Turn a live webinar into an evergreen one: the conversion job.

Use the run that converted, not the one with the most people in it.

The biggest room is the one you remember, and it is rarely the one that sold. Attendance is a fact about your promotion; what happens after minute forty is a fact about the session, and only the second one survives automation.

Put your runs side by side and compare four things: how many of the people who joined were still there when the offer started, how many clicked it, how many bought, and what the questions were like. Rank on the middle two. Use the questions as a tiebreaker, because a run where people asked buying questions is a run where the argument tracked. Use attendance last, if at all.

A room of thirty that produced four purchases is a better master than a room of four hundred that produced two. The first tells you the session works and needs traffic; the second tells you the promotion works and the session does not, and automating it converts a bad hour into a bad hour that runs unattended.

Three awkward cases

1. 01The best run is unusable. Bad audio, a doorbell, a name-drop that dates it, twenty minutes of shared-window video at the wrong aspect ratio. You can re-record a damaged segment. You cannot re-record the energy of the run that worked — so try the repair before you fall back to a weaker master.
2. 02The offer landed but the teaching wandered. Usable, and the cut list will be long. Prefer this over the reverse: a tight teach with a weak offer needs the part you are least able to fix in an edit.
3. 03A guest carried a section. Confirm you have the right to use their contribution in an ongoing commercial recording before you build anything around it. This is a five-minute conversation now and a rebuild later.

Three kinds of cut, in the order they get harder.

A live recording carries a room, a date and a set of mechanics that assume both. Each of them quietly tells the viewer that they are watching something that already happened to somebody else — which is fine if you say so, and corrosive if you pretend otherwise.

Dead air is the easy pass and the one that returns the most minutes. Waiting for latecomers, the promise to give it another two minutes, poll-result pauses, the tech hiccup, the gap before the first question arrives. None of it is content, and all of it is expensive: a viewer alone at a screen has nothing socially holding them there, so silence is where they leave.

Live-only mechanics are harder, because they are load-bearing sentences rather than gaps. Anything that asks the room to act, names an attendee, or references how many people are present reads as a lie when the room is one person. Some cut cleanly. Some are wrapped around the point you were making, and those become re-record candidates rather than snips.

Anything that dates it is hardest, because it hides in plain sound. A season, a specific date, a recent announcement, a current price, a product version, a tool that may not exist next year. This is a credibility problem rather than a maintenance one: the moment a viewer works out they are watching something old, everything else you said becomes older too.

The Q&A is the hard one

In most live runs the last ten minutes contain the best material in the session and the least usable footage in the file. The answers are excellent; the questions are addressed to named people, half of them are specific to one business, and the whole block is stitched together with waiting.

Cut all of it, then re-record the three questions that came up in every run as a short, deliberate block. It will earn more than the raw footage ever did, because it answers what everybody wondered rather than what two people typed. The question queue you kept from the live runs is the source material.

Re-record the frame. Keep the teaching.

The instinct after a first edit is to re-record the whole thing properly, in a quiet room, without the mistakes. Resist it. The version with the room in it is the version that converted, and the empty-room version reliably sounds like one.

The mechanism is simple enough: you already know where the argument goes, so you deliver it faster, drop the pauses that earned their keep, and stop selling the parts you are bored of. Everything that made the live run persuasive was a response to a room that is no longer there.

So the default is keep, and four things are exceptions: the open, any transition that spoke directly to the room, the offer section, and the FAQ block you are building to replace the raw Q&A.

The open has a new job

A live open managed arrivals. An evergreen open has to do three different things: say what this is, say how long it takes and what the viewer will be able to do afterward, and say when the offer arrives. That third line costs nothing and buys back every minute people would otherwise spend waiting to find out whether they are about to be sold to.

Welcome, and thanks for making the time.

This is a recording of a session I originally ran live for
[audience]. Nothing has been added to make it feel live — it is the
session, tightened up, so you get the useful hour without the
part where we wait for people to arrive.

Here is how the next [45] minutes go. First, [the first thing].
Then [the second thing]. Then [the third thing], which is the
part most people came for.

At around [00:42] I will tell you about [the offer], and you will
have [72 hours] from today to decide. If that part is not for
you, everything before it still stands on its own.

