item is held to the bar the first one was. Consistency is the part people cannot sustain.
It cannot act on its own
Metrics Analyst has no path to sending, spending or committing. That limit is why its output is safe to act on.
How it runs
This agent runs server-side through the PROMIVO runtime. Each run is logged step by step and every tool call is permission-checked before it executes.
- 1InputValidated against the agent input schema
- 2Modelclaude-opus-5 with automatic fallback
- 3Toolsextract_structured_data
- 4PermissionsRuns READ actions automatically; anything higher waits for approval
- 5OutputStructured result matching the output schema
- 6LoggingTokens, duration and every step recorded
Requirements
Integrations
No integrations required.
Before you start
- The data, as a table or export
- The question you want answered
- Requires the All-Access plan or higher
Use cases
A number moved
Find out what actually changed underneath it.
Monthly review
What genuinely changed, separated from noise.
Before a decision
Check whether the data supports what everyone is assuming.
Pricing
All-Access
Billed monthly through your PROMIVO subscription. Cancel at any time.
Runs consume your plan allowance for agent executions and tokens. See plan limits
FAQ
Will it tell me why revenue fell?
It will tell you what changed — which segments, which customers, which mix. Whether that caused the fall is usually a hypothesis, and it will label it as one and say what would confirm it.
Does it connect to my database?
Not yet. You supply the data — a table, an export, a summary. Direct connections need credentials and query permissions, which is a separate decision from analysis.
Can it forecast?
Only where the history genuinely supports one, and it will state the assumptions and the range. It will not produce a single confident number from a short or erratic series.
Reviews
Reviews open once customers have run this agent.
Need this agent to work differently?
Tell us what to change and our team will scope a customised version for your business.
Customize this agent