Email course software: the four kinds, and what separates them

Four kinds of product answer to the name

A general email service provider with per-subscriber automation is the commonest answer: the course is one workflow among many, built from a trigger, a wait and a send, in an account that also does broadcasts, segments and forms. A creator platform bundles the list with a landing page and a checkout, so the course is a feature inside a larger product and the bundle is the reason to stay. A learning platform emails lessons as a side effect of hosting them. A dedicated course tool does the sequence and close to nothing else.

The fourth answer is to build it — a sending API, a subscribers table, a scheduled job — a weekend of work and then a permanent obligation to everything further down this page: confirmation, bounces, unsubscribe headers, retries, the export somebody eventually asks for. All four put five emails on five mornings. What separates them is how much of the rest of your publishing ends up in one account, and how much of the machinery underneath it you are expected to own yourself.

They meter the same job in four shapes, and the shape decides more than the headline figure does: per subscriber in bands, which is the industry default; per email sent, used by credit-style senders and transactional APIs; a flat monthly fee, rarer and usually with a ceiling written into it; and a free tier, almost always defined by how many addresses are stored rather than by how much is sent. Which shape suits a course depends on how a course's list behaves over a year rather than on the number on the plan, and the companion guide on what an email course costs works that through with figures.

Timing measured from the reader, not from a date

Day one has to be relative to a person. Somebody who confirms on Thursday gets day one on Thursday and day five on Monday, and somebody joining three weeks later gets the same five days three weeks later. Five broadcasts a day apart works exactly once, for the people already on the list, and everyone arriving on Wednesday starts at day three of something whose beginning they never saw. Automation, sequence, drip, journey: the vocabulary settles nothing. What settles it is whether each step's delay is counted from that reader's own entry, and whether the reader's timezone was captured at sign-up — a tool that knows only the sender's clock delivers day two at three in the morning to anyone who is not local.

The cost of a tool built only for sequences is the thing it cannot do: no announcement, no correction to the people who already passed day three, no note to finishers when a second course goes up. A provider that does both jobs adequately can beat one that does a single job properly, and which trade is right turns on whether you expect to have anything to announce.

Every door an address can come in through

Confirmed opt-in is offered by everything, so the buying question is not whether it exists but which paths go round it. A list grows through more than one door: the hosted form, an embedded form on your own site, an API call, a spreadsheet import, a manual add, and whatever an integration with a shop or a webinar tool pushes in. A service can enforce confirmation on its own form and still take an unconfirmed address through the other five.

Enforcement on every door has a cost worth stating. A list you built yourself, with records, on a previous tool cannot be carried in as it stands; it has to be re-permissioned, and that campaign returns a fraction of what it started with — the companion guide on confirmed opt-in puts a number on how many never click a confirmation they asked for seconds earlier, and a re-permission campaign against a list that has not heard from you in a year does considerably worse than that. The loss falls on somebody who did nothing wrong, which is the strongest case for a tool that lets you assert consent you already hold. The case on the other side is about who else is sending: where every customer shares one domain, a confirmation step with no switch restrains the account that would otherwise degrade delivery for everyone, yours included.

Ask the same question in the other direction: what an unsubscribe removes. This list, your whole account, or every course on the service. A shared suppression list is what makes shared-domain sending survivable, and it also means a reader who leaves a stranger's course is gone from yours before they ever saw it — and on most of these services, gone for good. That is not a reason to avoid a shared service, and it is not a defect anyone is hiding. It is a thing to know before a launch is counted, because it is never written on the page that sells the plan.

Whose name, and whose domain, the mail goes out on

Three things travel with every lesson: the name in the From line, the address a reply goes to, and the domain the message is authenticated against. A course invites an answer far more than a broadcast does, and a reply from a reader on day three is close to the best thing that can happen to one. Where it lands was decided for you — your own mailbox, a queue the service reads first, or a noreply address that teaches a reader there is nobody there — and it is settled by replying to a real message, not by reading a settings screen.

The domain is the larger question. The service's own starts you in minutes with no DNS records and no warm-up, which matters when the alternative is a fortnight of not publishing; the price is a reputation shared with every other customer, and a failure mode that is quiet: delivery sags for reasons you cannot investigate. Your own costs SPF, DKIM and DMARC records and a few weeks of ramping volume, and returns a reputation that is yours, problems you can see, and an audience that moves with you.

Ask first whether your own domain is on offer at all, or only the service one. A tool that sends every customer from a single domain is not giving you the cheaper half of a choice — it is a ceiling. Nothing built on that domain is portable, the decision cannot be taken again when the list is finally worth moving, and none of it shows until the list is big enough to want the move. That answer belongs in the first five minutes of evaluating anything, because it is the one axis a later upgrade cannot fix.

