The 21:1 stat is real - and mostly useless
Email's famous 21:1 average return is cited everywhere and understood almost nowhere. It masks enormous variance: a dead list returns 1:1, a segmented one returns 40:1. The number only becomes useful when you can measure email's contribution to your own sales - which requires attribution, not vibes.
Attribution models that survive contact with reality
Last-click is where most brands start and where most brands stay wrong - email rarely closes in the first click. The workable frameworks: multi-touch linear (credit spread across touches), position-based (40% first, 40% last, 20% middle), and platform-native models like Klaviyo's channel attribution, which use data science to resolve which channel created the sales.
The rule: pick a model, apply it consistently, and change it only deliberately. Attribution consistency beats attribution perfection.
- Last-click: simple, biased, fine for daily ops
- Linear: fair across short journeys
- Position-based: honors both opener and closer
- Platform models: best for e-commerce stacks
The dashboard your CFO will accept
Four numbers, every month: sales attributed to email, email as a percentage of total sales, cost per acquired email customer, and LTV of email-acquired customers versus all others. When email-acquired customers spend more over time - which they usually do, because they opted in twice - that LTV gap is your true ROI story.
Present it alongside the operational metrics - open, click, unsubscribe - as context, not conclusion. CFOs buy sales, not open rates.
Raising the ratio, not just reporting it
The fastest ROI lever is segmentation, not volume. An RFM segmentation - recency, frequency, monetary - sends the 90-day buyer a different email than the six-month-lapsed one, and both outperform the blasts they replace. The second lever is flow architecture: welcome, cart, win-back. The third is cadence discipline: one fewer email that performs is better than one more that doesn't.
Measure, attribute, then let the data dictate the send. Email rewards the scientific, not the prolific.
