The mistakes that hurt a podcast the most are almost never the big creative ones, right, they are the small, boring ones that slip through on a Thursday night, a guest's name spelled wrong in the lower third, a sponsor read from last month's advertiser still sitting in the middle of the episode, a camera angle that drifts two frames out of sync after the forty minute mark, or an export that sounds fine in the editor's headphones and noticeably quieter than every other show in the listener's app.
None of those mistakes say anything about whether the edit was good, and that is exactly why they are dangerous, because a host who receives a beautifully paced episode tends to watch the first few minutes, feel happy and hit publish, and then a guest or a sponsor emails two days later. So I want to walk through what we actually check before an episode comes back to you, why each check exists, and the part you should still check yourself, because no editor knows your show's facts as well as you do.
What actually gets checked before an episode is delivered?
Before delivery the episode should be checked for sync, audio levels and loudness, names and titles on screen, sponsor reads, captions and the export settings for every platform it is going to, in roughly that order.
The order matters more than people think, basically, because the checks build on each other. There is no point proofreading lower thirds on a timeline where the audio is about to be remixed, and no point checking captions before the final cut is locked, since every trim shifts the timing. So the technical foundation comes first, then the words on screen, then the commercial pieces like sponsor reads, and only at the end the exports, which is where a perfect edit can still go wrong because of one wrong setting.
I keep a fuller version of this list in the podcast editing checklist I run on every episode, and if you look at the overall flow on how it works, all of it happens inside the editing step, after your files arrive and before the first cut reaches you, so the checks sit in time you are already waiting for rather than adding a stage of their own.
How do you check sync across cameras and audio?
We check sync by watching the cut at several points across the whole episode, not just the opening, because drift between cameras and a separate audio recorder usually shows up late in a long recording rather than at the start.
Tools are good at lining up a clap or a waveform at the top of a recording, and we use them for exactly that, but cameras and recorders do not always run at precisely the same speed, so an episode that is perfect at minute two can be a few frames off at minute fifty, and on a close-up of someone talking, a viewer feels that even if they cannot name it. So the editor scrubs to the middle and the end, watches lips on the tight angles, and on multi-cam shows checks every angle switch where the audio crosses from one speaker to the other. If something drifted, it gets fixed section by section, and this is one of those really really unglamorous jobs that nobody thanks you for and everybody notices when it is skipped.
What loudness should a podcast episode be delivered at?
The number I aim an audio master at is the one Apple Podcasts publishes, an integrated loudness of roughly minus 16 LKFS, give or take 1 dB, with true peaks kept at or below minus 1 dB FS, because that stops your episode sounding obviously louder or quieter than whatever the listener played just before it.
Apple publishes this in its audio requirements for podcasters, and it also recommends AAC for RSS feeds, and in practice those numbers are a sensible reference point for any podcast app. The check here is not only that the number on the meter is right, it is that the voices inside it are balanced, because an episode can hit the target overall while the host is comfortable and the guest is too quiet, which a meter alone will happily pass. So the editor listens on more than one playback setup, headphones and a small speaker at least, and checks the moments where people laugh or talk over each other, since that is where levels jump. Our page on podcast audio cleanup and the post on podcast mixing and mastering cover the mix itself in more detail.
Why do names and titles on screen need their own check?
Names and titles on screen need their own check because they are the only part of the edit where a single wrong letter is public, permanent and personal to the guest, and spellcheck cannot catch a misspelled surname.
A guest will forgive a slightly long tangent, right, but they will screenshot a lower third that gets their company or their title wrong, and the person who notices first is usually their boss. So every lower third, every on-screen graphic and every title card gets read against the source you gave us, the guest's bio, the booking email, the LinkedIn profile you linked, and not against what the editor heard in the audio, because people say names fast and spellings are a trap. And on top of that, we check that a lower third appears when the person is actually on screen, not three seconds late or over the wrong angle, which is an easy thing to break when a cut changes during revisions.
