Own Your List

What You're Actually Getting

Four layers, all in your name. Not a login to our account — your accounts, your domain, your database, your subscribers. If you fired us tomorrow you would keep every one of them.

The path a single issue takes

Worth stating plainly: no language model can send an email in this system. The model writes and stages. A mechanical cron with no intelligence in it does the sending, and it only fires on a checkbox that a human ticked. Those two facts are the whole safety design.

Dedicated, not shared. You get your own Listmonk deployment and your own database, not a tenant slot inside ours. That is what makes the confirmation emails, the opt-in pages and the subscribe URL carry your brand instead of ours — on a shared instance those three are technically impossible to white-label, so we don't put clients there.

What It Reads Before It Writes

A newsletter that runs on a schedule with nothing to say is filler, and your list can feel it. Every issue starts by checking whether something real is worth telling people about.

Why the live research step exists

A language model's knowledge is months stale, and it does not know that. On a real business-credit issue the 30-day pull surfaced two things that changed the draft: the net-30 shortcut everyone teaches is being actively debunked right now, and the myth being hammered hardest is that an LLC hides your personal credit. Both would have shipped as confident, slightly-wrong advice under your name.

What it will never do

• Invent a story about you. No origin story, revenue figure, headcount, credential or award appears unless you supplied it. Your story bank starts empty on purpose and only fills with what you tell us.
• Invent a link or a price. Every URL is generated and tagged from a real record. Unpublished content does not get linked "in anticipation."
• Stack two asks. One call to action per issue. Two asks reads to a subscriber as no ask.
• Open with pleasantries. Never "Hope you're doing well." Every issue opens on the most interesting sentence in it.

Sounding Like You, Not Like AI

Your readers can tell. Increasingly they can tell fast, and the moment they decide an email was machine-written they stop reading it and start ignoring the sender. That is a conversion problem, not a taste problem, which is why this step is mandatory rather than optional.

Every draft goes through a two-pass humanizer before anyone sees it

Pass one is mechanical. A scanner flags the tells that give machine writing away: em-dash overuse, hype vocabulary, hedging, sycophantic openers, corporate filler. Pass two is by ear — three-fragment rhythm stacking, signposting phrases like "here's what you need to know," manufactured punchlines at the end of every paragraph. The draft is rewritten, then re-scanned. Nothing is staged until it comes back clean.

A real catch, not a hypothetical one: a first-pass draft came through with seven em dashes, a three-fragment stack, and two signposting phrases in the same section. All of it read as competent marketing copy. None of it read like a person. That is exactly the failure mode this step exists for.

The issue gets the shape its purpose needs

Every link is tagged, every logo is clickable

An automated gate runs before anything reaches you, and it hard-fails on untagged links, images that aren't hyperlinked, missing alt text, and any affiliate link without its required disclosure. Attribution can only be captured at the moment somebody clicks. It can never be backfilled, so an untagged link is data destroyed permanently, not data collected later.

Nothing Sends Without You

One checkbox is the entire difference between a draft and an email in 40,000 inboxes. Here is exactly who touches it and when.

1. The issue is drafted and humanized, then staged as a record in your JJOS dashboard — subject line, preview text, body, publish date, and the single call to action.
2. A real preview image is generated and attached. Not raw HTML. We render the actual branded email, screenshot it, and pin the image to the record, so what you review is what your subscriber will open.
3. The link and logo gate runs automatically. If it fails, the issue does not reach you until it is fixed.
4. Approved to Send sits unchecked. This is the whole gate. No automation, no model and no scheduled job ever ticks it. It is only ever ticked by a human being who was explicitly asked to.
5. You review and approve — in JJOS, in Rubric, or by replying to us. Edit the copy in place if you want; the record is the source of truth, not a document somewhere else.
6. A cron fires it. Every 15 minutes it checks for issues that are approved, dated for today, and not yet sent. It creates the campaign, sends it, and writes the campaign ID back.
7. The campaign ID is the re-send guard. Once written, that issue can never send a second time, no matter how many times the job runs.
8. Opens and clicks sync back into your dashboard and your Rubric KPIs, automatically.

The failure modes, and what happens in each

Send time is explicit. A date with no time would go out at midnight, which is the worst possible slot for opens. Your send time is set per brand at setup — 9am local is the default — and it is a field you can change without touching anything else.

