What an email course costs: the sending, the domain, and your hours

Three line items, and the one with no invoice

An email course has three costs and only two of them ever arrive as bills. Sending recurs, metered either in messages that leave or in contacts held. A domain is a fixed annual charge in the low tens of dollars. The writing is hours, paid once before anything else can happen, and almost always the largest of the three.

Eight to thirteen hours is realistic for a five-day course by somebody who already knows the subject. The whole course is shorter than one long blog post, so the words are the small part: the time goes into deciding what the five days are, finding that two of them are the same day, and cutting one. Ten days is nearer three times that rather than double, because text scales linearly and structure does not.

Where those eight to thirteen hours go, for somebody not researching the subject while writing it:

• Deciding the promise and the five days, on paper, before drafting: one to three hours, and the ones that decide whether the rest are wasted.
• Drafting, at fifty to eighty minutes an email: four to seven hours across five days.
• Two read-throughs, one aloud and one in a mail client on an actual phone, plus the subject and preview lines: about two hours.
• The opt-in page and the confirmation email: one hour, and the only copy a stranger reads before deciding to trust you.
• Then an hour or two a year afterwards, which is maintenance rather than part of the eight to thirteen: a sequence naming a price or a rule keeps sending itself long after it stops being true.

The four ways sending is billed

Four billing models are in common use and they are not variations on one price. Per subscriber in bands, charging for contacts held whether or not anything went to them, which is the industry default. Per message sent, usual among sending APIs, where the meter runs only when mail leaves. A flat monthly fee, rarer, and usually with a ceiling written into it somewhere. And a free tier, almost always defined by a subscriber count rather than by a send volume. Which model is in front of you moves the annual figure further than any headline rate does.

A five-day course on confirmed opt-in sends six messages a person, counting the confirmation, so a thousand readers over a year is six thousand messages. That is a few dollars at usage rates, which is where per-message pricing looks unanswerable. Per-subscriber pricing charges for the same thousand people every month afterwards, because a course accumulates finishers: somebody who read day five in March is still a contact in November, still billed, and will never be sent anything again.

The defence of billing per contact is real, and worth hearing before that arithmetic is mistaken for a verdict. The charge pays for a consent record, a history of everyone who left or bounced, and an opt-in page and an unsubscribe that go on working long after the last lesson — none of which stops when the course does. It also buys a bill knowable a year ahead, with no overage to discover in the month a course goes unexpectedly well. A raw sending API is cheap precisely because it hands all of that back: rebuilt yourself, with an unsubscribe header to the shape RFC 8058 describes, it is a week of work and then a permanent obligation.

Two details decide the real number and neither is in the headline. The first is a monthly minimum: several widely used sending APIs sell plans with a floor around fifteen or twenty dollars a month, under which the message count is irrelevant and the year costs a hundred and eighty to two hundred and forty dollars whatever you send. The second is what a free tier does when it runs out — some stop accepting subscribers, which is a problem you can see, and some stop sending, which strands readers halfway through a sequence they were promised five days of.

The domain line, and what it does not buy

Ten to twenty dollars a year buys the name, the renewal is often dearer than the first year, and that is the whole dollar cost of a sending domain of your own — which is why it is the least interesting number on this page. What the money does not buy is the part that takes time. SPF, DKIM and DMARC records have to be right, and a new domain carries no reputation at all, so volume is ramped over a few weeks rather than switched on. Priced in hours that is an afternoon and then a habit of checking. Priced in dollars it is nothing.

The alternative is sending from the service’s own domain, which is what makes some tools instant to start. It costs nothing on the invoice and is charged in something else, and what that something else is belongs with the choice between tools rather than with the bill. One number from that side does belong here, though, because it is the one people size wrongly. Google’s sender guidelines ask bulk senders to keep spam complaints under 0.10 per cent and never to reach 0.30 per cent — three complaints on a single mailing to a thousand people, because the rate is measured against messages delivered rather than against the size of the list. Crossing it is free on the invoice and expensive everywhere else, which is the shape of most costs in this category.

What an hour on a trial settles about the bill

Every figure above is an estimate, and four of the things that decide whether the cheap option stays cheap are settled in an hour on a trial account, for nothing. None is on a pricing page, and each changes the annual number rather than the monthly one.

What to check before a card goes in, in the order that moves the bill most:

• Whether contacts are counted while they are stored or only in a month they were actually mailed. On a course, where most of the list has already finished, that single definition is what separates the two annual sending figures at the end of this page.
• Whether a monthly plan floor sits under the usage rate, because a per-message price of almost nothing is not the same thing as a per-message bill of almost nothing.
• What the free tier does at its limit: refuse new subscribers, or stop sending to the people already halfway through. Ask before the answer arrives as somebody’s complaint.
• Submit the form twice from an address that has already confirmed, then count what arrives. A second copy of day one is a duplicate, and on a per-message plan it is mail you paid for twice. A second confirmation request is not that finding at all — it is a tool dealing with somebody it has no consent for yet, which is a question about permission and is settled in the companion guide on confirmed opt-in.

What the dearer tools are actually charging for

Most of the price difference in this category is not sending. It is everything arranged around the sending, and the only honest way to value any of it is in the unit this format is short of, which is hours rather than dollars.

Integrations are the clearest case. A tool that can tag a finisher and hand them to whatever does the selling replaces a job somebody would otherwise do by hand; a tool without one charges half an hour a month instead, which is six hours a year, between half and three-quarters of what writing the course cost in the first place. Two years of that and the cheaper tool has cost more than the dearer one, in the currency that ran out first. The same arithmetic prices a writing surface that forces drafting somewhere else and pasting it in. What else separates one kind of tool from another — and which kinds will never have it at all — is a question about the product categories rather than about the bill.

The same course at three list sizes

What follows is illustrative and dated September 2026: plausible bands in United States dollars rather than any vendor’s prices. Take a five-day course on confirmed opt-in, so six messages a reader, and vary nothing but how many people join over twelve months.

The two sending prices below differ by one to two orders of magnitude at every size and both are correct, because they are not the same product: one is a pipe, the other is everything arranged around it. So the crossover is not a list size but whether a week or a year is the scarcer thing — and at a thousand readers the whole spread between cheapest and dearest is a few hundred dollars, the same order as the writing once the hours are priced at all.

The same five-day course at three sizes, in September 2026 orders of magnitude:

• A hundred readers a year, six hundred messages: usually nothing at all, inside a per-message free tier and inside most per-contact free tiers too. At this size the software price is not the decision.
• A thousand readers, six thousand messages: a few dollars for the year on pure usage, or a hundred and eighty to two hundred and forty if a plan floor of fifteen to twenty dollars a month applies; against ten to thirty dollars a month on stored contacts, which is a hundred and twenty to three hundred and sixty a year.
• Ten thousand readers, sixty thousand messages: tens of dollars for the year on usage; against something near a hundred dollars a month on stored contacts, twelve hundred for the year, and still climbing in year two whether or not anything is sent.
• The domain, at every size: ten to twenty dollars a year, the renewal often dearer than the first.
• The hours, at every size: eight to thirteen for a first course, unchanged by the length of the list, and the only figure here that never reaches a statement.