How to promote a webinar: fourteen days to a full room.

Count who you can actually reach before you pick channels.

Promotion advice usually starts with tactics. Start instead with arithmetic, because it tells you within ten minutes whether the plan you are about to make can possibly produce the room you want.

Write down, per channel, the number of people you can genuinely put this in front of: the portion of your list that opens things, the partners who have actually said yes and what each of them reaches, the size of any community you run, and the individuals you are willing to message personally. Not the theoretical audience — the reachable one.

Then resist the urge to apply a conversion rate you read somewhere. Published webinar benchmarks average across industries, price points and traffic sources that have nothing to do with yours, and borrowing one produces a forecast with false precision. For a first session, the useful question is not “what percentage will register” but “if this performs poorly, is the room still worth running?” If the honest answer is no, the fix is upstream: more partners, a longer window, or a smaller expectation.

What you are promoting has to work first

Every channel below points at one page. If that page does not make the trade obvious — a specific title, the date and time in the visitor’s timezone, three to five things they will be able to do afterward, and a form asking for as little as you can defend — then promotion multiplies a problem rather than solving one. Build it before you send anything; the Build stage of the pillar guide covers what belongs on it.

Three sends, and none of them is the same email twice.

Your list is the channel you control completely, which is exactly why it gets used carelessly. Three sends, each with a distinct job, with registrants removed before every one.

1. 01The announcement (T−14d). The only send allowed to be about the event. Title, date, time with timezone, three outcomes, the replay policy stated plainly, one link. Send it the day the page goes live.
2. 02The different angle (T−7d). Leads with the problem and mentions the session second. It has to be worth reading by someone who never registers — if it is not, you are sending the announcement again with a new subject line, and your list can tell.
3. 03The last call (T−1d). The send people cut, and the one with genuinely new information in it: this is happening tomorrow. Most of the remaining registrations arrive in this window.

One resend of the second email to non-openers, with a different subject line, is a reasonable fourth touch. A fourth distinct email is usually not — by then you are adding volume rather than argument.

Segmenting registrants out

Anyone who registers should stop receiving invitations immediately and start receiving the confirmation and reminder sequence instead. Two clocks, two audiences. Nothing reads as careless faster than being invited to something you already signed up for, and on a list of any size this has to be automatic rather than a manual export the night before.

One more piece of hygiene: if a large part of your list has not opened anything in months, sending all three invitations to all of them can cost you more in deliverability than it returns in registrations. Send to the people who open things, and treat the dormant segment as a separate decision.

All three sends, plus the confirmation and reminder sequence that follows a registration, are written out in full in the webinar email sequence templates.

Put the link where a click already happens.

Owned surfaces cost nothing and are usually worked hardest by the people who need them least. The trick is not to post more — it is to place the link on the surfaces that already get traffic for other reasons.

• →Your email signature. Every reply you send for two weeks carries the invitation. It takes one minute to set and one minute to remove.
• →A pinned post on each platform you actually use. One is enough per platform. Pinned beats posted, because it works for the whole window rather than for an afternoon.
• →The top of any community you run. Announcement post, and one useful post later in the window that happens to mention it. The second one usually does more than the first.
• →The confirmation and thank-you pages of everything else you run. Someone who just downloaded something or bought something is already engaged and already looking at a page with nothing else on it.
• →A site banner and the relevant blog posts. If you have articles on the topic, the invitation belongs in them. That traffic is already problem-aware, which is the whole reason it converts.
• →Your newsletter or podcast, if you have one. Not as a dedicated send — as a line in the thing you were sending anyway.

Two posts over fourteen days is a sensible default per surface: the announcement, and one later post that teaches something and mentions the session at the end. The second post is the one that works, for the same reason the second email does.

You are borrowing trust, not buying attention.

For a session without an audience behind it yet, this is the channel with the most room in it — a recommendation from someone the reader already trusts does work that no amount of your own copy can.

It also has the longest lead time, because it runs on somebody else’s calendar. Ask before the fourteen-day window opens. A partner who would happily have promoted you cannot rearrange their sends on three days’ notice.

Who to ask

People who have seen you teach. A partner who has not is being asked to lend credibility they cannot verify, which is why those asks get polite non-answers. Serve the same audience without competing for the same purchase, and be specific about what their audience gets — a promotion request framed around what you need is the easiest kind to decline.

Offer the reciprocal, and mean it. The partners worth having are the ones you will still be trading with at session six.

The yield is inversely proportional to the work you leave them

Almost every partner who says yes and never sends had every intention of sending. What stopped them was a blank page. Send a kit that reduces their side of the job to a paste and a click, and ask for a window rather than a date so it can slot into their calendar rather than fight it.

