Tags: commerce concept

Break-Even Analysis

Date: 2026-08-16

---

The volume at which a decision pays for itself. It converts “is this worth doing” from an argument into arithmetic — and the answer is frequently a number nobody believes is achievable, which is the finding.

---

What it is

Break-even is the point where a change’s incremental contribution equals its cost.

$\text{break-even volume}=\frac{\text{cost}}{\text{contribution per unit}}$

At business level, on the running model:

fixed costs             £180,000 / year
contribution per order       £15.00

break-even             180,000 ÷ 15  =  12,000 orders
actual                                   15,000 orders

margin of safety        3,000 orders  =  20%

The margin of safety is the useful output — how far volume can fall before the business stops covering its costs.

For a project or a test

The version you’ll use more often. A change costing £15,000 to build, on 500,000 annual visitors at 3% and £15 contribution:

break-even contribution        £15,000
÷ contribution per order        £15.00
                              ─────────
extra orders needed              1,000

current annual orders           15,000
required relative lift    1,000 ÷ 15,000  =  6.7%

A 6.7% lift to break even in year one. Now check whether that’s detectable: at a 3% baseline, a 6.7% MDE needs roughly 120,000 visitors per arm — around twenty weeks at 12,000 a week — Minimum Detectable Effect.

So the honest conclusion is often: this change cannot be validated by experiment at our traffic, and it needs a lift we’d struggle to detect even if it happened. That’s a genuinely useful answer, and it’s available in ten minutes before anyone builds anything.

Why it’s the right first question

It reframes the conversation from “will this work?” — unanswerable — to “how much would it have to work?” — arithmetic.

And it exposes a common failure: proposals whose break-even lift is implausibly large. If a change needs a 30% conversion improvement to pay back, the discussion is over regardless of how good the idea is.

In plain terms: you don’t need to predict the effect. You need to know what effect would be required, and then judge whether that’s plausible.

Where it’s used

• Test prioritisation. Break-even lift is a better filter than a scoring framework, because it’s derived rather than assigned — Test Prioritisation
• Discounting. How much extra volume must a promotion generate — Discount Impact on Margin
• Free shipping thresholds — Shipping Thresholds
• Channel decisions. What conversion rate makes this channel’s CAC affordable — Customer Acquisition Cost
• Feature and build decisions, as above

Getting it wrong

• Using revenue instead of contribution. A £15,000 build needs 300 extra orders at £50 revenue and 1,000 at £15 contribution. Using revenue understates the requirement by more than three times
• Ignoring ongoing cost. A feature costing £15,000 to build and £500 a month to maintain has a break-even that keeps moving
• Assuming the lift persists. Break-even calculated over a year assumes the effect lasts a year — Novelty and Primacy Effects, Winner’s Curse
• Ignoring cannibalisation. Incremental contribution, not gross — a change that shifts demand between categories may contribute nothing
• Forgetting the opportunity cost. The test slot and the engineering time had alternative uses

The habit

Run it before the work, not after. It costs ten minutes, it kills roughly half of proposals, and the half it kills are the ones that would have consumed months to reach an inconclusive result — Guide - Unit Economics and Commercial Decisions.