A video call to action works when there is exactly one, and it matches the job the video was built to do. A reach video asks for the follow or the share. A conversation video asks a question. A reference video asks for the save, and a conversion video sends people to one link. Decide the job before you film, write the final line for that job, and cut everything after it.
Endings get a fraction of the attention hooks get, which is strange, because the ending is where the viewer decides what happens next. Most creators spend twenty minutes on the first line and zero on the last one, then close with "so yeah, hope that helps" while the viewer's thumb is already moving.
How many CTAs should a video have?
One. I know the temptation, because every ask feels free: follow me, and comment your questions, and grab the link in my bio, and share this with a friend who needs it. Each additional ask splits attention and shrinks the odds of any single action happening. A viewer at second 38 has about one decision left in them. Spend it on purpose.
The follow-everything ending also reads as insecure, and audiences smell it. One confident, specific ask signals that you know what this video was for. Which raises the actual question you should answer before filming: what was this video for?
Match the video call to action to the video's job
Every video has one primary job, and the ending should collect on it. Four jobs cover nearly everything:
- Follow. For videos aimed at strangers, your broadest topics and strongest opinions. Earn it with a reason: "I post one of these teardowns every week, follow for the next one" beats a bare "follow me."
- Comment. For videos built around a debate or a common question. Ask something specific and answerable in five words. "Which one are you, checklist or chaos?" outperforms "thoughts?"
- Save. For reference content: steps, checklists, templates, anything a viewer will need at a future moment. Name that moment. "Save this for your next pricing call."
- Click. For conversion videos pointing at your lead magnet, product, or newsletter. One destination, named plainly, no mystery about what's behind the link.
Notice the pattern in the good versions: every ask comes with a viewer-shaped reason. The action is for them, and your phrasing should make that obvious. This mapping starts before scripting, by the way. When Batchly writes a script, each one carries a single CTA matched to what that video is for, because a job-less video is the thing that produces the mumbled ending.
The job also tells you which metric to check afterward, and why chasing likes misleads you. If you gave a video the save job, judge it on saves. I broke down why those signals beat likes in shares and saves vs likes.
The last two seconds decide the save
Here is the mechanic worth internalizing. A viewer saves a video at the moment they realize they will need it again, and that realization usually lands at the end, when the payoff clicks. If your payoff line is followed by eight seconds of trailing outro, the impulse dies inside the mush. If the video ends crisply, ask included, within two seconds of the payoff, the impulse has somewhere to go.
People decide to save your video in the last two seconds, right when most creators are mumbling "so yeah."
So the edit rule is blunt: find your payoff line, place your one ask immediately after it, and cut the take there. No wave, no wind-down sentence, and it will feel abrupt to you the first ten times. Trust the cut.
Here is what I call the one-job ending, a checklist to run for every script:
- Write the video's job at the top of the script, one word: follow, comment, save, or click.
- Write one closing line that asks for exactly that action.
- Attach a viewer-shaped reason or a future moment to the ask.
- Place the ask within two seconds of the payoff line, then end the video.
- Reread the script and delete any second ask hiding anywhere, including in your caption's first line.
Stop ending with "so yeah"
The trailing ending has one cause: the ending was never scripted. You wrote a strong open, three beats, and then assumed the close would take care of itself on camera. Live, in front of the lens, "so yeah" is what unwritten endings sound like.
A composite example. A resume writer I'll call Alex had solid middles and dead endings. His videos on interview answers closed with "so yeah, hopefully that's useful, I have more stuff like this." His save numbers ran around 3 or 4 per video at a few thousand views each. We changed nothing but the endings. Each script got a job written at the top, and each close became one line, like "save this for the night before your next interview, you'll want the second question." Over the next dozen videos, similar views brought 15 to 20 saves per video, and one video's comment-job ending, "what's the interview question that always trips you up?", produced his best comment thread and his next four topics. The middles were never the problem. The last two seconds were.
If you script the ending, you also get a filming benefit: takes end cleanly instead of trailing until you break eye contact, which makes editing faster. The ending is one of the seven lines in the script template for exactly this reason. It's a line you write, never a moment you improvise.
Do this next
Rewatch your last five videos and write down each one's final sentence. Grade them: does each have one ask, matched to a job, delivered within two seconds of the payoff? Rewrite the failures as one-job endings and use them as the closing lines of your next five scripts. If you want the whole script handled, Batchly turns your analytics export into pillars and writes each video with a single matched CTA already in place, so "so yeah" never makes it to the lens.