How to Write a Prompt for a Personalized Video

Start with the right kind of prompt

For a personalized business video, write a brief that separates what stays the same from what changes for each recipient. That distinction helps you plan the script and the data you’ll need; it does not mean a prompt alone generates the finished video.

BHuman’s documented campaign workflow uses a recorded video template and recipient details supplied through a data source. Its documentation also presents Speakeasy for creating presenter-led videos from a prompt, but does not establish that campaign personalization works by entering a prompt.

Use this prompt worksheet

Fill in the blanks before recording or preparing a template. Keep recipient-specific details in clearly named fields, and write the surrounding script as fixed text.

• Goal: What should the recipient understand or do?
• Audience: Who will receive the video?
• Fixed script: What words will every recipient hear?
• Personalized fields: Which details change, such as first name, company, or product?
• Call to action: What should the recipient do next, and which field supplies the destination link?
• Tone and length: How should the presenter sound, and how concise can the message be?
• Checks: How will you review names, pronunciation, timing, subtitles, and links?

Write the fixed script around the variables

Mark the exact words or media that should change. Use a stable, readable name for each field, such as first_name, company, or booking_url. BHuman recommends lowercase snake_case names and keeping them stable once a campaign or automation is live.

Only send the replacement value for a variable. If the template already says “Hi,” the first_name value should be just the recipient’s name—not “Hi Taylor”—or the greeting may repeat.

Keep personalized spoken sections short. BHuman notes that short variables produce more predictable timing. Its Keep Original feature supports up to 100 personalized sections per video, with each section limited to 10 seconds or less.

A clearly hypothetical example

For example, imagine a sales team preparing a brief introduction for each prospect. The following is a hypothetical script plan, not a claim about a particular campaign or its results.

Fixed script: “Hi {first_name}, I noticed {company} is focused on {priority}. Here’s one idea to consider. You can choose a time here: {booking_url}.”

Possible fields: first_name, company, priority, and booking_url. In the video template, mark only the changing phrases or link slot. In the data source, provide the corresponding value for each field; do not repeat fixed words such as “Hi” in the first_name value.

Turn the brief into a BHuman campaign

BHuman’s documented campaign flow starts with a ready AI Studio template that has a source video and at least one variable. Then choose a viewing mode, load campaign data, map each template variable to a source field, configure any needed options, and generate the videos.

Campaign data can come from a CSV or Excel upload, Zapier, Leadr, Pabbly, or API, subject to the documented workflow for each source. Map each variable to the field that supplies its value. Exact header matches may be selected automatically when possible.

Before generation, check that every required variable has a mapped field and every row has a value. For Mini mode, personalized moments need background coverage. Max mode can be restricted by plan limits or face-distance checks. Campaign generation is asynchronous, so review the generated results page for render status.

Test before you send

Generate a few test rows and review the details that can make a personalized video feel broken: name pronunciation, replacement timing, subtitles, and background assets. If a spoken replacement sounds awkward, shorten the variable or revise the fixed script around it.

Check that your field names and values match. A missing mapped column or blank mapping can stop campaign generation before the render request is sent. Keep text and media in separate variables when they need different values.