Why the number matters
Several recurring email decisions are really questions about what one more subscriber is worth. How much can a paid campaign that collects sign-ups spend per address? Is a first-order discount for joining the list worth its cost? Which sign-up sources deserve more effort, and which should stop? How much net growth does the list need to reach next year's email target? The list growth rate entry makes the last point directly: once each subscriber is worth a known amount, the growth required can be calculated rather than guessed.
Without a figure, teams fall into one of two habits. Some treat every sign-up as free and valuable, so they buy addresses through giveaways and lead forms whose subscribers never buy. Others use the platform's attributed revenue divided by the list size, which flatters every subscriber and makes almost any acquisition cost look affordable. Both lead to a list that grows in size and not in value.
The figure here is different from the customer's lifetime value (LTVTotal value a customer brings over time.) in the LTV and CAC guide. That guide values a customer. Many subscribers never become customers, and many customers would have bought without the list. This guide values one more address on the list: the extra margin the business earns because the person is on it.
Attributed revenue is not value
The figures most programmes already report each measure something real, and none of them is a subscriber's value.
| Figure | What it tells you | Why it is not a subscriber's value |
|---|---|---|
| Email-attributed revenue divided by list size | Revenue the programme touched, per address | Revenue rather than margin; set by the attribution rule; includes purchases that would have happened anyway |
| Revenue per email | What a send earns under the attribution rule | Values a send, not a person; falls as frequency rises even when total revenue grows |
| Spend of subscribers compared with non-subscribers | How different the two groups are | People who sign up were already keener; the gap measures who joins, not what the email does |
| Customer lifetime value | What a customer is worth over a horizon | Values a customer, not an address on the list |
| Incremental contribution per subscriber, from a holdout | What being on the list adds, in margin | This is the figure the guide builds |
Three gaps separate the first row from the last. The first is margin. A pound of revenue from a discounted order with expensive delivery is worth far less than a pound from a full-price order, and the guide to contribution margin and break-even explains why contribution marginRevenue left after every cost that varies with each sale., revenue less every cost that varies with the order, is the honest basis for any marketing target. Discounts carried by email belong in those costs.
The second is attribution. The email metrics and testing guide explains that email revenue depends on the window and model chosen, and that a long window lets a newsletter claim orders other channels earned. The third is selection. Comparing subscribers with everyone else looks like a natural experiment and is not one, because the people who choose to join a list are already more interested in the business than the people who do not. Only a random split removes that difference.
Building the estimate from cohorts
Start with what subscribers are worth in total, before asking how much of it the programme causes. The cohort analysis tutorial builds the table: group subscribers into cohortsA group of customers who share a starting event and period, tracked together over time. by sign-up month, record each one's sign-up source at the time of joining, and follow each cohort period by period.
Four choices make the table usable for this purpose. The measure is contribution margin per original subscriber, so the cohort's orders are joined to the subscribers through the email address or a customer identifier, and each order's variable costs, including any discount, are deducted. The denominator is everyone who joined in that month, fixed at sign-up: people who later unsubscribe or never buy stay in it, so their lower value lowers the average instead of disappearing from it, the same reason the email frequency test fixes its denominator at assignment. The horizon is fixed and stated, usually the same horizon the business uses for lifetime value, with any projection beyond the observed periods labelled as a projection. And subscribers who were already customers when they joined are recorded as such, because their value behaves differently from that of people who joined before their first order.
The resulting curves show when value arrives, often early around a first order, and whether newer cohorts behave like older ones. They do not yet show what the email programme adds. A cohort of subscribers would have spent something without a single email, and the next section separates that baseline from the programme's contribution.
Measuring the incremental part with a holdout
The metrics guide describes a holdout groupA randomly chosen part of an audience kept from exposure to measure what exposure caused. for automated flows: a random share of the people who qualify are entered but sent nothing, and their outcome is compared with that of the people who received the sequence. The same design answers the subscriber question at the level of the whole programme. At sign-up, a random share of new subscribers is held out of marketing email for a fixed window. Everything else continues for both groups: the website, advertising, any transactional mail, and anything that was promised at sign-up, such as a code, which the held-out group still receives.
At the end of the window, compare contribution margin per assigned subscriber in the two groups, measured on the fixed denominators from the cohort table. The difference, per subscriber, is the programme's incremental contribution for that window: the part of a subscriber's value that the email programme causes, which is the incrementalityConversions caused by marketing. question applied to the list. The incrementality testing guide explains how to size the groups and read the uncertainty around the difference; for a small list, that may mean running the holdout across several monthly cohorts before the difference is larger than the noise.
Three cautions apply. The window has to be long enough for the programme's effect to show, which usually means covering more than one purchase cycle. The held-out group has to be kept out of marketing email only, not out of the business, or the test measures something else. And value beyond the window is an extrapolation: report the measured window as the figure and any longer projection beside it, labelled. Where a holdout is not possible, report the cohort figure from the previous section as the total value of a subscriber and say plainly that it includes purchases the programme did not cause.
