The real cost of corporate video is the rewrite, not the shoot

Internal video programmes are approved on a production budget and killed by an amendment budget that nobody calculated.

The production number is visible and finite. A studio day, a presenter, an editor, a set of deliverables. Finance can evaluate it, and for a company of any size it is not a large line.

The second number never appears in the proposal. It is the cost of changing four sentences eleven months later, and it is not four sentences worth of the original cost. It is a studio, matched lighting, matched wardrobe, the original presenter's availability, and an editor who has to make a re recorded segment sit inside an existing cut without an audible seam.

That asymmetry, cheap to make and expensive to change, determines what happens to almost every corporate video library.

How often this material actually changes

The assumption behind most internal video is that the content is stable. Measured rather than assumed, it usually is not.

Onboarding material references systems, tools, org structure and role names, and any of those changing makes part of the video wrong. Most companies touch at least one of them per quarter.

Policy and compliance material changes when the underlying requirement changes or when an incident prompts a revision, and revisions are not scheduled events.

Procedure documentation changes whenever the procedure does, which in an operations context can be monthly.

Product and customer facing internal briefings change every release.

Set against a filmed asset that costs a studio day to amend, this cadence means the library begins degrading almost immediately after it is delivered. Within a year a meaningful share of it is describing things that are no longer true, and the failure is silent. A stale document carries a revision date. A confident recording does not, and staff act on it.

The amendment cost model

The number worth putting in the proposal is cost per change, not cost per minute.

For filmed video, cost per change is roughly the cost of a shoot regardless of the size of the change, because the fixed costs dominate. Amending forty seconds and amending four minutes cost about the same, which is why organisations batch changes and why batched changes are made late.

That single property explains the observed behaviour of internal video teams. They resist small corrections, they accumulate a list, they schedule an annual refresh, and by the time the refresh happens the material has been wrong for months.

The alternative worth evaluating is not a cheaper shoot. It is a production method where cost per change is close to zero and cost per minute is slightly higher.

Where a photo based approach fits

Generating a presenter from a still photograph rather than filming one changes the amendment cost specifically.

The presenter is not a recording, so there is nothing to match. Platforms such as Leadde.ai build what they call a photo avatar from a single image and drive it from a script, which means amending narration means editing text and regenerating. Lighting, wardrobe and audio conditions are consistent because they were never conditions.

Two constraints belong in the evaluation rather than being discovered later. The photograph has to meet the platform's requirements, which typically exclude group photographs, hats, sunglasses, heavy filters, low resolution images and screenshots, and it should be one that will still look current in a year, since it becomes infrastructure. And expressive animation modes generally carry a duration cap per video, sixty seconds in Leadde's case, which means longer material is structured as segments rather than as one continuous piece.

The trade off is real and should be stated. Generated presenters carry information competently and carry presence poorly. The method suits the material that changes, which is also the material where presence matters least.

The internal inventory, sorted by change frequency

A useful exercise before choosing anything is listing internal video needs by how often the content changes rather than by department.

High frequency material includes system walkthroughs, policy updates, quarterly business briefings and anything referencing current org structure. This is where the amendment cost dominates and where a regenerable method pays for itself in the first year.

Low frequency material includes company history, values, and the parts of onboarding covering what the business does rather than how it currently operates. This changes rarely, and it is also the material that benefits most from real people on camera, which is a convenient alignment.

Material that should not be video at all sits in a third category: anything requiring signed acknowledgement of specific wording, anything consulted for a single value, and anything with genuine interpretive latitude where the exact phrasing determines outcomes.

Most organisations find the first category is larger than they expected and has been served worst.

Consent and disclosure are governance items

A synthetic likeness of a named employee raises two questions that belong in policy rather than in a production brief.

Consent has to be specific. Agreement to a photograph being used on a website does not obviously extend to that likeness delivering scripted narration the person never recorded. The consent should state what the avatar will be used to say, over what period, and what happens when the person leaves, since a company whose onboarding presenter left eighteen months ago has a problem that was easy to prevent and awkward to fix.

Disclosure is owed to the audience. Staff should be able to tell that a presenter is synthetic, and the cost of a line stating it is negligible against the cost of employees working it out for themselves and reassessing everything else they were shown. The Partnership on AI framework for responsible synthetic media sets out the disclosure expectations that are becoming standard practice, and internal use is not exempt from them.

Captions belong on every version. Internal video is frequently watched in an open office without headphones, and captions for prerecorded video are a WCAG success criterion.

Where to start

The wrong first project is the company overview, because it is the most visible and the least representative.

The right first project is the piece of internal content that has been wrong for the longest time because nobody could justify a shoot to fix it. Every organisation has one, and everyone involved knows what it is.

Rebuilding that piece establishes the two numbers that matter: what it costs to produce, and what it costs to change when the next revision arrives. The second number is the one the decision should be made on, and it is the one no proposal currently contains.