Your first driver tree

Ten builder sessions, one skill: take any top-line business metric, break it into the drivers beneath it, and use that tree to diagnose a drop in minutes. You will learn the three decomposition patterns (b2), how to find leading indicators and run the diagnosis playbook (b3), then apply all of it to six real domains - ecommerce (b4), marketing (b5), branding (b6), internet/traffic (b7), B2B SaaS (b8), and B2G (b9) - before a capstone where a metric drops and you diagnose it end to end (b10). The leader track (a1-a6) teaches the same thinking for reading trees and running the weekly and monthly metric reviews.

Almost every top-line number is a product or a sum of smaller numbers. GMV is not a thing you control directly - it is what you get when traffic, conversion rate, and average order value multiply together. Each of those is itself controllable, measurable, and ownable. The tree just writes that relationship down so everyone argues about the same picture.

GMV = Traffic x Conversion x AOV is a multiplicative tree: the top line is the product of its drivers. Its defining property - the one that makes diagnosis fast - is that a percentage change in any single driver flows straight to the top line as the same percentage. Traffic down 10% with everything else held? GMV down 10%. That clean pass-through is exactly what you exploit when a metric falls.

Here is the GMV tree as a live simulator. Every leaf is editable and every parent recomputes instantly. First read it, then edit a driver and watch the top line move, then press Simulate a drop - one driver gets knocked down at random and the whole path from that leaf up to GMV lights coral. Your job: name the guilty driver from the trail, every time.

{
  "unit": "$",
  "root": {
    "label": "GMV", "op": "x",
    "children": [
      { "label": "Traffic",     "value": 120000, "unit": "visits" },
      { "label": "Conversion",  "value": 0.028,  "pct": true },
      { "label": "Avg order value", "value": 62, "unit": "$" }
    ]
  }
}

The tree tells you which driver fell. Segmentation tells you where. If conversion dropped, was it every channel or just one? Real diagnosis ends in a GROUP BY. These editors run live SQL against a small store's orders - the same move you make when the tree points at a guilty driver.

SELECT o.channel,
       COUNT(DISTINCT o.order_id)             AS orders,
       ROUND(SUM(oi.quantity * oi.unit_price), 0) AS revenue
FROM orders o
JOIN order_items oi ON o.order_id = oi.order_id
WHERE o.status = 'completed'
GROUP BY o.channel
ORDER BY revenue DESC;

SELECT o.channel,
       ROUND(SUM(oi.quantity * oi.unit_price) * 1.0
             / COUNT(DISTINCT o.order_id), 2) AS avg_order_value
FROM orders o
JOIN order_items oi ON o.order_id = oi.order_id
WHERE o.status = 'completed'
GROUP BY o.channel;

• Pick the top-line metric your team reports. Write it as a product or sum of three to five drivers. If you cannot, you do not yet understand what moves it - that gap is the point.
• For each driver, name the team that owns it. A driver nobody owns is a driver nobody watches.
• Bookmark this page. The simulator is your scratch pad - paste your own numbers into any later session's tree.
• Next time someone says "the number is down", ask "which driver?" before "why?". Watch how fast the room re-orients.
• Bring your metric's tree to b2 - we will decide whether it is multiplicative, additive, or a flow bridge.

Metric decomposition has no single certificate - it is a craft distilled from a handful of durable frameworks. This track teaches the working core of these, applied live instead of described. This page draws on: