IV
The Attention Factory
6 What the factory makes
The factory’s product is not email. It is the reason to open the next one.
The factory produces four things, and none of them is a promotion.
The first is a publication. The Brand Digest is a short edition from the customer’s world, sent under the brand’s name: a few things worth reading, one thing worth doing and something worth keeping. A travel brand curates places. A jobs platform explains salaries and skills. A grocer explores food. The brand is the editor, not the protagonist. The Digest is useful on a day the reader buys nothing, which is most days.
The second is the Magnet: a thirty-to-sixty-second interaction that leaves the reader better off. A quiz on salary bands in their city. A prediction about a match. A puzzle, a benchmark, or a choice between two places to visit. The Magnet turns reading into participation, and participation into a signal that a real person is present.
The third is the object. A Magnet that leaves nothing behind is a moment, and moments do not form habits. A NeoCard gives the edition something to keep. It arrives by email with a permanent front and a back that can change. It may come with a Magnet, ask for a prediction before it is revealed, or carry a new challenge after it has been collected. Cards form sets, sets fill an album, and yesterday’s collection gives today’s edition another reason to matter. I described this in NeoCards: Objects Worth Keeping.
The fourth is the ledger. Mu, the attention currency, rewards completed actions under clear rules, and the ledger records what each reader has earned. The number in the subject line, the reader’s own balance, is a small promise that this email remembers the last one.
Each of these is expensive to build once and cheap to reuse. A quiz engine, a card system, a Mu ledger and an editorial workflow do not need to be rebuilt for each brand. What changes is the content: the places, the questions, the card designs and the category knowledge. Even content has a shared structure. A career Digest for one jobs platform and a money Digest for one consumer-finance brand draw on the same production line, staffed by editors working with AI and organised by vertical rather than by client.
This is where the portfolio does its work. A single brand would pay the full cost of the studio for one Digest. A hundred brands each pay a hundredth.
7 The way back
An open is a rumour. An action is evidence.
A factory that produces good editions still needs a careful way to introduce them to dormant customers. Sending a dormant list a daily Digest would be the fastest way to damage the domain the brand depends on. So the way back has two programmes, not one.
The first is Recover: a short, time-boxed sequence of lighter editions, called Minis, sent only to addresses that pass eligibility checks. They must be validated, unsuppressed and inside consent. Each Mini carries one strong reason to engage: a new card to collect, one question worth answering, or, where the brand has issued Mu before, a balance the reader had forgotten. The sequence ends with a hard exit rule. Readers who take no action by the end are suppressed. A dormant customer never receives open-ended Digest cadence. I set out the reasoning for this rule in 10+1 Email Innovations.
The way back then has three stages, in a fixed order. Recover finds the people who respond. An open inside Recover is logged as attention detected, but it changes nothing, because machines can generate opens. Graduation requires a verified human action: a Magnet answered, a card collected through a real interaction, a purposeful click or a stated preference. Graduation does two things. It returns the reader to the brand’s own marketing programme, at no charge. And it unlocks the Standard Digest, at a cadence the reader chooses. Habit is what the Standard Digest is then meant to produce: repeated, voluntary engagement, measured by Habit Rate, the share of readers who open at least 60% of the editions they chose over four weeks. Because opens are noisy, Habit Rate counts estimated human opens, with machine opens filtered out, and is checked against clicks, interactions and repeated returns.
The order matters. Reactivation qualifies a reader for the Digest. Habit is the Digest’s result, not its entry ticket.
Figure 3. Recover, Graduate, Habit. An open is never enough to graduate, and habit is earned inside the Digest.
Every Recover programme runs with a randomised holdout from the first day: a slice of the eligible dormant cohort that receives nothing. The holdout answers the question every brand should ask: how many of these customers would have come back anyway? Reactivation is counted as the difference between the two groups, never as the raw number in the treated group. Dormant cohorts allow a cleaner test than almost anything else in marketing, because the control group is drawn from the same people at the same time and left alone.
Recover is the bridge. It is deliberately narrow, short and suspicious of easy signals. Its job is to find the part of the dormant half that can still become an audience. The Digest’s job is to keep it. I described why customer retention begins with attention retention in Before the Slide.
