← Notes·2026-07-14·behind-the-video / startup-vlog
How to film a startup vlog: real numbers and real weather in Saigon.
How to film a startup vlog when the work is invisible: say the real number on camera, borrow structure from what the lens can see, and end on the next problem.
By Baptiste Vanpeltupdated 7 min read
Story pass and shot list: Planned in Vanpelt Studio →
A startup vlog works when it says one real number on camera and borrows its structure from something the camera can see, because the actual work (typing, thinking, waiting for users) is invisible. This note is for founders and creators who build in public and want a weekly episode that people finish. The short answer: pick the number before you film, let the weather or the city carry the shape of the episode, put the number at the turn rather than the top, and end on the next problem.
The worked example is Week 1 in Saigon: 100 Users, $0 MRR, and a Monsoon, a 6 minute 10 second vlog published in June 2026. A few weeks earlier my girlfriend and I had quit our jobs and moved from Belgium to Vietnam to start our own business. In that first week I reached 100 users and zero monthly recurring revenue (MRR, the subscription income a product brings in each month), and it rained every single day. Every step below uses that episode.
What are the six steps of a weekly startup vlog?
- 01Before filming
Pick the number and its unflattering partner
The metric you will say out loud, plus the one you would rather not. Output: one line, like 100 users, $0 MRR.
- 0210 min
Choose a visible stand-in for the work
Weather, a city, a move. Output: a cold open line the stand-in can pay off later.
- 0315 min
List what changed this week as segments
Things you can film anywhere: noise, food, a workplace, the sky. Output: four to six segments.
- 04Every day
Shoot arrivals first, say the date daily
Establishing shot on arrival, one spoken date and number per day. Output: footage you can search.
- 05Edit day
Place the number at the turn
Around two thirds in, after the viewer is invested. Output: an edit order.
- 06Last 30 s
End on the next problem
The thing that has not happened yet. Output: a cliffhanger into the next episode.
Step 1: pick the number before you pick up the camera
A startup vlog with vague numbers is a travel vlog with a laptop in it. Decide which number gets said on camera before the week starts, and pair it with the one you would rather leave out.
In this episode the pair was 100 users and $0 MRR, and both went into the title. The app was mostly free, so a hundred people found it and none of them paid. I said it plainly, then said the less comfortable part too: I was putting that project aside for now. A hundred users sounds like progress and zero revenue sounds like failure; most founders would report the first and skip the second, and anyone who has built something would notice the gap.
Omission is expensive because it compounds. Once an audience decides a channel reports selectively, every later number arrives discounted. Saying the bad one first is what makes the good one worth anything next month. Your output here is a single line you would be happy to put in the title.
Step 2: let something visible carry the shape
When your subject is invisible, borrow structure from something the camera can see. A desk on Monday looks exactly like a desk on Friday. Weather does not.
The episode opens on a joke: "The weather is actually so nice here in Saigon. I'm so glad we left Belgium." It is a setup. Near the end I admit we had rain every single day, that arranging a day around the rain is not easy, and that there is not much you can film when it pours, especially for the camera lenses. The cold open is paid off five minutes later, and the episode gets a shape that no metric could give it.
This was not luck with the forecast. Ho Chi Minh City's rainy season usually runs from May to November, with about 150 rainy days a year, so a June arrival was always going to be wet. Check the season where you are and write the cold open around what you know is coming. Your output is one opening line that the end of the episode can contradict or confirm.
Step 3: list what changed this week as segments
Week-one episodes are dangerous because everything feels worth filming: the city is new, the product is moving, and none of that is a story yet. A short list of "what changed" segments gives you a frame that survives a slow week.
This episode uses four. The noise (scooters, a construction site under the flat, an Airbnb door that does not shut). The food (a banh mi for around 2 euros at a place called Banh Mi 123). The coffee (Vamos, a workplace open 24/7, and my first coconut coffee). And the weather, listed as the third best thing about Vietnam just before it turns on us. A fake montage near the top, where I claim my routine has not changed at all, sets up the list by being obviously untrue.
- Ironic cold open about the weather 30 s
- Why we moved, and a joke montage 60 s
- What changed: the noise 45 s
- What changed: food and coffee 60 s
- What changed: the weather 35 s
- The number: 100 users, $0 MRR 40 s
- The startup beat, about a ninth of the runtime.
