7 min read
Podcast to TikTok: a workflow that holds up
The whole path from a two-hour recording to daily vertical posts, including the parts the platforms make awkward on purpose and the three-tap workaround that works.
Every guide to repurposing a podcast stops at “and then you post the clips”. Posting is exactly where these workflows fall apart, so we will start at the end and work backwards.
The end of the pipeline is a phone
TikTok’s API only lets applications publish posts as private until the application has passed an audit. In practice that means no tool can publish public TikTok posts on your behalf unless it has gone through that process, and most have not. Anything advertising fully automatic TikTok publishing is either audited, in which case they will say so plainly, or the posts arrive private and you have to open each one and make it public, which is more work than posting it yourself.
Instagram and YouTube are more accommodating, but building three different publishing paths with three different sets of caveats produces a workflow nobody can remember. The honest version is simpler and it works on all three: the clips arrive on your phone, and you share them out through the system share sheet. Three taps per post.
Two details make that path bearable rather than miserable:
Never share text and video together. On iOS, passing both a file and a string to the share sheet makes the receiving app frequently prefer the text, and you end up publishing a text post with no video. Share the file alone and put the caption in the clipboard separately, so pasting is one more tap and never a surprise.
Review on the same device you post from. If you approve clips on a laptop and post from a phone, you have invented a file transfer problem for yourself. Approve and post in the same place.
Working backwards: what has to arrive on the phone
For three taps to be enough, the clip has to be final when it reaches you. That means the vertical crop is right, the captions are burned in, and the caption text is written. If any of those is missing, the three taps become a fifteen-minute editing session and the workflow collapses.
Final also means reviewable in a couple of seconds. Each clip needs its hook visible as text, and its in and out timecode against the source, so that when a choice looks wrong you can see why immediately and reject it without opening the original episode.
The middle: reframe and captions
A podcast is shot 16:9, usually with two or three people, sometimes with a camera switcher. Cropping the centre gives you a vertical frame of the gap between two heads.
The rule that matters: reframe follows the speaker, and it only moves at real scene cuts. A crop that drifts inside a sentence looks like a mistake even when the framing is correct; a crop that snaps when the shot changes looks intentional because it is. If you are doing this by hand, put your keyframes at the cuts, and work through the rest of the manual pass in the same order every time.
Captions should be readable at arm’s length, positioned clear of the bottom third where every platform puts its own interface, and consistent across every clip. Consistency is the part people underrate: the same type, the same position and the same colours across three hundred clips is what makes a feed look like a channel rather than a folder.
The front: choosing, which is the whole job
A two-hour episode contains about nine publishable moments. Not thirty. The mechanical operation of cutting a two-hour file into thirty pieces is trivial and produces mostly filler, and filler posted daily teaches an audience to scroll past you.
Choosing well means reading rather than scrubbing. A transcript of a two-hour episode takes fifteen minutes to read and you can mark candidates as you go. Look for self-contained thoughts: a claim someone might argue with, a story with a turn, a number with its context. Skip anything that needs the previous five minutes to make sense.
Whatever does the choosing, whether that is you, an assistant or software, should hand back three things per candidate: the start, the end, and the sentence that earned it. If you cannot see why a moment was chosen, you cannot correct the process when it chooses badly, and you will end up re-reading the whole transcript anyway.
A weekly rhythm that holds
Here is the shape that survives a real schedule, for a weekly two-hour show:
- Record once. Nothing about this workflow requires extra filming, extra takes, or a second setup.
- Process the same day. Transcript, candidates, cuts, reframe, captions. This is the part to automate first, because it is the part that is identical every week.
- Review in one sitting. Nine clips, two seconds each to accept or reject, plus a minute for the two you want to trim. Under ten minutes.
- Post across the week. Nine clips across three platforms is twenty-seven posts, which is about three a day. Spread them; do not dump them.
- Read the numbers once a week, not once a day. Daily numbers are noise. Weekly numbers tell you which kind of moment your audience actually watches, and that is the only feedback that should change your selection.
What to measure
Watch time and completion rate per clip, and where the views come from. That last one is the useful one: a high proportion of views from the recommendation feed rather than from followers means the clips themselves are doing the distribution. A low proportion means you are being seen by people who already know you, and the hooks are not landing with strangers.
Do not measure follower count weekly. It moves too slowly to steer by and it tempts you into posting things that perform for the wrong reason.
The part nobody can promise you
None of this guarantees anything. Distribution is controlled by platforms that change their behaviour without notice, and anyone offering you a view count is quoting a number they do not control.
What a working pipeline does guarantee is that the constraint stops being you. When a two-hour recording reliably becomes nine finished clips and twenty-seven scheduled posts, the question changes from “why am I not posting” to “which of these ideas deserves an episode”. That is a much better question to be stuck on.