Profit growth lever tree (Leads × Conversion × Transactions × Value − Costs)
![[assets/2026-11-06-profit-growth-lever-tree.jpeg]]
Mental model in one line: profit breaks down into four multiplicative revenue terms minus three cost buckets, so every growth idea should name the one leaf it moves.
What it says (author's terms)
Profit = Leads × Conversion Rate × Transactions per customer × Value per transaction − Costs
- Increase Revenue
- Get and keep more customers
- Acquire new customers → increase lead volume, increase conversion rate
- Retain or win back customers
- Increase revenue per customer → increase transactions per customer, increase value per transaction
- Get and keep more customers
- Decrease Costs → reduce cost to deliver · reduce cost to acquire · reduce fixed costs
Green levers grow revenue and red levers cut costs. The caption says "Profit has four terms", meaning the four revenue factors.
Why this is in the vault
The founder flagged it as the model to use "when we want to analyze growing a business." It's a compact, teachable first-pass decomposition. It gives us one shared vocabulary when we look at an acquisition target, a Scribble Works pricing question, or a client value conversation.
Mapping against Ray Data Co
- Acquisition funnel ([[2026-08-30-shortlist-refresh-scored]]): for each shortlisted business, ask which leaf an owner-operator plus agents moves cheaply. In most SMB targets the best leaves are conversion rate and cost to deliver, not lead volume. That's a sharper screen than "AI can help this business."
- CAF / Business Value framing: the money branch of the phData value tree (profit / quality / people) is this same shape, so this diagram can serve as the client-facing simplification of that branch. Note that it covers only profit. Quality and employee/community value have no node here, which is the gap the three-tree framework exists to fill.
- Agent use cases: each leaf is a natural home for a use-case tag. Speed-to-lead and follow-up agents are conversion rate. Win-back campaigns are retain. Support deflection is cost to deliver. Tagging use cases this way makes the business case one line instead of a waterfall.
Where the model is weaker than it looks
- The formula and the tree disagree.
Leads × Conversiononly produces new customers, but the tree has a "Retain or win back" branch that the formula has no term for. A correct version would beCustomers = (Leads × Conversion) + Retained − Churned. For a recurring-revenue or repeat-purchase business, retention is often the biggest lever, and the formula hides it. - The levers interact. Raising value per transaction (price) usually lowers conversion. Cutting cost to acquire usually lowers lead volume. [[2026-06-13-cfo-secrets-value-atoms-growth-cfo-ii]] makes this exact critique of KPI trees: the branches feed and constrain each other. And [[2024-07-13-cfosecrets-7-levers-pl-unit-economics-ii]] treats the P&L as a system with second-order effects, not separate dials.
- Mix and time are missing. There's no product or customer mix term and no lifetime or cohort view. Use the tree for a quick read, then switch to a cohort or unit-economics model before you bet money on a lever.
Ray's recommendations (added 2026-11-06 at founder request)
1. Use a corrected formula
Replace the four-term version with one that has a place for retention and churn:
Profit = Active customers × Transactions per customer × Value per transaction − Costs
Active customers = Retained customers + New customers
New customers = Leads × Conversion rate
Retained = Last period's customers × Retention rate (+ win-backs)
Costs = Cost to deliver (variable, per transaction)
+ Cost to acquire (Leads × cost per lead, or New customers × CAC)
+ Fixed costs
This keeps every leaf in the original tree. It also makes two hidden links show up in the math: acquisition cost scales with lead volume, and delivery cost scales with transaction count.
2. Add the missing leaves to the tree
- Under Get and keep more customers: split "Retain or win back" into retention rate and win-back rate as separate leaves. They're different motions with different owners.
- Under Increase revenue per customer: add mix, meaning moving customers toward higher-margin products, alongside transactions and value.
- Under Decrease Costs: split cost to deliver into unit cost and utilization. Utilization is the agent-native lever: the same people or tools serving more volume.
3. Run it as a four-step analysis, not a picture
- Baseline: put a real number on every leaf for the business (last 12 months). A leaf with no number is the first finding.
- Pick one primary lever: choose the leaf where a 10% move is cheapest. Write down why in one line.
- Name the knock-on effects: for that lever, list the 1-2 leaves it pushes the wrong way (price ↔ conversion, lead volume ↔ cost to acquire, discounting ↔ value per transaction). Estimate the net effect, not the gross.
- Check against cohorts: before you commit money, confirm the lever on a cohort or unit-economics view, not just on the totals. The tree is the map. The cohort model is the test.
4. Where to apply it
- Acquisition scans: add
primary_leverandlever_rationalefields to each scored target in the monthly rescan. Default hypothesis for small owner-run service businesses: conversion rate (speed-to-lead and follow-up) and cost to deliver (utilization) beat lead volume. Treat that as a hypothesis to check per target, not a given. - Client value conversations: use this tree as the plain-English front of the money branch. Tag each agent use case with the leaf it moves (speed-to-lead → conversion · win-back campaigns → retention · support deflection → cost to deliver · quoting/upsell assist → value per transaction). Quality and people value still need their own branches, because this tree only covers profit.
- Scribble Works / own bets: for any pricing or growth proposal, require one line saying which leaf it moves and which leaf it might hurt.
Open questions
- Founder call: add
primary_leverto the acquisition rescan scoring (recommendation 4)? It would make month-over-month diffs easier to compare. - Is there a version of this tree for the quality and people branches, so the client value story isn't only about profit?
Related
- [[2026-04-30-amazon-wbr-metric-trees-business-stack-view-research]]: metric trees as an operating surface
- [[2026-06-13-cfo-secrets-value-atoms-growth-cfo-ii]]: the interdependency critique of KPI trees
- [[2024-07-13-cfosecrets-7-levers-pl-unit-economics-ii]]: seven P&L levers, a finer cut of the same idea
- [[2026-08-30-shortlist-refresh-scored]]: where the lever screen would be applied