Illustrative scenario
The publisher, the newsletter, the people and every figure on this page are invented so that a growth loop can be shown end to end. The page does not describe a real client, a real publication or the author's own results, and the monthly table is a worked exercise rather than evidence of what a referral programme achieves. Do not read any number here as a benchmark.
About this scenario
This scenario is invented. The newsletter, which the page calls the Marrowline Dispatch, does not exist, and neither do its publisher, its readers or its sponsors. The figures in the monthly table were made up to show a loop starting, stalling and compounding; they are not drawn from any real list. Nothing here describes the author's own work.
In the scenario, the Dispatch is a trade newsletter for people who run or work in independent bookshops: owners, buyers and front-of-house staff. It is published by a two-person company — an editor who writes it and a partner who sells sponsorship to publishers and suppliers who want to reach the trade. Revenue is sponsorship, so the value of a subscriber (LTVTotal value a customer brings over time., in the sense of the sponsorship income a reader's attention earns over their time on the list) depends on the reader being a genuine bookseller who opens and reads. That constraint runs through every decision: the loop has to bring in the right readers, not just more addresses, and anything that fills the list with people sponsors do not want to reach is growth the business cannot sell. The scenario covers six months, from the decision to build a loop to the point where referrals are the largest source of new readers.
The starting point
The list has been growing slowly for a long time. New readers arrive from a sign-up form on the publisher's website, from a stall at a trade fair twice a year and, the editor suspects, from forwarded issues, because a reader survey keeps turning up "a colleague sent it to me" as the answer to how people found it. Nothing in the set-up captures that last route. Forwarded copies carry no sign-up link of their own, nobody is asked to share, and the only call to action in an issue is to reply.
The sending set-up is also fragile. Issues go out twice a week — a news round-up early in the week and a long read at the end — from a shared sending domain supplied by the email platform, with no authentication records the publisher controls and no view of complaint rates. A large share of the list has not opened anything in months, and bounces are quietly accumulating. The metric the editor watches is open rate, which is a weak signal and getting weaker as mailbox providers pre-fetch images.
The growth lead's first job is to map the loop that already exists informally and to name the leak points: no mechanism for a forwarded reader to subscribe, no moment where a reader is asked to share, and no reason to. The second job is to pick a KPIA metric tied directly to an objective.: net new subscribers per issue, with supporting measures underneath it — referrers per issue, invites per referrer, invite-to-subscriber conversion (CVRShare of visits or clicks that convert.) and the retention of referred readers after a quarter. The growth loops in practice guide covers how to draw and instrument a loop like this one; the scenario applies it to a list that already had an unmeasured loop running.
The loop design
The mechanic. Each subscriber gets a personal referral link. It appears in a block near the top of the issue built for sharing, and in the sign-off. A forwarded issue now ends in a line inviting the recipient to subscribe, carrying the referrer's link so the referral is credited. The referral landing page shows what the newsletter is, who it is for and a recent issue, and asks for a confirmed opt-inA sign-up process in which the subscriber confirms through a link sent to the address given., which keeps the list clean and makes incentive abuse hard. The list building and consent guide covers why confirmed opt-in matters more on an incentivised flow than on an organic one.
The reward. Rewards are things only a bookseller would want: after a few successful referrals, access to a supplier contact directory the editor has been compiling for years; after more, a listing on the member-shops map that appears in every issue; after more again, a mention in the issue. No cash, no gift cards. The reasons are cost, abuse and audience quality: a cash reward attracts referrers who are not booksellers and referred readers who are not either, and sponsors pay for booksellers.
Onboarding. The welcome series tutorial shapes the first week: the first email delivers the thing the reader signed up for, the second introduces the editor, and only the third — after the reader has received a full issue — explains the referral programme. Asking before the reader has had any value produces few referrals and some unsubscribes.
Content and cadence. The long read at the end of the week is replaced by a short issue built to be forwarded: one practical artefact each time — a seasonal window-display planner, a returns-handling template, a checklist for an author event — with the sharing block at the top. The news round-up stays as it was. Among the distribution channels the publisher uses, the loop becomes the main one, but the website form and the trade-fair stall stay in place.
Deliverability first. Before any of this launches, the publisher moves sending to a dedicated subdomain it controls, publishes SPF and DKIM records and a DMARC policy, enables one-click unsubscribe, and sets up postmaster monitoring so that complaint rates are visible against the thresholds documented in the Google sender guidelines in the sources. The new subdomain is warmed gradually, starting with the most engaged readers. The deliverability guide explains each of these pieces; the scenario's point is their order. A referral loop sends more mail to more new addresses, and doing that from an unauthenticated shared domain with a list full of dead addresses is how a loop ends up in spam.
What happened
The monthly table below is invented. It is worth reading for three things: month 1 has no referrals at all because the loop had not launched yet, and month 2, the loop's first month, starts slowly, with 95 new readers from referrals against 190 from the website and events, because the welcome series only asks after the third email; the fourth month is net negative because the publisher deliberately removed inactive addresses; and from the third month referred readers outnumber every other source combined. The invite-to-subscriber conversion rate creeps up over the period as the landing page and the sharing block are revised.
