Segmentation, broadcasts and knowing it worked

After the welcome: silence or spam

The welcome sequence handles arrivals. Then most small businesses do one of two things to everyone who stays: go silent (the list decays — subscribers forget they subscribed, and the eventual reappearance earns spam complaints), or blast everyone identically (relevance decays — the discount-hunter and the devoted customer get the same email until both leave). This lesson is the small amount of structure that prevents both: two or three segments, one sustainable rhythm, a scorecard of numbers that mean something, and the discipline to let cold contacts go.

Segments that earn their keep

A segment exists for exactly one reason: you would genuinely say something different to these people. Not "could report on them separately" — say something different. Three ways of cutting that reliably pass the test:

• By source or magnet — Ana's Lesson 8.3 seed: guide-readers and discount-seekers joined for different reasons and respond to different sends.
• By lifecycle stage — prospect versus customer, the Lesson 8.1 map made operational. Tomas should not pitch the €80 Check-up to people mid-treatment.
• By behaviour — clicked the pricing link, bought twice, downloaded the self-check: actions that change what the next message should be.

Two or three segments. Twelve is a hobby — each one you add multiplies writing work and divides send sizes until nothing is learnable. If a proposed segment gets the same emails as another, it isn't a segment; it's a label.

A rhythm you can keep

Broadcasts are the emails you write fresh: the newsletter, the announcement, the seasonal offer. The rule is unglamorous: a cadence you can sustain for a year beats an ambitious one you abandon in March — not least because mailbox providers distrust silence followed by a blast (Lesson 8.4's steady-volume habit). Fortnightly with substance is a fine default; monthly, honestly kept, beats fortnightly abandoned.

And one bar for every send: would a subscriber be mildly disappointed to miss this? Not thrilled — mildly disappointed is achievable and sufficient. If the answer is no, don't send it. The list's attention is capital; a pointless broadcast spends it and buys nothing.

Metrics, held with tongs

Email is measurable — the reason Module 1 promised this module could attack your funnel's bottom — but one flagship number is now broken. Open rate is inflated by Apple Mail Privacy Protection, which pre-loads tracking pixels and records "opens" no human made. Treat opens as directional — fine for comparing subject lines against each other on the same send — never as a KPI.

The scorecard that means something:

• Click rate — interest you can act on, per send and per segment.
• Conversions per email — did the send cause orders, bookings, trial starts? Tagged links (your Module 6 UTM habit) are how you know.
• Net list growth — signups minus unsubscribes and suppressions. The honest size of the asset.
• Unsubscribe rate per send — watch your own trend, not a universal norm; a spike means that send missed.
• Spam complaint rate — the one that damages delivery for everyone remaining: keep it under the mailbox providers' 0.3% line, ideally under 0.1%.
• Revenue or leads per send — for the version of you that reports to a budget-holder.

The sunset policy

Some contacts go cold and stay cold. The sunset policy is the standing rule for letting them go: after N months without a click, send one honest re-engagement email — "Should I stop writing to you?" — then suppress the silent. It feels like shrinking. It is growth: engagement rates rise, sender reputation improves, delivery to everyone who remains gets better — and it operationalises the retention promise you wrote in Lesson 8.2, which until now was just words in a policy.

Ana, three months in

Ostara's list stood at 2,300, emailed quarterly. Three months of Module 8 later, it looks like this.

Segments. Two, by magnet source, plus the customer flag: guide-readers (and the organic pop-up signups), discount-seekers, and past customers. The Lesson 8.3 tags did the work in advance. On a typical broadcast, guide-and-organic clicks 4.1%; the discount segment clicks 1.2% — except on promotion sends, where it wakes up. So Ana stops sending new-collection broadcasts to the discount segment entirely and sends it two promo emails a season instead. Unsubscribes drop; nobody gets email they've demonstrated they don't want. That is segmentation earning its keep: one different thing said to one group, for a reason the data shows.

Rhythm. Fortnightly to the guide-and-organic segment, mostly derivatives of the Module 5 content engine — which was already designed to feed email. Sustainable inside Mateus's twelve hours.

Sunset. Applying the 8.2 retention rule for the first time: 540 contacts with no click in 18 months and no purchase get the re-engagement send. Thirty-seven click to stay. Five hundred and three are suppressed. The list shrinks by more than a fifth, Ana winces — and every rate on the scorecard improves from the next send, because the deadwood was dragging every denominator. The list now fits its ESP tier, too. Shrinking the list is the most counter-intuitive act in this module, and the one that separates people who read this lesson from people who did it.

Back to the baseline

Open your Module 1 funnel baseline — the page this Program promised every later module would attack. Lesson 1.6's exact words: Module 8 attacks the bottom. Now you can see the mechanism.

Your list is a new stage in the funnel: a reservoir between found us and bought, where people who weren't ready to buy wait — on your terms, at €0 per contact — instead of vanishing. That gives your baseline new numbers with real rates between them: visitors → subscribers (is the magnet converting?), subscribers → customers (is the sequence working?), customers → repeat (are the broadcasts causing anything?).

For Ostara, the baseline said 6,500 sessions → 91 orders → 19% ever order again, and the pack called the bottom a hole. Email-attributed orders per send, measured by tagged links, is the first instrument Ana has ever had pointed at that hole — around eleven tagged orders on a typical fortnightly send, against a 91-order month. For Herzog, the baseline's leak was 0.75% of visitors enquiring; the self-check magnet gives the other 99% something smaller than a booking to say yes to, and subscriber → Check-up booking becomes a rate he can watch. For Storkflow, the welcome sequence's E4 produced four trial starts in its first six weeks — small, attributable, and incremental to a 41-trial month.

None of these numbers is dramatic. All of them are owned: they'll be there next quarter, unchanged by any algorithm, compounding as the list grows. That is what you built this module. Finish the plan below, and ship the project.

---