I have cut somewhere north of a thousand podcast episodes now, and the question I get most often from people who have not started yet is some version of "what gear should I buy." That is almost never the thing standing between them and episode 20. The thing standing between them and episode 20 is that they built a show they cannot stand to make on a bad week.
So here is the honest answer to what I would do if I were starting a podcast today, from the chair of the person who has to open the project file afterwards. It is not a gear list. Some of it will annoy you.
I would decide the format before I bought anything
Format is a promise about your calendar. A weekly interview show means you are running a booking operation forever. A solo show means you are running a writing operation forever. A co-hosted conversation show means you are running a scheduling operation between two humans who both have lives.
Pick the one that survives a bad month. I would take a fortnightly solo show over a weekly interview show every single time for a first podcast, because the failure mode of the interview show is silent. Nobody cancels. The guests just get harder to book, the gaps get longer, and one day you notice it has been six weeks.
Most shows that die do not die from bad audio. They die from a format that needed more of the host than the host had. I wrote about the shape of that collapse in why most podcasts quit at episode 7, and the pattern has not changed since.
I would shoot video from episode one, even if I published audio
This is the part where people push back, so let me be direct about it.
If I were starting today I would record video from the first episode, even if I had no plan to publish it for six months. Not because video is magic. Because the version of your show that grows in 2026 almost always has a visual surface, and retrofitting that later means your first fifty episodes are unusable for it.
I have watched hosts spend a year building a good audio-only show and then hit a ceiling they did not see coming. Audience stops compounding. Nothing is broken. There is just nowhere for a new listener to accidentally find them. I broke down the mechanics of that plateau in why audio only growth stalled, and I stand behind the conclusion: audio is a great listening format and a terrible discovery format.
Video does not mean a studio. It means a camera pointed at your face, on a tripod, in the same spot every week. If you are solo, one camera is genuinely fine, and the whole workflow for single-camera shows is simpler and cheaper to run. If you have guests in the room, two cameras is the jump that actually changes the edit, and everything after that is refinement.
The best camera setup is the one you can repeat without thinking about it on a Tuesday morning when you are already late.
I would spend on the room, not on the microphone
Every beginner buys the microphone first. The microphone is the least of your problems.
A moderately good mic in a treated corner beats an expensive mic in a kitchen with tile floors. Room reflections are the one thing that no editor, no plugin, and no AI cleanup tool can fully undo, because the reflection is baked into the same waveform as your voice. I can fix a hum. I can fix inconsistent levels. I can fix a fridge. I cannot un-echo a room, not really.
So if I had a budget of a few hundred dollars for audio, I would put most of it into soft surfaces, a stand that keeps the mic close to my mouth, and a quiet time of day. Then whatever is left goes to the mic.
The second thing I would do is record a fifteen second test and actually listen to it on headphones before the guest joins. Not the levels. The sound. Half the audio disasters I receive were audible in the first fifteen seconds and nobody checked.
I would not let a machine decide what to cut
You can now feed an episode to a tool and get a shorter episode back in minutes. I have used most of these tools seriously, not as a stunt, and my position has hardened rather than softened.
They are genuinely good at mechanical work. Transcription, filler word detection, leveling, noise reduction. That part is solved and I use it daily.
They are still bad at meaning. Silence detection does not know that the four second pause before someone answers a question about their father is the best moment in the episode. It sees dead air and removes it. It also does not know that the story starting at minute 34 should have opened the show. A machine cutting by waveform will confidently deliver you a tighter, faster, worse episode.
That is why every episode we cut is watched end to end by a person. Nothing gets auto-cut by silence detection. I put a year of real testing into ai editing tools one year in if you want the longer version, and if you are weighing an automated audio chain against a human mix, Auphonic versus human audio mixing lays out where the line actually falls.
I would build the clip habit before I needed it
Here is the mistake I see most in month three. The host records a great episode, publishes it, then goes looking for clips. That is backwards and it is why clip production feels like a chore.
If I were starting today I would do two things during the recording, not after:
- Say the guest name and the topic clearly in the first minute, so the episode is self-contained wherever it lands
- Note the timecode any time I feel the energy jump in the room, even roughly, in a notes app
That second one takes no effort and it saves an hour per episode forever. When you hand an editor three rough timestamps, you get back clips that carry the actual moment instead of clips assembled by someone guessing. Our clips work gets noticeably better when the host does this, and the difference is not small.
I would also say the boring truth out loud: one good clip a week beats seven mediocre ones. The volume advice you have read is written for people with teams.
I would treat packaging as part of the edit
Titles and thumbnails are not marketing that happens after the show. They are the show, from the perspective of someone who has never heard of you.
I would write the title before I recorded. Not to force the conversation, but because a title you cannot write is usually an episode you have not thought through. If the working title is "chat with Dave," the episode does not have a subject yet.
For anything going to YouTube, the thumbnail carries more weight than everything I do in the timeline. That is an uncomfortable thing for an editor to admit and it is still true. Our sister studio ClickTheory exists specifically because so many hosts we work with were shipping good edits with packaging that buried them.
I would decide early where the audience actually lives
Downloads are a vanity number and you already suspect that. The metric I would care about in year one is whether anyone comes back on purpose.
The cheapest way to find out is to give people somewhere to go that you own. An email list is unglamorous and it is the only channel that survives a platform changing its mind. Even a short note per episode, sent to two hundred people, tells you more about your show than a download chart ever will. I sketched out how that loop actually works alongside a podcast in a separate post on building a newsletter loop around a show.
The other thing I would do is pick one niche and be almost annoyingly specific about it. Broad shows compete with everyone. A show for a defined group of people gets shared inside that group, which is the only distribution that compounds without spend. The most consistent growth I see comes from shows aimed at a real profession, which is part of why work aimed at a single profession, fitness coaches for example, tends to have a clearer path than a general interest show with better production.
What I would actually do in week one
If you want this compressed into something you can act on today:
- Choose a format you can sustain on your worst week, then halve the frequency
- Record video from episode one even if you publish audio only
- Fix the room before you upgrade the mic
- Write the title before the recording
- Note the moments that felt alive while you record
- Batch two or three episodes before you launch so a bad week does not break the streak
- Decide who the show is for, specifically enough that you could name five of them
That is the whole thing. Notice how little of it is about equipment.
The reason I would give this advice rather than a shopping list is simple. A thousand episodes in, the shows that are still going are not the ones with the best gear. They are the ones whose hosts built something they could repeat. Everything I learned watching that play out is in what I learned cutting a thousand episodes, and the pattern is boringly consistent.
If you get to the point where the editing is the thing eating your week, that is the part we take off your hands, and you can see how it works without talking to anyone first. But start the show either way. The edit is a solvable problem. The not starting is not.