Email Frequency Optimisation Calculator

Written out

The optimal frequency is where the marginal send still produces more contribution than the future value it destroys through unsubscribes and disengagement.

Marginal send value = (orders from that send × contribution) − (unsubscribes × subscriber lifetime contribution) − production cost.

Frequency should rise while that figure is positive and stop when it is not, which is a calculable point rather than a matter of taste.

In practice

Sends one to four in a month each produce roughly $765 of contribution and 59 unsubscribes, clearly worth sending at $386 of net value each.

Send five produces $610 and 75 unsubscribes: $610 − $254 − $180 = $176. Still positive.

Send six produces $470 and 105 unsubscribes: $470 − $355 − $180 = −$65. Negative, and this is where frequency should stop.

The optimum on this list is five sends a month, and it was found by measuring rather than by adopting a convention.

The limitations

The marginal figures decline unevenly rather than smoothly, and a well-timed sixth send around a seasonal event can outperform the third in an ordinary month.

Subscriber lifetime contribution is also an estimate, and the whole calculation is sensitive to it, using an annual value rather than a multi-year one is the conservative choice.

Putting it to use

Test frequency by splitting the list rather than by changing it for everyone. A holdout receiving one fewer send a month, measured over a quarter, gives a direct answer including the churn effect.

Then set frequency by engagement segment rather than globally. The engaged third of a list will usually support twice the frequency of the rest without additional churn.

Why most businesses under-mail rather than over-mail

The fear of annoying subscribers is more vivid than the arithmetic, and unsubscribes are visible while foregone revenue is not. That asymmetry pushes most programmes below their optimum.

Businesses that measure the marginal send usually find they could send more, particularly to their engaged segment, and that the additional revenue substantially exceeds the churn cost.

The exception is a list built on a promise of infrequency, where raising frequency breaks the expectation set at signup. There the constraint is the promise rather than the arithmetic, and the honest fix is at the signup rather than in the send schedule.

Preference centres offering a frequency choice recover a meaningful share of people who would otherwise leave entirely, and they let subscribers self-select into the tier that suits them.

That is usually a better mechanism than guessing at the right frequency for a whole list, since the same schedule genuinely does not suit everyone on it.

Seasonal peaks justify temporarily exceeding the calculated optimum, provided the frequency returns afterwards rather than becoming the new baseline.

Automated flows sit outside the frequency calculation, since they fire on behaviour rather than schedule and are welcomed at frequencies broadcasts would not survive.

Watching complaint rate alongside unsubscribes catches the point where higher frequency starts damaging placement rather than merely costing subscribers, which is a more expensive threshold to cross.