Twenty messages that could only have been sent to that person.

The least scalable channel and the one that most reliably fills a first room. It also produces the attendees who ask the questions worth answering live, which makes it an input to the session and not just to the registration count.

The rule that makes it work: if you cannot say in one line why this specific person should come, do not send it. That single constraint is the difference between a direct invitation and a template blast with a first name at the top, and people can tell the difference immediately.

Send them wherever the existing relationship already lives — email, DM, the thread you were already in. Five a day for a week is manageable; twenty in one afternoon reads like what it is.

[Name] — you mentioned [the specific thing they said, or the
thing you know they are working on].

I am running a live session on [Day] at [Time] [Timezone] about
[the topic], and [the specific part] is most of it. Thought of you.

[Registration link]

No pressure either way — if the timing is wrong I will send you
the replay.

Keep the replies. Every “sounds useful, will you be covering X” is a question to answer in the room and a line of copy for the follow-up. This channel hands you the session’s content as a side effect of filling it.

Not until the session has sold something.

Buying registrations for a session that has never produced revenue buys you a more expensive way to find out that it does not. You learn what a registration costs, which is the number that does not matter, and nothing about what a customer costs, which is the one that does.

Run the first two sessions on the four channels above. If the session converts — people stay through the offer, click it, and some of them buy — you now have a revenue per registrant figure, and that is what tells you what a registration is worth paying for. Without it, a cost-per-registration target is a number with nothing on the other side of it.

When you do start, in this order

1. 01Retarget the registration page. People who visited and did not register are the warmest and cheapest audience you have. Set this up in the first window even if you spend nothing until the third.
2. 02Retarget your own audience. Video viewers, list members, past attendees. Cheap, and the message can assume they know who you are.
3. 03Audiences built from people who registered. Only once you have enough registrants for the platform to build something meaningful. Before then it is a guess with a budget attached.
4. 04Cold interest targeting. Last, smallest, and only with a session you have already watched convert. This is the expensive way to be wrong.

Judge the spend against attributed revenue, not against registrations. A campaign that fills the room with people who never attend is worse than one that fills half of it with people who do — and it will look better on every dashboard that stops at signups.

Fourteen days, laid out.

The channels above become a plan when they have dates attached. This is a working default to adjust against your own source data, not a schedule to obey — but running it once as written gives you the data to adjust it with.

Two things in that table do most of the work and are the two most often skipped: the partner ask before the window opens, and the source review at T−6d. The first is the only task with a dependency on another human. The second is the only moment where you still have time to act on what the data is telling you.

Tag every link before the first one goes out.

Source tracking is worth almost nothing for the first session and everything by the third — which is precisely why it has to be set up on day one, when it costs ten minutes, rather than reconstructed afterwards, when it cannot be.

One link per channel, one per partner

Use whatever tagging your tools already understand — campaign parameters on the URL are the universal option — and pick a naming convention you will still be able to read in three months. Something like list-announce, list-lastcall, partner-name, owned-pinned, direct. Boring and consistent beats clever, because the value comes entirely from being able to compare this session to the next one.

Then judge them on the right number

Registrations by source is the easiest report to produce and the easiest to be misled by. A partner who sends forty registrations who never attend is worth less than one who sends eight who buy. Judge channels on attendees at minimum, and on attributed revenue wherever you can trace it that far.

Three sessions of this and you will know which channel to stop working — which is the whole point, and the reason the first session’s tracking matters even though its numbers are too small to mean anything on their own.

Registrations have stopped. Diagnose before you promote harder.

Halfway through the window the numbers flatten and the instinct is volume — another email, another post, maybe some ad spend. Volume is the right answer to exactly one of the three possible causes. Work the questions in this order.

1. 01Are people arriving and not registering, or not arriving at all? This is the only question worth asking first, because the two answers have opposite fixes. If traffic is reaching the page and not converting, more promotion makes it worse — it spends the channels you have left on a page that is not working, and the fix is the promise, the title or the date. If the page converts what reaches it and the numbers are simply thin, the page is fine and you have a distribution problem.
2. 02Which channel was going to carry this, and did it actually happen? Nine times out of ten the answer is that the partners have not sent yet. Chase the specific thing that has not happened rather than broadcasting harder at the people who already heard you. One message to three partners beats a fourth email to your whole list.
3. 03Is the topic the problem? The uncomfortable one, and worth asking at T−7d while there is still time to act on the answer. If your own list, your own community and people who like you enough to take a personal message are all declining an hour on this, that is information. The fix is the angle or the title, not more sends — and it is much cheaper to learn now than after you have built the session.

Only after those three does it make sense to act. The list below is ordered the same way — by how likely it is to change the outcome, not by how much it feels like doing something.