I would rather an episode come back one hour later than come back with your guest's name spelled wrong.
How are sponsor reads checked before delivery?
Sponsor reads are checked for the right sponsor, the right version of the read, the right position in the episode and a disclosure that is clear, because a sponsor is paying for exactly those things and will notice if any of them is off.
The most common mistake I see across the industry is a stale read, where a host records a few versions over a month and the old one ends up in the new episode, or a promo code that changed is still being read out. So the check is simple but strict, the read in the timeline is matched against the latest one you sent, the placement matches what you agreed with the sponsor, and any on-screen promo code or link is checked character by character. Disclosure stays in, always, because both YouTube's rules on paid promotion disclosure and general advertising guidance expect a paid relationship to be clear to the viewer, and whether a specific read meets the rules for your situation is your call or your advisor's, not ours. The podcast sponsor integration page explains how we cut reads so they sound like the host rather than an ad.
Do captions get checked by a person?
Captions are generated with tools for speed and then read against the audio by a person, because automatic captions are good at common words and bad at names, product terms and anything said quickly or with an accent.
That is one reason, accuracy. Secondly, timing, because a caption that appears after the word has been said reads as a lag and makes a clip feel cheap, so the vertical clips, which are 25 dollars each and come captioned and ready to post, get watched with the sound off to see whether they still make sense. And the third thing is accessibility, which is not optional for a lot of viewers, and the W3C's guidance on captions for audio and video is a good plain explanation of why accurate captions matter for people who rely on them. More on how we handle them is on the captions and subtitles page.
Why does a second pair of eyes catch what the editor misses?
A second pair of eyes catches what the editor misses because after hours in a timeline the editor sees what they meant to cut rather than what is actually there, and someone watching fresh sees the episode the way a viewer will.
This is not about the editor being careless, to be very honest it is the opposite, the more carefully someone has worked on an episode, the more their brain fills in the gaps, so a doubled word at a cut point or a lower third that is one frame early just disappears for them. A fresh viewer has none of that memory, they simply watch, and the small things jump out. This is where a bigger team has a structural edge over a solo freelancer, nothing to do with talent, simply that with close to fifty people under one roof it is easy to put a fresh viewer on an episode before it goes out, and with more than 400 episodes a month going through our hands, the same handful of boring mistakes turns up often enough that people learn exactly where to look for them.
What should you still check yourself before publishing?
You should still check the facts only you know, things like whether a guest asked for a line to be removed, whether a number they quoted was correct and whether anything said is sensitive for your business, because no editor can know those from the footage alone.
Here is the thing, right, quality control on our side covers what can be checked from the files and your notes, and your one pass of notes is the final check that covers everything else. If a guest emailed you after recording asking to cut a remark about a former employer, we only know if you tell us, and if your show touches health, finance or legal topics, it should go through whatever review your industry expects before it is published. The post on the revision round that actually works explains how to give notes in one pass so nothing gets lost. Fair enough so far? The point is that a good process splits the work, we catch the technical and visual mistakes, and you catch the ones that need your knowledge of the show.
Where this fits in the price
Every check above is built into the editing price rather than sold as an extra, so a single-camera hour on Starter at 199 dollars goes through the same sync, loudness, name and sponsor checks as a multi-camera episode, and the differences between the plans, revisions included, are spelled out in the pricing section.
The one thing that can move is the clock, right, because a very long recording, a graphics-heavy cut or four camera angles simply has more to check, so an episode like that can need an extra day beyond the usual two, and I would much rather you hear that the day the files arrive than receive a rushed episode on time. If you want to test all this on your own show, send me an episode that has already gone out, by email to samy@pixelsamy.com or on a quick 15 minute call, and I will go through which of these checks it would have tripped and quote you before we finish. In the end, Pixel Samy Podcast gets judged on the boring stuff as much as on the pacing, because a listener forgives a long pause and a guest never forgets a misspelled name. So yeah. That's my way of saying it.