Marrowline Dispatch: six months of invented figures
The loop launched in month 2, after a month of authentication, warm-up and welcome-series work. Month 4 includes a deliberate prune of 600 addresses that had not opened or clicked in six months and did not respond to a re-engagement email. "Referrers" counts subscribers who sent at least one invite that month.
| Month | Subscribers at start | New from website and events | New from referrals | Removed (unsubscribed, bounced, pruned) | Subscribers at end | Referrers | Invites sent | Invite-to-subscriber rate |
|---|---|---|---|---|---|---|---|---|
| 1 | 4,000 | 180 | 0 | 120 | 4,060 | 0 | 0 | — |
| 2 | 4,060 | 190 | 95 | 110 | 4,235 | 210 | 620 | 15.3% |
| 3 | 4,235 | 210 | 260 | 130 | 4,575 | 480 | 1,520 | 17.1% |
| 4 | 4,575 | 230 | 340 | 700 | 4,445 | 590 | 1,900 | 17.9% |
| 5 | 4,445 | 260 | 410 | 120 | 4,995 | 680 | 2,210 | 18.6% |
| 6 | 4,995 | 280 | 470 | 130 | 5,615 | 760 | 2,480 | 19.0% |
Referrals were 33% of new subscribers in month 2 and 63% in month 6. The click-through rate on the sharing block was 1.1% when it sat at the bottom of the issue and 3.8% after it moved to the top of the forwardable issue in month 3. Referred readers were still on the list after a quarter at 91% against 89% for website sign-ups, so the loop was not diluting the list. Sponsorship was sold at a flat rate per 1,000 subscribers per issue, which made each net new reader worth roughly £0.12 per sponsored issue. That is the figure the publisher used to decide that a cash reward of even £5 per referral would take the better part of five months to pay back, before allowing for the referrals that would have been fraudulent.
Taken as a pattern rather than as figures, the table shows a loop whose output depends on the number of people asked, the share who act and the rate at which invitations convert, and shows all three being worked on separately. The prune in the fourth month looks like a failure on the headline count and is the opposite: it protected inbox placement for the growing volume of mail the loop was about to send. The conversion rate calculator is the quickest way to see how a small change in the invite-to-subscriber rate compounds across the invite volume of a growing list.
Decisions along the way
Cash or no cash. The partner who sells sponsorship wanted a cash reward because it is simple to explain. The editor argued that cash buys the wrong readers, and the sponsorship economics in the example settled it: a referred reader earns the business a small amount per issue, so any reward has to be cheap to give and valuable only to a bookseller. The directory cost nothing to give away and was something readers had been asking for.
Where the ask sits. The first version of the sharing block sat at the foot of the issue, after the long read, and few people reached it. Moving it to the top of the forwardable issue, and making that issue short, raised the block's click-through (CTRShare of impressions that result in a click.) markedly. The news round-up kept its block at the foot, because readers of that issue are scanning headlines and resented the interruption.
The prune. Removing a large block of inactive addresses in the fourth month was argued over for weeks. The case against was the headline number and the sponsorship rate card, which was quoted per thousand subscribers. The case for was that complaint and bounce rates on the new subdomain were creeping towards the levels the sender guidelines warn about, and that a loop sending more mail to more new addresses needs a clean base. The prune went ahead after a re-engagement email gave inactive readers a last chance, and the sponsorship rate was re-based on engaged subscribers, which sponsors preferred.
Not adding a third issue. Once the loop was compounding, the temptation was to send more. The team declined: a third weekly send would have raised the ask frequency, lowered engagement per issue and risked the deliverability gains. Growth came from more readers, not more mail.
Watching referred retention. The team checked, each month, that referred readers stayed on the list at a similar rate to organic ones. Had they churned faster, the reward or the landing page would have been changed before the loop was pushed harder.
What generalises
Seven things carry over to other newsletters.
- The order of work. Authentication, list hygiene and monitoring come before the loop, because a loop multiplies whatever the sending set-up already is.
- The structure of the loop. A credited link in every copy, including forwarded ones; a moment to ask that comes after the reader has had value; a reward that only the intended audience wants; and a landing page that sets expectations and confirms consent.
- Three separable rates. Instrument the loop as the share of readers who refer, the invites per referrer and the invite-to-subscriber conversion, and work on them one at a time, because each has a different lever.
- One issue rebuilt to be forwarded. Rebuilding one issue around being forwarded, rather than adding sharing buttons to everything, suits most editorial newsletters.
- The economics, in form. Whatever a subscriber is worth to the business over time bounds what a reward can cost, and the bound is usually lower than cash-reward schemes assume.
- Checking the referred readers. Confirm that referred readers behave like the readers the business wants, rather than merely counting them.
- Taking the net-negative month. Be willing to accept a net-negative month to protect the asset the loop depends on, which is inbox placement.
What does not generalise
Six things should not be copied from this scenario.
- The figures and the smoothness. The figures are invented, and so is the smoothness of the growth. A real loop shows seasonality, campaigns that flop, months where a mailbox provider changes something and placement drops without warning, and referrers who refer once and never again.
- The invite-to-subscriber rate. It is a made-up quantity that depends on a landing page and an audience that do not exist.
- The trade-newsletter advantage. Trade newsletters have an unusual advantage: readers know other readers professionally, and a non-cash reward can be genuinely valuable because it is scarce. A consumer newsletter has a wider audience with weaker ties and fewer rewards that cost nothing, so its loop will be slower and its reward decisions harder.
- The funding model. A publisher funded by paid subscriptions rather than sponsorship values readers differently and might reasonably pay for referrals.
- The sender-guideline thresholds. Sender guidelines and the thresholds they contain change; the sources link to the current version, and this page avoids restating numbers from it.
- The cadence. One round-up and one short forwardable issue suited an audience that reads at work; another audience might want a single weekend issue.
What transfers is the discipline, not the design. The case studies library holds other labelled scenarios built the same way.
