Having product video made: agency, in-house or AI

So this is a choice piece, not an explainer. Three routes, what each one really buys you, and the question that settles it more often than any quality argument.

Route one: a production company

You brief, they produce. Concept, pre-production, a shoot day, an edit, one or two feedback rounds, final files in the formats you asked for.

What you are buying is judgement and a finished thing. A good crew makes a hundred small decisions you would otherwise have to make yourself, and they carry the risk of the shoot day.

What it costs beyond the invoice is calendar. Between the first briefing and usable files sits a chain of other people's availability: the studio, the videographer, the editor, your product samples, somebody from your team present on the day. For a seasonal launch, that chain is usually the real constraint rather than the budget.

Where it fails is repetition. The second version costs roughly what the first one did. If your range changes four times a year, you are buying the same thing four times.

Route two: in-house

A camera, a light, a corner of the warehouse, and somebody on the team who is good at this.

What you are buying is availability. No scheduling, no minimum order, and a new clip the same afternoon a product arrives.

What it costs beyond the equipment is that person. In-house video tends to live with one enthusiast, and it moves at the speed of how busy they are that week. When they leave, it usually stops.

Where it fails is consistency across time. Not within a session, which is normally fine, but between March and September, when the light in that corner has changed and nobody wrote down where the lamp stood. This is the same failure mode we described for stills in one product, many interiors.

Route three: AI production with a person in the loop

Clips generated from your own product photography, with a creator checking and correcting before anything is delivered. This is how we work, so take the following with that in mind, including the part where we say what it does not do.

What you are buying is the ability to change your mind cheaply. A second version, a different season, another market, a vertical cut: these stop being projects and become small requests.

What it costs beyond the subscription is preparation. You need usable product photography to start from, and you need your visual rules written down somewhere other than one person's head. Brands that have never written them down find that out here rather than earlier.

Where it fails is the load-bearing clip: anything where the buyer is judging a mechanism, a material or a fit. We set out that boundary in what generated video does well, including the parts it does badly.

And it is not unattended. A creator from our team reviews and corrects every clip before it leaves, which is the same model we described in a person in the loop.

The question that usually settles it

Not "what is the quality difference". This one:

How many times will this clip have to change in the next two years?

Once or never, and the clip is the centrepiece of a campaign: hire a crew. You are buying one excellent thing and you want someone to own it.

Several times a year, across seasons, markets or variants: that is a production problem, not a filming problem, and routes two and three are built for it while route one is not.

Constantly, because your range turns over: do not solve it with shoots at all. Solve it with a library and a repeatable process.

Change frequency, not quality, is what actually separates these routes. Most budget arguments stall because both sides are comparing the first clip, where an agency obviously wins, instead of the twentieth, where it is not close.

What nobody puts in the budget

Three things, consistently.

The second version. Almost every clip gets one. Different length for a platform, different language, something a dealer needs in another format. Ask what version two costs before you choose, because that number differs by an order of magnitude between the routes.

Moving the product. This is invisible on a spreadsheet and real in a warehouse. On our use case page there is a brand called BosGrill that makes kamado barbecues. Their challenge was not that photography was expensive, it was that the products are heavy, hard to transport, and once they have been used in a shoot they can no longer be sold as new. That was about stills, and the logistics are the same for a video shoot, only with more people standing around. For heavy, fragile or made-to-order products, the shoot day is rarely the biggest line.

Your own team's hours. Briefing, sample logistics, attending the shoot, feedback rounds. It is real time and it never appears in the comparison.

Agree these four things before you sign anything

Whichever route you pick, the same four points decide whether the material is still useful to you in two years. They are easy to settle beforehand and expensive to renegotiate afterwards.

What you are allowed to do with it. Not just "we own it". Can you run it as paid advertising, can you keep using it after a contract ends, does it cover every market you sell in. Models, voices and music each carry their own terms, and those are usually where the limits sit rather than in the footage.

Whether your dealers may use it. This one gets forgotten constantly and it is the whole point for a producer who sells through other people. If your resellers cannot legally republish the clip, you have paid for something that stops at your own website. Write it into the agreement in plain words.

What you receive besides the final file. Source files, the graded master without text, and the clip without music. Those three are what make version two cheap. Without them, every future variant is a new production.

Where it ends up afterwards. A clip that lives in one person's drive is a clip you will pay to make again. It belongs in one library everyone works from, the same place as everything else.

Picking without agonising

A short version you can use in the meeting.

• •One hero clip, high stakes, long life: production company.
• •Fast, frequent, low stakes, and someone on the team enjoys it: in-house.
• •A large range, several markets, regular change, and consistency that has to hold across all of it: AI production with a person in the loop.
• •Any combination. The most common pattern we see is one proper shoot a year for the hero material, with everything else produced from it.

Before you decide anything, be clear about what the clip is for once it is live. That sounds obvious and it is skipped constantly, so we wrote it out in what the clip has to do on the page.

Curious what this looks like for brands like yours? Have a look at our use cases and see how it works, or try LensLab free with five credits.