How to turn policies and procedures into effective e-learning

There is a tempting way to create e-learning from a policy or procedure:

1. Split the document into sections.
2. Put each section on a slide.
3. Add narration.
4. Finish with a multiple-choice quiz.

It is also a good way to produce very dull, and ineffective, training.

Policies and procedures define rules, responsibilities and processes. They are reference documents;  thy are not, with all the will in the world, learning materials.

The job is to identify what people need to understand, decide, and do. The learning should be designed around those needs.

AI can help you reach that goal. It can analyse source material, draft scenarios, create scripts, and generate media. It can also help produce assessments and supporting resources. But it should support instructional design, not automate slide production.

Start with the required behaviour

Begin with one question:

What should someone be able to do after completing the learning?

For a policy, this might mean recognising when approval is required or knowing when to escalate an issue. It might involve making the right decision when the situation is ambiguous.

For a procedure, it might mean completing a task correctly or following a process in the right order.

AI can provide a useful first analysis of the source material. It can identify:

• Key obligations
• Responsibilities
• Exceptions
• Approval points
• Likely mistakes
• High-risk decisions
• Information people must remember
• Information they can look up when needed

Human judgement is still needed. While AI can identify what appears in the document, it cannot always judge which parts create the greatest risk in the workplace.

Do not recreate the policy on screen

A policy might contain six pages about conflicts of interest.

The learner probably does not need six pages of e-learning on conflicts of interest.

A better starting point would be a realistic decision:

Your team is choosing a new supplier. One of the companies bidding for the work is owned by a close relative. What should you do?

AI is good at drafting scenarios like this. It can produce examples for different roles, locations, and levels of responsibility. It can also suggest plausible wrong answers. and explain why they are wrong.

The course can then follow the situations people are likely to encounter, rather than the headings in the source document.

Keep the policy as the source of truth

If you copy large sections of a policy into a course, every policy update might require a corresponding course update. The two versions can quickly drift apart. You want to avoid creating a maintenance problem.

Use the e-learning to explain principles, decisions, and behaviours. Link to the current policy or procedure for detailed reference information.

AI can help when the policy changes. It can compare the old and revised versions, identify the differences, and flag the affected learning content.

That reduces the checking workload. However, a policy owner must still approve the revised interpretation.

Use AI for drafts, rather than instructional decisions

AI is good at producing first drafts. It can help create:

• Learning objectives
• Course structures
• Scripts
• Scenarios and examples
• Explanatory feedback
• Job aids
• Captions
• Assessment questions

The output can look polished even when the instructional design is weak.

For example, AI might produce a concise summary of every section in a policy. But employees might only need to recognise four situations in which they must stop and seek approval.

Producing more content, more quickly, does not answer the question of what people need to learn.

Human presenters add authority and context

Human presenters can add authority, context and personality to e-learning. A subject matter expert can explain why a policy exists, where people make mistakes, and what the consequences are.

The problem with doing this has been production time. Until recently, editing presenter-led videos could take far longer than recording them. Someone had to select the best takes, remove pauses, add captions, and place supporting visuals.

AI can reduce that workload. It makes it more practical to include human presenters without turning every video into a lengthy editing project.

AI video should support video editing

AI video tools can generate short illustrative scenes, create captions, and identify useful moments in recorded material. They can also remove pauses, and add visual emphasis, by generating motion graphics.

Video editing tools help you stay in control of the project. For some tasks, they are also much quicker than asking AI to create or revise the result. For that reason, we believe AI should support video editing tools, rather than replace them.

AI video tools can also add video zoom-ins automatically. They can draw attention to a field, button, menu, or other important detail, during a software demonstration.

AI can also help add B-roll

Traditionally, that might mean searching a stock library for a clip of “employee in an office” or “manager talking to colleague”. The problem is that these clips are usually generic and often bear little resemblance to the organisation or situation being described.

Generative video provides another option

Instead of choosing from existing stock footage, you can create a short clip specifically for the course.

