There are two halves to this, and most teams only work on the first. The capture half is planning: a shoot day yields more finished videos when the setups are built around several outputs instead of one hero cut. The retrieval half is what happens eight weeks later, when someone needs the 40 seconds where the customer explained why they switched, and it is somewhere in four hours of takes nobody has opened since the edit shipped. A shoot that was planned well and then buried is still a shoot you got one video out of.
Plan the day around outputs
Write the list of things you want to publish before you write the list of things you want to shoot. If the list has a 4-minute customer story, a set of social cuts, and stills for the landing page, the day looks different: you ask each interview question twice, once inside the conversation and once as a clean standalone answer that makes sense with no setup around it. You run a second camera wide so the same take gives you a cutaway. You shoot the vertical inserts while the set is still standing, because rebuilding it later costs another day.
This is real work with a real cost. Every extra setup eats time on the most expensive day in the whole process, and a crew that is chasing eight deliverables often comes back with eight mediocre ones. Two or three planned outputs plus deliberate coverage beats a long wish list.
Getting back into a shoot that already happened
The cheaper source of extra videos is footage you already paid for. The difficulty is not creative, it is access. People who shoot a lot describe the same pattern: during the day they notice a moment they liked, and afterward they are trying to remember roughly when it happened so they can get back to it.
The honest routes, in order of what they cost:
Watch it back and log it. One person, a spreadsheet, and a timecode column. This is slow, roughly real time or worse, and it is still the most reliable method there is, because a human notices things nobody would think to search for.
Search the transcript. Cheap and fast when the thing you want was said out loud. Useless for the shot of hands on the product, the reaction, or anything visual.
Tag at ingest in a DAM. Works when the discipline holds. Most teams keep it up for two quarters and then a rush job breaks the habit, and after that the metadata is worse than nothing because you trust it. Faster logging conventions help, but they are conventions, not automation.
Upload the footage into a retrieval layer, index it once, and search it by describing what you want. You describe the form of the thing rather than knowing what was said: a short direct instruction to camera, a wide of the room, a moment where someone corrects themselves. What comes back is a set of time ranges you can open, usually with a line of reasoning per range, and you still make the call on each. Once a shoot day is uploaded to a project in something like Vivu and indexed, asking for the setup you half remember becomes a question you type instead of an afternoon of scrubbing, and the ranges that come back are short enough that you usually want to extend them a little at each end before cutting.
One failure mode is worth knowing before you build a habit on it. An empty result is ambiguous. If you ask for a tight single on the speaker with hands moving and nothing comes back, you cannot tell from the result whether the day lacked that shot or your description was too narrow, and the only way to find out is to loosen the query and ask again.
When you don't need any of this
If a shoot produces two deliverables, one editor cuts both inside a fortnight, and the project folder is still open on their desk, none of this applies. Memory and a well named folder are enough. The problem starts at the point where the person who was on set is not the person who needs the footage, or where enough months have passed that nobody remembers what is in the file.
Which half is your problem
Look at the last three shoots. If the deliverable count was low because the shot list was thin, fix the planning and nothing else. If the shot list was generous and the extra videos never got made, your bottleneck is access, and the next thing to fix is how anyone finds a moment in footage that already exists. The same test applies before you book the next shoot at all: check whether you already have the shot first. And if the answer is that you have plenty and the editor is the queue, the constraint has moved again, to how you hand over selects.
FAQ
How many videos can you realistically get from one shoot day?
It depends on how many distinct setups you captured, not on how many hours of footage you have. Four hours of one person in one chair against one background gives you one long video and some cut-downs that all look the same. Three locations, two camera angles, and a set of product inserts gives you material that can carry several separate pieces without them feeling like reruns of each other. Plan for variety of setup, not volume of runtime.
Should we shoot vertical and horizontal on the same day?
Usually yes, if vertical is genuinely part of the plan. Cropping a horizontal frame to 9:16 works when the subject is centered and there is headroom to spare, and it falls apart with two people in frame, wide shots, or anything with text along the bottom. Framing a few takes natively vertical while the set is up costs minutes; recreating them later costs another shoot.
What do we do with footage from a shoot that happened two years ago?
Treat it as an inventory question before a production question. Find out what is actually in the archive and whether it still represents the product and the brand, then decide what is worth pulling from. Old footage is often fine for b-roll and human moments and useless for anything showing the product UI, so the usable half is smaller than it looks but it is rarely zero.
Is it worth logging footage on the shoot day itself?
A short version is. Nobody has the energy for a full log at wrap, but a few lines noting which takes were good and roughly when the unplanned moments happened will save far more time than it costs. The detail that matters most is the one nobody writes down: what you liked about a take, in your own words.