The floor underneath both is written down rather than folklore. Google and Yahoo's bulk sender requirements, in force since February 2024, ask for SPF, DKIM and DMARC authentication, one-click unsubscribe in the message headers rather than only in the footer, and a low rate of spam complaints. Google's sender guidelines put numbers on that last one: keep spam reports below 0.10 per cent of the mail Gmail accepted from you, and never reach 0.30 per cent — a narrower band than the single figure usually quoted. The rate is reported per sending domain, so on a shared one only the service can see it, blended across everybody on it.

The parts a trial account cannot show you

Some of what matters here is invisible from outside, and a page that pretends otherwise is selling something. A send is a request to a mail provider, and the failure that decides whether a tool is trustworthy is not the one that returns an error; it is the request that times out after the provider already accepted the message, where the sender cannot know whether it went. Sent late is an inconvenience. Sent twice is a complaint. The shape that prevents it is a uniqueness rule held in the database rather than in a code path, plus a sending key byte-identical on every retry. No vendor publishes its schema and every one of them says retries are handled.

Two questions about it do have checkable answers. The first is whether a duplicate is prevented by the database or by the sending code: the reply either names a mechanism in one sentence or describes a promise. The second is what happened the last time their provider timed out, because anybody who has lived through one describes it in detail. A third needs no account at all: mail providers only remember a de-duplication key for roughly a day, so a message already handed over cannot be pulled back, and a retry arriving after that window, or a pause longer than it, resolves as either a repeated lesson or a missing one. Tools that have hit the case choose the missing one and say so; a tool that promises a pause of any length with no consequence has not hit it yet.

Where a dedicated course tool is the weaker choice

A narrow tool is missing things a general provider has had for a decade, and a comparison that leaves them out is not a comparison. A template gallery and a drag-and-drop layout editor. A visual builder with branching, so day four differs for the people who clicked day three. Tags, segments and conditional content. Native integrations with a shop, a webinar or a CRM. None of it is decoration: branching earns its complexity for anybody whose next step depends on what a customer already did, and a bundled checkout earns it for anybody selling the course rather than giving it away.

Measurement is the other axis, and narrow tools measure less on purpose. Open rates stopped meaning much once mail clients began fetching tracking images for readers who never opened anything, and click tracking rewrites every link in a message. What survives is completion: the sequence ran to the end, or it stopped at day three because somebody unsubscribed or the address stopped accepting mail. That is a record of leaving rather than of reading — somebody who receives all five in silence counts as finished — but no mail client can inflate it, and the day it clusters on is the day to rewrite. The question every tool answers by default and few tools raise is what day six is: nothing, one note saying the course is over, or a door into something else.

What you can take with you when you leave

Three things are yours and only one is easy to get back. Every tool exports addresses. The consent record is harder: GDPR Article 7(1) requires a controller to be able to demonstrate that consent was given, and does not prescribe what the demonstration looks like. The familiar trio — the wording somebody agreed to, the date, and something identifying the submission — is supervisory guidance and ordinary practice rather than the text of the regulation, which is why tools differ so widely here. A service that stores a subscribed flag and nothing beside it has made a list non-portable in the sense that matters when a complaint arrives two years later.

The third is the suppression list: everyone who unsubscribed, bounced hard or filed a complaint. Moving without it means the first send from the new tool reaches every person who already asked you to stop, the worst available opening act for a fresh reputation. It is rarely offered beside the subscriber export and usually has to be asked for by name. And where the opt-in page sits on the service's domain, that URL belongs to the service — fair for a first course, expensive for a page that has been collecting links for a year.

How to test one in an afternoon

Every claim above can be checked on a trial account, for an hour of setup and a day of waiting. A comparison table cannot be checked at all, because it is written by the people being compared. Use two addresses at two different mailbox providers, put a real course behind them rather than five messages saying test, and read what arrives on a phone.

The rehearsal, in the order that finds problems earliest:

• Join with the first address today and the second tomorrow. The second should receive day one. If it receives day two, the tool is scheduling from a date and the trial is over.
• Put the second address in a timezone several hours from yours before it joins, then look at the clock on day two. Arrival hour is the thing no interface describes honestly.
• Submit the form again from the first address after it has already confirmed, and count what arrives. A second copy of day one is a duplicate. A second confirmation email is not — resending one to an address that never confirmed is reasonable, and a tool that rate-limits it is being careful rather than sloppy.
• Reply to day one, and find out where the reply landed, how long it took to reach a person, and whose name was in the From line.
• Unsubscribe from the second address on day three, using the mail client's own unsubscribe button rather than the footer link. Days four and five should never come.
• Ask for the export before the trial ends, and ask in writing whether a duplicate send is prevented by the database or by the sending code. Addresses alone or addresses with the consent wording, the timestamps and the suppression list; a mechanism or a promise — the two replies together are the answer.