Playbooks · 5 min read
The Reel Edit Decides Whether Anyone Types Your Word
Nothing in the automation builder affects whether somebody types your word. By the time a comment exists, every decision that mattered has already been made — in the edit, in the caption, and in the two seconds where a stranger decided this was worth a thumb.
So this post is about those decisions. If you want what happens after the word arrives, the Reels to DM funnel covers it end to end. Everything here is upstream of that, and it is where almost all the losses are.
Picture the person you are writing for
Sound off. One thumb resting on the screen, already half-committed to the swipe. Watching in a queue, or in bed, or with a toddler in the other arm. They have never heard of you. They will give you somewhere between four seconds and the whole video, and they will make the decision to type or not type in the moment your ask appears — not later, not after re-watching.
Every rule below is just that picture, applied.
Say the ask three times, in three channels
On screen as text. Out loud in the voiceover. Again in the caption.
This feels redundant when you are editing and it is not. Muted viewers never hear the voiceover. People who read captions are a different group from people who read on-screen text. If the instruction lives in only one channel, you have quietly decided that a large part of your reach will never learn there was anything to type.
Of the three, the on-screen text is the one that has to be right. It is the only channel that reaches everybody.
Where the ask sits in the edit
Most creators put it at the very end, after the outro, which is the worst available position. The end of a Reel is where the fewest people are, and it is competing with the next video already trying to start.
Put the ask immediately after the payoff — the moment the video has just delivered the thing it promised, while the viewer is still holding the feeling that made it worth watching. That is when a small request gets said yes to. Then repeat it at the end for the people who stayed.
Two mechanical details that decide more than they should:
- Leave it on screen long enough to act on. A word that flashes for six frames is a word nobody typed. It has to survive being read, understood, and then acted on with the same thumb that is currently on the screen.
- Do not put it under the caption bar or the profile row. The interface covers the bottom of the frame and the right edge. A perfectly written ask sitting behind the share button is invisible for reasons that have nothing to do with your writing.
The word has to be typeable with one thumb
Short. One word. No numbers, no punctuation you have to reach the second keyboard for, nothing you would have to spell out loud.
Matching folds case and punctuation and Latin accents, so Price?, PRICE and price are one keyword and you do not maintain a list of shapes. What it does not do is fold Indic vowel signs — दाम and दम stay two different words — and it does not do partial matches, so price will not fire on pricing. If your audience writes in Devanagari or types the plural, those spellings go in the list yourself.
Then add the wrong spellings. Not a clever pattern — the four or five specific mistakes your audience makes, harvested from your own comments in ten minutes. Every one you add is a person who tried and would otherwise have got silence.
The ask has to be worth typing
This is the part nobody wants to hear. A public comment is not free to give. It sits under your Reel with their handle attached, visible to their friends, permanently. A stranger will pay that for something concrete and will not pay it for the phrase "DM me for details".
They are not typing a word. They are putting their name in public in exchange for something you promised. Make the promise specific enough to be worth it.
Specific means the thing, named. The price list. The size chart. The recipe. The three-page checklist. "More info" is not a thing, and the reason a vague keyword produces a thin response rate is that a vague keyword has nothing behind it worth the exchange.
Name what happens next
Say the mechanism out loud: comment PRICE and I will send the list to your DMs.
Plenty of people have typed a word under someone else's Reel and received nothing, or received four messages. That memory is doing quiet damage to your conversion, and one clause naming the outcome undoes most of it. It also sets an expectation the automation can actually meet, which matters when the reply lands.
Then the automation gets its turn
The public reply is a receipt — two or three words, human, readable on its own, because on a Reel that travels it is the most-read thing you write. The private reply does the actual work: one clause of orientation for someone who has no idea who you are, then the thing they asked for, then one closed question, because their answer is the only thing that restarts your twenty-four hours.
Run Test on myself before you post the Reel, not after. It sends the real reply through the real pipeline to your own account, so you read it the way a stranger will — on a phone, cold.
Diagnose in this order
When a Reel gets views and no comments, work down the list rather than opening the builder first. Was the ask visible with the sound off. Was the word typeable. Was the offer worth the exchange. Only then, was the automation firing.
Activity tells you which half you are in. No matching comments at all means the problem is the edit. Matches with blocked rows underneath means the problem is the gate, and that is a different investigation. If it turns out the word itself is the weak link — which happens more often than anyone expects on a mixed-language audience — keyword choice has its own post.