For example, you might generate:

• an employee entering a restricted area
• a manager reviewing a request
• someone handling sensitive information in a public place
• a visitor arriving at a secure site
• an employee dealing with a difficult customer situation

The clip can be designed around the learning rather than the other way round.

The latest generation of video tools can also work from image references.

This is particularly interesting because it allows you to create greater visual consistency.

You might provide a reference image of a workplace and generate several variations:

• the same location during the day
• the same location after hours
• the same employee interacting with a visitor
• the same type of scene in another office
• the same process taking place at a different time or under different conditions

This makes it possible to create visual material that feels specific to the course, rather than relying on generic stock video.

Generated video does still needs checking. It can contain visual inconsistencies or details that are inappropriate for the real procedure.

But for short illustrative scenes, it can be a powerful way to support scenario-based learning.

Show procedures rather than describe them

If learners need to use an application or follow a process, show them.

If they must submit an incident report, demonstrate the submission process. If they must approve a request, show the approval steps. If they must complete a form, let them see a realistic example. That gives the person producing the training a practical sequence for the demonstration.

Build scenarios around decisions

Many policies exist because employees must make judgement calls. The training should let them practise those decisions.

For example:

You are offered hospitality by a supplier.

The learner chooses what to do. Then the circumstances change:

The supplier says the event will take place after the contract decision.

Does that affect the answer?

Or:

The value is below the stated approval threshold.

What should the learner do now?

This tests how someone applies the policy, which is is more useful than asking them to recall a threshold from memory.

AI can draft branching paths, alternative situations, and feedback. A subject matter expert should check that the scenarios reflect real decisions and do not introduce misleading exceptions.

Test more than recall

Automatically generated quizzes tend to favour recall questions:

Within how many days should an incident be reported?

Which of the following is confidential information?

These questions are easy to produce and mark, and sometimes they are appropriate. But recognising an answer in a list is weak evidence that someone can make the right decision at work.

Instead, you could ask learners to:

• Classify a situation
• Identify the level of risk
• Put actions in the correct order
• Find missing information
• Complete a form
• Draft a short response
• Review a document and identify problems
• Explain the reason for a decision

These activities can be delivered through interactive exercises and small web-based simulations. The learner produces a response that is closer to the real task.

Create focused simulations

Employees sometimes need to practise a procedure without using a live business system.

A simulation can reproduce the relevant actions and decisions. It does not need to copy every screen or function.

For example, an incident-reporting simulation might ask the learner to:

• Select the correct category
• Enter the relevant information
• Attach the appropriate evidence
• Decide who should be notified
• Submit the report

A purchasing simulation might present a request and ask a manager whether it should be approved.

The simulation only needs to reproduce the parts of the system that matter for the learning. That is enough to reveal whether someone can carry out the procedure.

Use a controlled production process

A practical workflow is:

1. Analyse the policy or procedure. Identify the rules, decisions, risks, exceptions and required actions.
2. Define the learning requirement. Decide what learners need to understand and do.
3. Create the first draft. Develop scenarios, scripts, media ideas, assessments and supporting resources.
4. Review the interpretation. Ask the policy owner or subject matter expert to check the content.
5. Produce and test the learning. Build the activities, demonstrations, simulations and media.
6. Test with representative learners. Check whether they understand the scenarios and can complete the required tasks.
7. Maintain the content. Use the approved policy as the reference point when either the document or the learning changes.

AI can accelerate several stages. It does not replace ownership, review or testing.

From policy writing to learning content

The opportunity is to turn an approved policy into a connected set of learning and performance-support resources. Each resource should match what people need at that point in their work.

That is easier when the policy, procedure and learning content are planned together.

Cherryleaf can help your organisation create clear policies and procedures, then turn them into useful e-learning, demonstrations, scenarios, and supporting guidance. This creates a consistent route from the organisation’s rules to the decisions and actions expected from employees.

The technology can shorten the production process. Someone still has to make the policy accurate, the procedure workable, and the learning useful.