For humans · Act
Start the day knowing exactly what needs you.
Every morning, each person gets a plan built from the metrics they own: what moved, why it moved, and the actions worth taking, sized in money. Not a dashboard to interpret: a plan with your name on it.
A plan with your name on it. Built with Canopy Agent Workflows.
Personalised by ownership, not preferences. RACI decides
The plan is built from the metrics where you are named Responsible or Accountable. No saved filters, no persona dropdowns: the ownership graph decides what makes your list.
Every number arrives interpreted. Drivers attached
Each item carries the driver behind the move, the statistical evidence for it, and the expected impact in money, so the first question of the day is never what does this mean.
Reasoning you can audit. Every step logged
The plan lists the steps it took and the evidence it used, and every run is logged, so you can audit why an item made your list before you act on it.
Dashboards wait to be read. This reads them for you.
The morning ritual of the modern data stack is a dozen tabs and a question: is anything wrong? The action plan inverts it. Overnight, an agent reads every metric you own, classifies what is off track, traces each move through the causal tree to its driver, and ranks the actions worth taking with the expected impact attached. What lands at 08:00 is not a report about data. It is a short list of decisions, already argued for, with the evidence one click away.
Starts from what you own
Built from the metrics where you are named Responsible or Accountable. Nobody else receives this briefing.
Numbers from the actuals pipeline
Like-for-like comparisons are precomputed for every date, so MTD never means two different things.
Drivers tested, not guessed
The causal claim carries its statistics from the tests that run on every tree edge daily.
Today's briefing
Tuesday · 08:00Good morning, Sarah. Revenue is tracking £31.5k/day behind target, down 13.1% MTD, though the year-to-date trend holds at +24%. The shortfall is concentrated: the online store and Liverpool Street account for most of the decline, and the pattern points to checkout failures and a staffing gap, not to demand.
Where the gap is
| Location | MTD | Change | Status |
|---|---|---|---|
| Online store | £188.4k | -9.8% | Needs attention |
| Liverpool Street | £26.9k | -40.9% | Critical |
| Shoreditch | £48.1k | +3.4% | On track |
| Manchester | £31.2k | -2.1% | Monitor |
Like-for-like days, precomputed by the actuals pipeline.
Diagnosis · Online store
Conversion is 23% off pace while sessions are flat, so this is not a demand story. The drop traces down the tree to checkout, where payment errors began minutes after Monday's checkout release. Card-on-file customers are unaffected, which isolates the new payment form.
Your checkout fix from February is verified at +£32k, measured against the causal baseline, 14 days on. The EU store shows the same error pattern today.
Do next · ranked by expected impact
Roll back Monday's checkout release
Sarah ChenTodayBackfill the two open shifts at Liverpool Street
David MitchellThis weekApply the checkout fix to the EU store
Sarah ChenThis weekKPI Tree App 08:00
Good morning, Sarah. Revenue is tracking £31.5k/day behind target, down 13.1% MTD. The gap is concentrated in the online store and Liverpool Street: checkout failures and a staffing gap, not demand.
Do next: 1. Roll back the checkout release (you, today) · 2. Backfill Liverpool Street (David) · 3. Apply the fix to the EU store.
FromKPI Tree <briefings@kpitree.co>
Tosarah@acme.co
SubjectToday's briefing: £31.5k/day behind target · 3 actions
Good morning, Sarah. Revenue is tracking £31.5k/day behind target, down 13.1% MTD, though the year-to-date trend holds at +24%. The shortfall is concentrated: the online store and Liverpool Street account for most of the decline, and the pattern points to checkout failures and a staffing gap, not to demand.
Where the gap is
| Location | MTD | Change | Status |
|---|---|---|---|
| Online store | £188.4k | -9.8% | Needs attention |
| Liverpool Street | £26.9k | -40.9% | Critical |
| Shoreditch | £48.1k | +3.4% | On track |
| Manchester | £31.2k | -2.1% | Monitor |
Like-for-like days, precomputed by the actuals pipeline.
Your checkout fix from February is verified at +£32k, measured against the causal baseline, 14 days on. The EU store shows the same error pattern today.
Do next · ranked by expected impact
Roll back Monday's checkout release
Sarah ChenTodayBackfill the two open shifts at Liverpool Street
David MitchellThis weekApply the checkout fix to the EU store
Sarah ChenThis weekKPI Tree
online
Good morning, Sarah. Revenue is £31.5k/day behind target (-13.1% MTD). Biggest drags: Online store -9.8%, Liverpool Street -40.9%. Checkout errors are the driver, not demand. Your February fix is verified at +£32k.08:00
Do next: 1. Roll back Monday's checkout release (you, today) · 2. Backfill the Liverpool Street shifts (David) · 3. Apply the fix to the EU store.08:00
The in-app briefing on your workspace home, annotated with where every claim comes from.
Classified before a word is written
Every metric is checked for anomalies, staleness and decline first. The prose is drafted from those verdicts.
Track record, verified
Impact is measured against the metric by the same pipeline that calculates your actuals, not self-reported.
Owners from the live org chart
Named people, not a mailing list. If nothing happens by the deadline, it escalates.
Starts from what you own
Built from the metrics where you are named Responsible or Accountable. Nobody else receives this briefing.
Numbers from the actuals pipeline
Like-for-like comparisons are precomputed for every date, so MTD never means two different things.
Drivers tested, not guessed
The causal claim carries its statistics from the tests that run on every tree edge daily.
Classified before a word is written
Every metric is checked for anomalies, staleness and decline first. The prose is drafted from those verdicts.
Track record, verified
Impact is measured against the metric by the same pipeline that calculates your actuals, not self-reported.
Owners from the live org chart
Named people, not a mailing list. If nothing happens by the deadline, it escalates.
Personal because of RACI, not preferences
Personalisation here is not a saved filter. KPI Tree holds a live RACI assignment on every metric, so the plan is built from the metrics where you are named Responsible or Accountable, and it knows who else to mention because the org chart is part of the model. The same morning produces a different plan for marketing, growth and finance, and each one stays short, because it only contains what actually falls to that person.
Sarah Chen
Accountable · ConversionDavid Mitchell
Responsible · Paid traffic
Emma Thompson
Accountable · Cash collectedIt shows its working
A plan you cannot interrogate is just another opinion. Every plan starts from your briefing, walks the tree beneath the metrics that moved, checks the statistical evidence on each driver, and looks up whether similar actions worked before. Each of those steps is recorded. The receipt is the difference between advice you follow and advice you trust: when an item says the checkout fix is worth scaling, you can see exactly why it made the list.
Built your plan in five steps
Every step is logged in the run, so you can audit why an item made the list.
Delivered where your day already starts
The plan is on your workspace home when you open KPI Tree, and it comes to you everywhere else: a Slack DM at 08:00, an email digest, or WhatsApp when you are on the move, scheduled in whatever timezone your morning happens in. Teams that live in Slack turn on the weekly action plan workflow and the plan lands in the channel before the Monday stand-up, with the driver context attached.
KPI Tree app · 08:00
Good morning, Sarah. Tuesday briefing: revenue is tracking £31.5k/day behind target, down 13.1% MTD. Three things need you.
Four weeks later, the plan is smarter
Actions taken from the plan are tracked against the metric they were meant to move, and the impact is verified by the same pipeline that calculates your actuals. That verdict feeds the next plan: actions that proved out get recommended with their track record attached, and the ones that did not get discounted. Compare a plan with the same plan a month later and you can see the learning, because the recommendation now cites the verified result.
Investigate checkout payment errors. Likely driver of the conversion drop. Unverified.
Scale the checkout fix to the EU store. The same error pattern, and the fix is verified at +£32k here. Ranked first because the last one worked.
An agent you can govern, not a black box
The plan is generated by a platform agent that runs with your permissions and read-only tools, so what it sees is exactly what you see, and it cannot touch what you cannot. Anything that would change the world routes through approvals with named reviewers and escalation, and you choose the model and cap the spend.
Weekly revenue review
scheduled · Mondays