- Celebrating with some tourism 30 s
- Rain every day, packing for Da Nang 40 s
- The bus cliffhanger 30 s
The turn: the startup is the smallest segment
Look at the figure again. The business beat, the one in the title, takes about 40 seconds of a 370-second video. That is the part I did not expect when I started making these, and the part many build-in-public vlogs get backwards.
The number works because it is rare. If every minute is about the product, the 100 users is one more statistic in a stream of them. Placed at roughly 3:50, after the viewer has spent three minutes with the noise, the banh mi and the coffee, it lands as a turn: the week was hard and loud, and something real happened anyway. Then it costs something, because the revenue is zero and the project goes on the shelf. A segment list is the frame; the number is the event inside it.
Your output is an edit order with the number placed about two thirds of the way in. Not at the top, where it is just a claim, and not at the end, where it has nothing left to change.
Step 4: shoot a real week, not a staged one
Documentary coverage means the desk where the work happens, the streets between, the sky changing its mind. The discipline is in what you do not keep, because most of a working week looks like the rest of it.
Two habits make that survivable. Shoot the establishing shot of a place the moment you arrive, before you get absorbed in why you came, because you will not go back for it. And say the date and the number out loud on camera once a day, even if it feels silly. A speech-to-text model such as Whisper turns those spoken markers into a searchable transcript, which is what makes a week of near-identical desk shots findable rather than scrubbable. The footage organizing guide covers the rest of the system.
Your output is a day of footage with one spoken timestamp and at least one shot that proves where you were.
Step 5: end on the next problem
The episode does not end on the milestone. It ends with us packing for Da Nang, then the line that we have 20 hours of bus ahead and 2 hours to wait for it, in a station that looks like an old warehouse. "Adventure begins now." That is a cliffhanger, and it paid off: the bus became the next episode, written up in turning a travel mistake into a vlog.
A weekly vlog lives on the click into the next one. Close on the thing that has not happened yet, not on a summary of what did. Your output is a last 30 seconds that asks a question the next episode answers.
Weekly startup vlog checklist
- The number and its unflattering partner, written down before filming
- A cold open line the end of the episode can pay off
- Four to six what-changed segments you can film anywhere
- Establishing shot on arrival at every new place
- Date and number said on camera once a day
- Footage reviewed the same evening, while a pickup still costs twenty minutes
- The number placed about two thirds through the edit
- A last 30 seconds that sets up the next episode
- Title says the real number, not the flattering one alone
If it went wrong, it is one of these
- Viewers leave in the first 30 seconds. The opening does not match the title. YouTube's key moments report shows the share of viewers still watching after 30 seconds; if it is low, make the cold open set up the number or the weather the title promises.
- The week blurs into one long day. No spoken markers and no arrivals. Add the daily date line and the establishing shots.
- The number feels like bragging. It was reported without its partner. Say the uncomfortable half in the same breath.
- Nothing happened this week. There is no story without a turn. Make the segment list carry the episode and keep the number short.
- Rain stopped the shoot. It will, in a monsoon season. Plan indoor segments (a café, the flat) and protect the lens, then use the rain on screen as the payoff it is.
The footage review checklist is what I run each evening to catch the second and fifth problems while they are still cheap.
What carries over to client work
Founder stories and launch films ask a brand for the same exercise, and most brands fail it in the same place. The film is fine until the part where a real constraint should be named, then it goes vague, and the viewer feels the swerve. I cover how to brief around that in how to brief a founder story video. If the true thing cannot be said on camera, that is worth knowing before anyone books a shoot.
The whole recipe on one line
Pick the real number, let the weather carry the shape, list what changed, put the number at the turn and end on the next problem. The Saigon week is the proof: a hundred users, zero revenue and a monsoon made a better episode than a good week would have. More of this kind of work is on the video page.
Sources and further reading
- Ho Chi Minh City: climate
WikipediaThe rainy season usually runs from May to November, with about 150 rainy days a year, which is why a June week in Saigon is a wet one.
- Measure key moments for audience retention
YouTube HelpDefines the intro metric (share of viewers still watching after 30 seconds) and what dips and spikes mean, for judging where a compressed week loses people.
- Whisper
OpenAI on GitHubA general-purpose speech recognition model; one way to turn spoken date markers into a searchable transcript.
- How to organize YouTube footage
Vanpelt StudioTurning a week of near-identical clips into something you can search.
- Footage review checklist (after filming)
Vanpelt StudioWhat to check the same evening, while a pickup shot still costs twenty minutes.
Want a video like this made for your startup?