The inbox
is a surface.
Email is where the brand most often loses its voice. Customers receive ten emails a day; ours is the one that reads like a magazine, not a marketing tool. Signature, transactional templates, Journal campaigns, list hygiene · the rules below cover every kind of send the brand makes.
Six rules that make an email feel like ours.
600 pixels wide. One column.
Email is not a website. Single column, 600 px max, fluid below 480. No sidebars, no two-column blocks, no carousel.
Georgia and Helvetica. That is the type stack.
Fraunces and Euclid do not render in most inboxes. The safe substitutes are Georgia (for the serif voice) and Helvetica Neue / Arial (for the sans). Use them with the same pairing rules as the web.
No tracking pixels. Open rates are not the metric.
No 1×1 transparent pixels. No UTM-bombed links. Quote-rate, reply-rate, and citation-rate are the brand's metrics; open-rate is the SaaS metric, and the SaaS metric is not ours.
No images in the body. Text first.
Hero images get stripped by image-blockers and inflate send weight. The brand mark renders as text. The one exception: a single retina-doubled artefact in a Journal campaign, sized to load under 80 KB and captioned in alt text.
Plain-text twin, hand-written.
Every send ships a plain-text twin generated by a human, not stripped by the ESP. Plain text is the brand at room tone · easier to write, harder to fake, more likely to be replied to.
Reply-to is a person.
Never noreply@. The brand sends from a named human; the inbox they reply to is a named human's. If we cannot answer the reply, we did not have the right to send the email.
One block. No social icons. No quotes.
No avatar. No company logo image. No social icons. No motivational quote. No "sent from my iPhone".
Eleanor Mansfield Editor, The Journal BetterReviews eleanor@betterreviews.io betterreviews.io MMXXVI · By application
Five lines. ASCII separator dot. No fancy underscores or pipes. Reads identically in every client from mutt to Apple Mail.
Every send shares the same skeleton.
Named human · eleanor@betterreviews.io · never noreply@.
Sentence case. 40 to 65 characters. No emoji. No exclamation marks. Cross-reference 02 \u00b7 Voice microcopy gallery, row 05.
A single complete sentence. 80 to 130 characters. Continues the subject; never repeats it.
Smallcaps in Helvetica · Terracotta · 10px · 0.22em tracked. Names the template family.
Georgia 700 · 30px · one line. Sets the rhythm.
Georgia 400 · 16/1.6 · up to four short paragraphs. Long sentences earn their length here, the same way they do in the Journal.
One pill button or one underlined link. Never both. Never a row of buttons.
As shown in section II. Hairline rule above.
Helvetica 10px smallcaps. Three lines max: address, unsubscribe, why-you-received-this. Centred.
The most-sent template, rendered in full.
NO. I · APPLICATION RECEIVED
Your application is in the queue.
Hello Maud,
Thank you for the note about Maud's. We answer every application by hand, in the order it arrived, and we read each one before we write back. That takes us a week, on average. Sometimes longer.
If we offer you a place in the cohort, the reply will come from this same inbox, and it will tell you exactly what happens next: the install, the cohort calendar, the editorial review schedule. If we do not offer you a place this round, we will tell you that, too.
In the meantime, two things you can do without us:
Read The Refusal, the longest piece we have written about what BetterReviews is and is not. And read What Maud's Did, and What She Stopped Doing · the case study a different Maud might appreciate.
A note on time: if anything in your store changes between now and when we reply, send a one-line update. The freshest version is the version we want to read.
Read The RefusalUnsubscribe · You received this because you applied to the cohort.
Four kinds of email the brand sends.
From: Eleanor · To: applicant
Acknowledges the application, sets the timeline, points the reader at the Journal. Sent within 30 minutes of submission. The most-read email the brand sends.
Cadence: one send per applicant. No drip.
From: Eleanor · To: accepted merchant
After acceptance: cohort calendar, install link, the editorial schedule, the name and email of their assigned reviewer.
Cadence: one welcome, then a four-email onboarding over two weeks.
From: the engine · To: merchant
When a customer's review is cited by an answer engine: a quiet note with the quote, the engine that cited it, and the buyer's first name.
Cadence: digest, sent weekly on Mondays. Never per-event. The brand is patient.
From: Eleanor · To: opted-in readers
Long-form pieces from the Journal, delivered in full email body. Never a "click to read more" teaser. The reader who opens the email gets the whole essay.
Cadence: two per month at most. The Journal is published on the calendar of the writing, not the marketing.
The full piece, in the inbox.
The Journal newsletter delivers the article body, not a marketing teaser. That single decision changes everything about the send.
The reader who opens the email does not click through to a website. They read the essay, in serif, full length, in the inbox. The footer of the email links back to the Journal index for the back-catalogue, but the current piece is fully present.
Reply-rate is the campaign metric we track. A Journal email that earns one reply per 200 sends is performing. Anything over one in 100 is exceptional.
Kicker · Issue number, date
02Title · the article H1, verbatim
03Dek · one sentence, italics-free
04Byline · name, role, read time
05Body · full essay, no truncation
06Sign-off · \u00b7\u00b7 fin
07Footnotes · sources
08Three related reads from the back-catalogue
09Signature \u00b7 footer
The quiet weekly note that wins the relationship.
DIGEST · 4 \u00b7 5 \u00b7 11 MAY 2026
Six citations
this week.
Your customers' words showed up in four answer engines, on six different queries, between Monday and Sunday. Two of those queries are new this week.
Most-cited sentence
"doesn't strip everything, smells like rosemary not perfume"
Eleanor M. · 14 February 2026 · cited 4 times this week (ChatGPT, Perplexity, Google AI Overviews, Claude).
One to read
Frances L. left a short review yesterday morning. It uses a phrase ("face wasn't tight at all") that overlaps with the cluster ChatGPT is currently quoting from. Worth a reply, in your own words, in the next 48 hours.
Open the reviewUnsubscribe · You receive this every Monday at 09:14 because Maud's is in cohort I.
Smaller, replied-to lists. Always.
Double opt-in. No exceptions.
Single opt-in leaks language across the brand and earns no replies. Double opt-in keeps the list dense.
90 days of silence, off the list.
If a reader does not open three consecutive Journal sends, we remove them. We do not run "we miss you" campaigns.
Two per month, maximum.
Across all opted-in lists combined. If two sends collide in a calendar fortnight, one of them was the wrong send.
Patterns the inbox lets through but the brand does not.
Don't send
- Emoji in subject lines
- Hero image as the first thing under the meta line
- "View this email in your browser" header
- Multiple CTAs in a single send
- Open-tracking pixels
- "P.S." in the body
- Drip flows over five emails
Do send
- Sentence-case subjects, 40 to 65 characters
- Title-Then-Body, in serif
- A plain-text twin, hand-written
- Exactly one action per send
- Reply-tracking via the inbox, not the ESP
- A second short paragraph if the first did not say it
- Four-email onboarding, then silence
What we send with, and why.
Postmark · chosen for deliverability, sub-second send, no tracking by default.
Buttondown · plain-text first, paid plan disables open-tracking and click-tracking at the ESP level.
Front, configured so every Journal reply lands in Eleanor's queue and is answered within 48 hours.
All three configured at strict policy. The brand is not in a position to be spoofed.
Mailchimp template builders. HubSpot sequences. "AI-personalised" subject-line generators. Pixel trackers. URL shorteners.