How to Write a Training Video Script (+ Free Template)

At Synthesia, we’ve made it really easy to make a training video. You can upload a slide deck or write a prompt and have our AI Assistant generate a first draft in minutes. 

Which leads people to ask, do I really need to write a script for my training video? 

Yes, unequivocally. 

A script is a plan for what learners will see, hear, and do in a video. Even if you’re using an AI video platform, it’s only going to make suggestions or act on your behalf. 

AI shouldn’t replace learning design. You still need to conduct a needs analysis and transform that into a learning objective and observable performance outcome for the training video. 

A script, then, becomes the way that you think about the learning design: how you’ll guide the AI to create your scenes and build a video that you can measure.  

Define the measurable outcome 

Before writing your script, I recommend completing the learning objective and performance outcome for your training video. You can use a template like this:

This video is for [specific role] who currently [context or gap]. After watching, [specific role] should be able to [observable action] so that [business outcome].

If you can't complete these sentences, you're going to have a hard time designing an effective training video.

The biggest mistake I see people making when they create videos is forgetting to plan for observable performance outcomes. By that, I mean evidence that the video is changing behavior in a way the business can measure. A VP of Sales may not care that 1,000 reps completed a training video, but they will likely care that it led to a 10% increase in average deal size.

Decide what kind of script to write 

Once you have your target audience and outcome, you can decide what kind of script you need to write. Training scripts exist on a spectrum. On one end, you have highly prescriptive scripts. These are commonly used for compliance and regulatory cases, like if you’re creating a training video about an SOP. 

On the other end, you have lightly structured scripts. These are commonly used for scenario-based learning, especially where there’s ambiguity. For instance, if you’re designing a training video around conflict resolution for managers, you may intentionally leave room for interpretation in your script. 

In the middle of the spectrum, you have scripts that provide clear steps, but still require judgment. These types of scripts work well for new hire onboarding, where multiple roles may need to learn a workflow but decide how to apply it in their role. 

Write your script

Now you have the inputs you need to write a script. You have your goal (and how to measure it), and have considered the kind of script you'll need for your video. I recommend writing your script one of three ways.

Option 1: Use an LLM 

A Large Language Model (LLM), like Claude, ChatGPT, or Gemini, is my go-to for script writing. At my organization, like many others, our LLM is integrated with our tech stack. That means it's able to search across our knowledge base for information that I may not even know existed. If I'm writing a script on an internal process, this ability to search internally within the same tool as I'm writing is invaluable.

When writing scripts, I'll start by doing research in an LLM chat, and then, when I'm satisfied I have the most up-to-date and comprehensive information to work from, I put in a structured prompt.

A structured prompt framework I particularly like is CIDI, which stands for Context, Instructions, Details, and Input. The way it works is you write a prompt that does the following:

• Gives the LLM a role and enough context to understand the scope of the task
• Provides step-by-step instructions of how you want the task to be performed
• Details any constraints that will guide the style or quality of the script
• Includes any relevant materials the LLM should consider as input

I've given an example of a CIDI prompt for script writing below.

An LLM gives you a draft quickly, which helps on days when you're staring at a blank page. Sometimes it's easier to look at a draft and say "that's not it" than it is to start from nothing. Even on good writing days, it's helpful to have another perspective to respond to or challenge.

Option 2: Use a template  

If you'd rather work from a structure than a blank prompt, a template may suit you better. In my experience, video training scripts are most compelling when they're well-organized.

Depending on the use case, that may include the following components:

• Hook
  Why someone is spending their precious time on this training 
• Outcome
  What will someone be able to do immediately after?
• Step-by-step instructions
  The minimum number of steps to complete the task or make the decision 
• Pitfalls
  Common mistakes and how to avoid them
• Next action
  What someone does next, and how they get help if needed

We have several templates to help you get started here. You can make these templates as prescriptive as necessary for your use case. (You can also upload them into an LLM to complete.) ‍

Option 3: Convert existing materials

I recognize that you may not be creating a brand new training, but rather adapting something that already exists into a video. In that case, you'll still follow the same process to get started (define your outcome and decide how prescriptive your script needs to be). Then, you can use an LLM to turn it into a script with a structured prompt.

Alternatively, you can use Synthesia's AI video generator to turn your existing training materials into a video, generating the script and building the video at the same time.

Evaluate your script

Perhaps the hardest part of script-writing is figuring out whether it's working. To truly know whether a script is landing, you have to start building the training video. Only then can you see if the scenes, with the narration, media, and interactivity, are working together towards your learning objective.

The best way to assess this before you publish a video for an entire audience is to pilot a rough draft. Ask a trusted SME who knows the audience to go through the video and provide feedback. If possible, watch them go through it, observing how they interact with scenes. Then, briefly ask them about the experience: where did they get stuck? Did anything not make sense or could have been explained better? Take this feedback and refine the script before you ship.

Only then will you be able to evaluate whether the training video, and the script by extension, is supporting your goals.

If this is your first training video, look for two types of evidence:

1. Snapshot measurement
   Are employees going through the video as intended? Monitor completion, drop-off, and rewatch rates, as well as any interaction results.
2. Impact measurement 
   Are employees doing something differently as a result of watching the training video? For this, you'll be looking for the observable evidence you identified earlier (you know, the behavior the business cares about).

These complementary data points will help you figure out what, if anything, needs reworking in your script. If you're losing employees early on, consider revising the introduction. If employees are repeatedly rewatching a specific scene, it may be too complex and need to be simplified into several scenes. If they're completing the video but nothing is changing behaviorally, revisit the original outcome: was the performance gap real, and is video the right medium to close it?