Welcome series

What a welcome flow actually is

A welcome flow is a promise being kept. Someone handed over an address in exchange for something specific: a code, a guide, a quiz result, entry to a draw. Everything the flow does afterwards is borrowed against how well it keeps that first promise.

The one rule: map the promises before you write a word. A discount-popup subscriber, a lead-magnet subscriber, and a giveaway entrant were each promised different things. A single generic email one breaks two of those three promises at the exact moment attention is the highest it will ever be.

How to think about the sequence

Five beats, decided before any of them is written: keep the promise, say why you exist, show proof, answer the objection, make the honest ask. The order is not decorative. Proof comes before the ask, and the objection before it too, because an ask that arrives before the reader has a reason to believe you is the one that earns the unsubscribe.

• Beat one is not a brand introduction. It delivers the promised thing in the first screen. The brand can wait one email; the code cannot.
• Sequencing compounds. Every beat inherits the attention the previous one earned, which is why a weak beat two costs beat five more than it costs itself.
• Exits are part of the sequence. A purchase exits the flow into post-purchase. Nothing reads worse than a ten percent code the day after someone paid full price.
• So is suppression. Campaign sends stop while a subscriber is in the early beats. Two competing emails in one day is how welcome flows earn their unsubscribes.
• The discount decision is a pricing decision, not a copy decision. Opening with a code teaches every future subscriber that signing up means a coupon, and turns every full-price sale into a small failure of patience. Free shipping over a threshold, a strong guarantee, or content that genuinely helps all keep the margin. A discount is right when the popup promised one, when the average order value absorbs it, or when the objection really is price. Decide it once, deliberately, in the Sequence phase.

Failure modes worth naming: one generic blast instead of a sequence; a discount handed to everyone whether they needed it or not; beat one delayed an hour by an ESP setting nobody checked. And the quiet one, a flow written once, launched, and never read against its own numbers again.

Numbers to hold it against

Welcome flows beat campaign sends by default, because they reach people at the one moment those people asked to hear from you. The honest questions are by how much, and where the flow leaks.

• Email one commonly opens between 50 and 60 percent. Below 40 percent, read it as a deliverability or timing problem before touching the subject line: something is landing in spam or arriving too late. Remember that Apple Mail Privacy Protection inflates every open rate here as it does everywhere, so treat the number as directional.
• Clicks on beat one usually run several times a campaign send, often in the 8 to 15 percent range, because the reader is collecting something they asked for. Beats two to five settle back towards your list baseline, and should still beat it.
• Revenue per recipient is where the flow justifies itself. A welcome flow routinely earns many times what an average campaign send earns per recipient. Read that as the floor for what good looks like rather than as proof your copy is good: these are the newest and warmest addresses you have.
• Automated flows commonly produce a quarter to a third of email revenue on a small fraction of the sends, and the welcome flow is usually the largest contributor after abandoned checkout. That ratio is the argument for spending real time here rather than on the next newsletter.
• Stay well under a 0.3 percent spam complaint rate, the threshold Gmail and Yahoo now enforce, and treat 0.1 percent as the working target. Unsubscribes between half a percent and two percent per beat are normal; a beat running above three percent is indicting itself.
• Three to six emails across seven to fourteen days covers the beats without wearing out the welcome. Longer is defensible when the considered purchase cycle is genuinely long.

All of these move with list source, category, and price point. Your own last flow is the better baseline; published ranges are for sanity-checking, not for targets.

What has to be true before an email asks for approval

• The promised thing appears in the first screen of beat one, before any brand talk and before any scroll.
• One primary call to action, repeated. A welcome email with six links teaches nobody anything.
• The subject line and the preheader do different jobs. A preheader that restates the subject line is a wasted line of inbox real estate.
• Expectation-setting lands in beat one: what you send and how often. It is the cheapest unsubscribe prevention there is.
• Every image carries alt text, and the email still reads correctly with images off.
• Any discount states its terms, its expiry, and its exclusions in the email itself, not only on the landing page.
• No image placeholder and no unresolved bracket survives into approval.

The seed cards are the first instance, not the ceiling

This board ships with six cards to show the shape. A real welcome programme is denser than that.

A healthy Map phase holds one entry point card per real source, usually two to five. Sequence normally holds one card, or one per source when the flows genuinely diverge. Write is where the count grows: one email card per beat per variant, so a five-beat flow whose first beat differs by source is six or seven cards, not two. Measure holds one result card, revisited monthly, plus a card for any cohort worth reading separately.

If the Write phase holds two cards and the flow is already live, three beats are running from somebody's memory of what they meant to send.

The layout is one arrangement of many

The phases here run by production step: map, sequence, write, launch, measure. That works because the hard thinking has to happen before the writing does. If you run genuinely separate flows per entry point, arranging the board by source instead is a better fit, and the method is unchanged by the move. Phases can be renamed, added, and reordered; the promise-keeping logic survives all of it.

When this is the wrong playbook

Subscribers who arrive by buying belong in a post-purchase flow, not in a pitch to buy what they just bought. A purchased or scraped list cannot be fixed by any flow, and this one will burn the sending domain the rest of the programme depends on. If signups run at a handful a week, build the thing that drives signups first: the flow will be starved of the volume it needs to read its own numbers. And if the real problem is that nobody can say why they should buy from you at all, that is a positioning problem, and beat two will keep coming back blank until it is solved.