From a Single Interview to Posts and Videos: Content Factory Operations

How to transform one expert interview into a series of posts and short videos: fact cards, material assembly, expert version approval, editing plan, and publishing.

By Artem Korsakov ·

A detailed expert interview can be used for posts, carousels, short videos, and answers to frequently asked questions. If you have created only one post from it, check what prevents you from using the rest of the material. Each subsequent format may require finding new quotes, verifying numbers, and coordinating them with the expert. Let us examine how to organize a common source review and reuse verified facts.

Source and Fact Cards

The work begins with the source: an interview recording, transcript, or expert notes. The source has a date, author, and client. A month later, these details will clarify where the number came from.

From the source, the editor extracts fact cards. A card contains one verifiable statement: a term, quantity, condition, or procedure. For example, "Planned commissioning date of the building: Q4 2027" for a developer, or "The first module of the course consists of 12 lessons" for a school. Expert opinions and formulations can also be saved, but separately from facts.

A good fact card passes a simple check: it can be shown to the expert in one line, and you get an answer of "true" or "false." If a context is needed for the answer, the statement should be split into several cards. From one hour-long interview, you usually get several dozen such cards, and this is a normal volume for a week's content.

Immediately mark facts that require special caution. Terms, prices, conditions, and result promises change more often than other information and are more likely to cause complaints. For them, it is useful to indicate the verification date and who conducted it.

Each card has a version. If the expert clarifies a number, a new version is created, and the old one remains in history.

Material Assembly

A post and a video script are assembled from cards. Each material stores links to specific fact versions included in it. The text can be prepared manually, according to a template, or with the help of an AI model, but the link to facts is recorded in any case.

For short videos, it is convenient to prepare a script by scenes right away: what is in the frame, what text is in the subtitles, and which fact it conveys. If the source video is long, the editing can be entrusted to a ready-made tool. For example, OpusClip offers editing and publication calendars for received videos. The script remains the basis for verification: it shows which facts will be in the video.

Expert Verification

The expert approves a specific version, for example, version 3 of the post about the building. The approval is linked to the version and to the fact versions within it. If the editor later changes the introduction, version 4 appears, and it must be approved again.

If there are multiple approvers, the order is set in advance: first, the expert on facts, then the client on tone and brand. Ready-made services support such chains, for example, multi-level approvals in Planable. For a content factory, there is one more condition: editing a fact removes approval from all materials where this fact was used.

When a Fact Changes

Suppose the developer has shifted the planned commissioning date of the building. The editor updates the fact card. The following should happen next:

  • all materials with the old version of the fact are marked as outdated;
  • approvals for these materials are revoked;
  • scheduled publishing of the old version is cancelled;
  • already published materials are marked as requiring correction.

The last point is often overlooked. A material that has already been published cannot simply be reassembled: the old version remains in the channel, and it must be corrected separately.

Подпись: Illustrative example. The status of each material depends on where it was located at the moment of fact editing.

Editing Plan and Publishing

For a video, the result of the work is an editing plan: scenes, duration, subtitle text, subtitle file. It can be passed to an editor or to a template rendering tool. The link between subtitles and facts is preserved, so when editing a fact, it is clear which videos will need to be remounted.

An editing plan is useful even without automatic rendering. The editor receives the exact text of each scene and does not rewrite subtitles from a draft. The expert can check the script before mounting, when correction still costs nothing. If the video is assembled according to a template, the plan becomes input data for the template, and the format of videos for one client remains consistent.

Publishing is scheduled by the client's channels. Each task has a status: in queue, published, failure, cancelled. If the platform does not respond, retrying should not create a duplicate.

Fact Card Template

To start without a system, set up a table with columns:

FieldFilling Example
ClientClient Name
SourceInterview from such-and-such date
StatementOne verifiable sentence
Version1, 2, 3
Who CheckedClient Expert
Where UsedPost such-and-such, script such-and-such

The "Where Used" column quickly shows how many materials one edit affects. When there are too many such lines, manual maintenance stops coping.

The table should be maintained for at least a month before choosing tools. By this time, it will become clear how often facts change, how many materials come from one source, and where delays occur. These observations will be useful for configuring ready-made services and for ordering a custom system.

In the demo scenario, you can trace the chain: source, post and script, fact correction, outdated versions, reassembly, expert approval, and local publishing queue.

View scenario

Related article: Ready-made services, automation, or your own platform for content production.

Sources and Example Boundaries

Sources confirm the described functions of ready-made products, not demand or return on investment for custom development. Scenarios and amounts are training examples. Demos work with fictional data; real integrations, AI, payments, and external publishing are not connected in them.