Say it in your own words rather than reading that; the point is the four moves, not the phrasing. And note what it does not do — it does not claim anyone else is watching, and it does not apologise for being a recording. Neither is necessary.

Matching a re-record to the original

Same room, same microphone at the same distance, same lighting, same clothes if you are on camera. If you cannot match it, do not try: cut to slides or a full-screen deck for the re-recorded segments and let the change read as deliberate. Nearly-right is worse than different — a jump in audio tone makes the viewer notice the edit, and noticing the edit is the first step to wondering what else was constructed.

The offer has to work with nobody reacting to it.

Live, the offer runs on feedback: questions arriving, chat moving, somebody asking the price out loud so you can answer it for everyone. None of that exists now. Every job the room used to do has to be done by the recording, in advance, in order.

• →The objections have to be answered inside the recording. In the room, three or four objections arrived out loud and you handled them as they came. They still arrive evergreen — you simply cannot hear them. Write them out from the live question queue and answer them in order, inside the offer section, before anyone has to ask.
• →The price gets said out loud. A slide is not enough. Some of these viewers are on a phone with the video small and their attention partial, and a number that was only ever on screen is a number nobody heard.
• →The link needs somewhere permanent to live. Live, you paste it into chat while you keep talking. Evergreen has no chat, so the offer has to appear as a persistent element beside or beneath the video from the moment you start describing it, and it has to appear at a timestamp you have written down.
• →Scarcity that depended on the room goes. “There are twelve seats and forty of you here” cannot survive automation — the second viewer of the day can do the arithmetic. Either the constraint is genuinely real per registrant, or the sentence comes out.
• →Bonuses tied to the live date get re-anchored. Anything that expired at midnight on the day either disappears or moves onto the registrant’s own window. Both are fine. Leaving it as it was is the one option that is not.

Then write down the timestamp the offer begins at, because three separate things key off it: the persistent offer element on the page, the milestone that marks a viewer as having reached the offer, and the follow-up branch that treats “left before the offer” differently from “heard it and passed”. Get the timestamp wrong and all three point at the wrong people, quietly, forever.

Those branches, and the emails that run off them, are written out in the evergreen email templates. Why the split matters more than any single message in it is the argument in the evergreen funnel playbook.

One shared close becomes one clock per person.

This is the translation people get wrong most often, and the one that costs the most when they do — because a deadline nobody believes devalues every deadline you set afterward, including the honest ones.

Four decisions make the window real. Pick the anchor — registration time or viewing time. Viewing time is fairer and matches what the person experienced; registration time is simpler and keeps a clock running for people who never watched. Either is defensible. Not saying which one, in the emails and on the page, is not.

Pick a length you can hold, and hold it. Make expiry real, which means the checkout stops honouring the price rather than the timer merely reaching zero above a working purchase link. And decide what an expired link shows — a real page saying the offer has closed and what is still available, because some of those people come back at full price later and will not if the last thing you showed them was an error.

One more thing the live version never had to handle: the same person registering twice. Decide now whether a second registration restarts the window, and be aware that if it does, you have built a deadline that anyone can reset by using a second email address. The deadline mechanics guide works through the anchors, the timer behaviour and the edge cases in full.

Audit what you already own before any of it goes evergreen.

A live campaign leaves a pile of assets, most of them reusable and a surprising number of them carrying a date. The recording gets all the attention; the emails and the pages are where the live version actually leaks through.

Work the list below before you switch anything on. Every item is either reused, rewritten or retired — and writing that decision down for each one is faster than discovering in week three that the post-purchase email still thanks people for joining you live.

The recording is ready. The funnel is a separate build.

You now have a master, a versioned cut, a re-recorded frame, an offer section that stands on its own, the timestamp it starts at, and a deadline rule. That is the conversion job finished.

What remains is the funnel around it: the schedule model, the email branches that key off watch time, the three QA passes and the maintenance interval that keeps the whole thing true. Those are the remaining phases of the build guide, and the first decision waiting there is which schedule model the recording is going to run on.

One last thing worth knowing before you flip the switch: the first month is when evergreen funnels break most quietly. Nothing reports a dead link, a stale reference or a checkout that stopped taking payments — the funnel simply produces less, and less looks exactly like a slow month. Put the maintenance checklist in a calendar on the day you launch, not on the day you first suspect something is wrong.