Every Dollar, Itemized

These accounts are in your name and you pay the vendors directly. We never mark them up, and we never resell you capacity. Figures below are current as of July 2026.

Listmonk is open source under AGPL-3.0 — there is no licence fee and no per-subscriber fee, ever. Railway is $5/mo including $5 of usage, then metered per second (RAM $10.00/GB-month, CPU $20.01/vCPU-month, disk $0.156/GB-month, egress $0.05/GB); Listmonk plus Postgres sits around 0.5–0.7GB. Resend Pro is $20/mo for 50,000 sends and 10 domains. Domain: $0 because newsletters send from a subdomain of a domain you already own; a brand-new .com is $10.44/yr at cost if you need one. Docker: $0 because Railway builds the image on its own machines — you never install it. Docker Engine and the CLI are free software; Docker Desktop only becomes licensable ($9/user/mo) if you exceed 250 employees or $10M revenue and choose to build images in-house, which this setup does not require. Cloudinary: free tier is 25 credits/month; newsletter imagery uses a small fraction of one.

The dashboard layer runs on our infrastructure, not yours

Both are multi-tenant, so your marginal cost is genuinely zero rather than discounted. Isolation is enforced at the database layer, not hidden in the interface — we tested it by requesting another tenant's record directly by ID and got a hard refusal, not a blank screen. Read-only viewers are free, so give access to as many people on your team as you like.

The Cost Of Getting Bigger

Hosted platforms charge you for subscribers, whether you email them or not. Owned infrastructure charges you for emails actually sent. That single difference is what the table below is measuring.

Assumes one send per subscriber per week. beehiiv's Scale tier is $43/mo on annual billing and steps up at 2,500 / 5,000 / 10,000 / 25,000 / 50,000 / 75,000 / 100,000 subscribers to roughly $290 at the top; the intermediate figures are interpolated between published endpoints, so confirm at checkout before budgeting.

Read that last row honestly — above 50,000 the relay is the whole problem

Resend is priced for convenience, and past roughly 25,000 subscribers that convenience gets expensive enough to beat the point. The fix is to swap the relay for Amazon SES at $0.10 per 1,000 emails, which is a settings change in Listmonk, not a migration. At 100,000 subscribers that turns a $350 relay bill into about $43, and your total infrastructure from $378 to roughly $71 a month — against beehiiv's $290. We do this swap for you when the numbers cross over, and we tell you when they do.

What you are actually comparing this to

beehiiv at $90 a month buys you a place to send from. It does not decide what the issue should be about, research it, write it in your voice, strip the AI tells, build the preview, or report on what happened. The honest comparison is a sending platform plus a newsletter writer, which is where the real market rate sits — and that is the line the next section is priced against.

What We Charge

The all-in number

For comparison: a hosted platform at that list size runs about $90 a month and writes nothing. A freelance newsletter writer for a weekly issue is commonly $800–$2,000 and touches no infrastructure. This is both, plus the asset ends up in your name instead of ours.

What This Costs You Besides Money

Any ownership pitch that skips this section is selling you something. These are real, they apply to you, and none of them are solved by paying us.

Deliverability starts from zero

Hosted platforms send from IP pools with years of reputation behind them. On day one your domain has none, which means a 4–6 week warming ramp where volume climbs gradually. You cannot blast 40,000 people in week one, and anyone who says otherwise is setting up a spam-folder problem you will spend months undoing.

You inherit the compliance thresholds

Every relay suspends accounts sustaining a complaint rate above roughly 0.1% or bounces above 5%. We wire that handling at setup and watch it monthly — but the reputation being protected is yours, and a list you bought or scraped will destroy it faster than we can defend it.

You give up the monetization layer

beehiiv is not just a sender. It runs an ad network that brokers sponsors into your newsletter, paid recommendations that pay per referred subscriber, and paid subscriptions at a 0% platform cut. A 50,000-subscriber newsletter clearing $1,500 a month through that ad network makes the $200 platform fee irrelevant, and rebuilding it independently is a business, not a config file.

Where we would tell you not to do this

What Happens After You Say Yes

Book 30 Minutes

Bring your list size, your current platform and what you send now. You will leave the call knowing whether owning this is the right move for you — including if the answer is no.

Calendar not loading? Book directly at cal.com/jonjason.