Value by sign-up source
The entry for list growth rateNet change in an email list over a period as a share of its size at the start. notes that sign-ups from a checkout opt-in, a content download, a competition and an event behave differently. The value estimate should be read the same way, because an average across sources hides the source that pays for the others and the source that costs more than it brings. Repeat the cohort table by source, and where a source has the volume, read the holdout by source as well.
Where a source is too small to read its own holdout, apply the programme-wide ratio of incremental to total contribution to the source's cohort value, and record that this is an assumption. Then add the costs a source carries beyond its acquisition cost: subscribers who unsubscribe within weeks, or who complain, cost the list some of the deliverabilityWhether sent email reaches the inbox rather than spam or rejection, decided by sender reputation. every other subscriber depends on. A source whose subscribers rarely buy and often complain can carry a negative value even when its sign-ups look cheap. The list building and consent guide explains how the sign-up offer and the consent wording decide who a source attracts in the first place. Whether a source's sign-ups should be confirmed before they join can be tested with the double opt-in test, a design that reads email revenue per form submission rather than per subscriber. The email sign-up source register template gives each source a row where its cohort value and holdout result sit beside its wording, opt-in method and welcome branch.
Using the figure: incentives, paid sign-ups and hygiene
What to pay for a subscriber. The incremental contribution per subscriber over the horizon is the most the business can pay to acquire one before the sign-up loses money, and a target below it leaves room for profit. Use the figure for the source the money will buy, not the list average: subscribers from a paid lead form rarely behave like those who opt in at checkout. The break-even calculator turns margin and acquisition cost into the point at which the spend pays back, and the LTV and CAC guide explains how to read payback against the business's cash position.
Sign-up incentives. A discount for joining is a cost on every subscriber who uses it, including those who would have bought at full price, and it changes who joins. Measure its effect the same way: subscribers recruited with and without the incentive, followed on contribution margin, with the discount deducted. The exit-intent popup test applies the same logic to a sign-up placement, judging it on revenue per visitor including later email revenue rather than on the number of sign-ups.
Cadence. How often to send is a question about value per subscriber too, and the frequency experiment reads it on revenue per subscriber rather than per email for that reason. Link the two rather than mixing them: the value estimate tells you what a subscriber is worth under the current programme, and the frequency test tells you whether a different cadence would change it.
Hygiene. Suppressing long-inactive addresses looks like throwing value away. Read the value of that tier from your own cohort table: addresses that have not clicked or bought for a long time carry little of the margin the estimate measured, while the engagement they drag down is shared by the whole list. The cohort figures make the trade visible instead of leaving it to the instinct that a bigger list is a better one.
A cheese club's subscriber value (illustrative)
Curdlewick Cheese Club caps what it pays for a subscriber
Curdlewick Cheese Club, a monthly cheese subscription box, gains about 1,500 newsletter sign-ups a month. It follows twelve monthly cohorts for their first six months on contribution margin, after cheese, packaging, chilled delivery, payment fees and any discount used. The platform's report credits email with £21 of revenue per subscriber over six months; the cohort table shows £9.20 of contribution margin per mailed subscriber. For the last six cohorts, 10% of new subscribers (900 people) were held out of marketing email for six months while 8,100 were mailed; the held-out group, which still received its welcome code, produced £6.10 per subscriber.
| Sign-up source | Mailed subscribers | Contribution per mailed subscriber | Contribution per held-out subscriber | Incremental per subscriber |
|---|---|---|---|---|
| Website footer and checkout | 5,200 | £11.40 | £7.80 | £3.60 |
| Paid social lead form | 2,300 | £6.10 | £3.70 | £2.40 |
| Giveaway competition | 600 | £1.90 | Too few to read | Not measured |
| All sources | 8,100 | £9.20 | £6.10 | £3.10 |
The paid social campaign had been paying £3.40 per subscriber against £2.40 of incremental contribution in six months, so it lost money within the horizon. The club caps the campaign's cost per subscriber at £2.00, keeps the footer and checkout placements, and stops the giveaway, whose subscribers barely bought and unsubscribed fastest.
Common mistakes
- Valuing subscribers on attributed revenue. Revenue, an attribution rule and purchases that would have happened anyway are three overstatements in one number.
- Comparing subscribers with non-subscribers. The gap measures who chooses to join, not what the email does.
- Dropping people who unsubscribe from the denominator. The cohort's average then describes survivors.
- Using the list average for a paid source. The money buys that source's subscribers, not the average one.
- Breaking a sign-up promise in the holdout. The held-out group still receives what it was promised.
- Projecting far beyond the measured window without saying so. Report the window as the figure and label the projection.
- Quoting an outside figure. Another business's subscriber value describes another business's list.
- Treating the figure as permanent. Re-measure when the programme, the margin or the sources change.
The fundamentals hub sets out the list building, deliverability and segmentation work that decides how much value each subscriber can carry.
