How to Create Employee Training Programs That Stick

It is estimated that while American industries annually spend up to $100 billion on training and development, not more than 10% of these expenditures actually result in transfer to the job.

Does that statistic feel familiar? 

Would you be surprised if I told you that it came from a 1988 article about transfer of training? 

The point is, it’s been nearly 40 years since Baldwin and Ford introduced their transfer of training model to illustrate how workplace training translates into job performance through inputs, outcomes, and conditions.

While I won’t be doing a deep dive into that model, I bring it up to illustrate that we’ve been having the same conversations (and citing the same statistics) for decades. These conversations lead us back to one place: how can we design programs that translate into measurable outcomes, for our employees and the business. 

But it feels like the conversation is finally progressing, and that’s due in large part to AI.

What is an employee training program?

As I reflected on how I would articulate the definition of an employee training program, I did something I often do these days. I went to an LLM and asked them to explain an employee training program to me, like I was 5 years-old. Here’s what they said: 

Someone teaches you how to do a job. First you watch them. Then you try it yourself. They help you when you mess up. You keep practicing until you're really good at it. Then you do the real job. But you have to keep doing it. If you stop doing it, you forget. So they keep reminding you and helping you practice again. Eventually it becomes something you just do without thinking.

Truly, it’s that simple. 

And yet, enterprises struggle to design effective training programs. That’s because we design, deliver, and then move on. We spend so much time crafting the perfect experience, that we rarely have the capacity to measure whether the capability has been developed. 

But the reality is training programs can shape expectations, only if they’re reinforced. If you teach a manager how to coach an employee in an hour session, but never give them an opportunity to practice (let alone gentle reminders of how to do it months later), then how can we expect them to become good coaches? 

That’s where AI is changing the way we should think about training programs. Instead of discrete experiences, they’re robust systems that include learning experiences, but are surrounded by scalable implementation efforts, like practice loops and reinforcement. 

And for the first time in my career, I’m confident that’s possible to execute sustainably at scale.

Let me show you how. 

Phase 1: Analyze

Before you can build any training program you need to know what's not working for your targeted audience.

Grab your detective hat, because we’re doing some sleuthing to find where employee performance is diverging from expectations.

This is a needs analysis. You want to move from generic training requests, like "we need to develop our managers" to concrete behavioral problems, like "our managers are not delivering constructive feedback to their direct reports in a timely manner."

1. ‍Define the performance gap
   ‍What should people be doing? What are they actually doing instead? Where can you observe this behavior and how?‍
2. Gather evidence
   ‍Search data you have access to, conduct listening sessions or interviews with stakeholders, or directly observe. Multiple sources give you a fuller picture than any single one.‍
3. Synthesize patterns
   ‍Collate the data to understand root causes of the performance gap. Look for consistent breakdowns across inputs.‍
4. Validate your findings
   ‍Ask 3–5 stakeholders (a manager, an HR Business Partner, a member of the targeted audience) to review your findings. Don't skip validation! ‍
5. Translate into a capability statement
   ‍Write something like: "After participating in this training program on [X], [target audience] should be able to [observable action] so that [business outcome]."‍
6. Align on success measures
   ‍Determine what observable evidence can support whether or not your training impacted the performance gap.

I have included sample prompts to illustrate how you might use an LLM to support your training program (you can copy and paste them). If you’re working with internal or sensitive information, follow your organization’s data and AI governance policies. Avoid including personal data or confidential business details unless you’re using an approved, secure environment.

Example: Manager training program

People leave managers, not companies. You've likely heard this before. It surfaces in conversations about engagement, burnout, or turnover.

But you also know how hard it can be to change how people manage. That's why building a manager training program is so difficult.

So throughout this post, I'm going to focus on managers, specifically training them to coach their employees using the GROW (Goal, Reality, Options or Obstacles, and Way Forward) framework.

During the needs analysis, I discovered that my target audience — newly promoted managers in engineering, more precisely employees who have never previously managed people — are struggling in their 1:1s with direct reports.

There's a shared belief, validated by their managers, HR BPs, and members of the group, that as managers they should have all the answers for their direct reports. They're spending too much time trying to solve their directs' problems, without helping their directs build the capability to take accountability and ownership for their problems and long-term development.

I'll return to this example to demonstrate how to build a training system that supports managers' coaching capabilities.

Phase 2: Design

The next phase builds off your needs assessment. You're going to take the capability statement you wrote, and the success measures you identified, and begin planning the learning experience.

My capability statement looks something like this:

"After participating in this training program on the GROW coaching framework, newly-promoted, first-time people managers in engineering should be able to ask goal-focused questions to surface obstacles and options and confirm the direct report identifies and owns their own next step so that their direct employees report increased clarity on expectations and meet or exceed performance targets."

I know, it’s a mouthful, but you’re not publishing it anywhere. This is for you to document so you can successfully design an effective training program. You’ll appreciate the level of detail later. 

With your capability statement drafted, you can also write your learning objectives. Learning objectives break down the capability into specific, observable actions people will practice and be assessed on. They're the building blocks of your program.

In our example, they could look like: 

1. Managers can guide a direct report to identify their own options and next steps in a coaching conversation, demonstrating that they trust the person's capability to solve the problem.
2. Managers can recognize when they're defaulting to problem-solving mode and redirect the conversation back to coaching.
3. Managers can end a coaching conversation where the direct report has identified at least one concrete action they will take independently.

From there, you can also determine what content and assessment methods may support the learning experience. In situations like this, where the training is focused on skill-building, I'd highly recommend identifying opportunities for practice and feedback, whether with facilitated exercises or even with peers.

Phase 3: Develop

With your learning design in place, you can finally start developing your program. This is what most people think of when they think of "creating a training program" – the content.

But this phase is about more than authoring courses or crafting workshops. It's about developing an intentionally sequenced system that supports learning, practice, feedback, and reinforcement.

That's why instructional designers often blend delivery methods. Different methods serve different parts of your program.

A common assumption is that training programs need a live component. While real-time interaction can build culture and provide immediate feedback, not every program depends on it. Importantly, you may need to shift to less live facilitation to reduce training costs and efficiently scale programs.

For our manager training program, here's what I might consider as the learning architecture: 

1. Live kickoff session 
   Share the purpose of the training and why it matters. (If the cohort is pressed for time, this could easily be delivered asynchronously.) Whenever possible, use these live moments to build psychological safety, but they're not required.
2. GROW training video
   Assign a brief lesson on the framework, like the video I shared above. Something that could be completed by a busy engineering manager on their own schedule.
3. Facilitated practice sessions
   Organize sessions where managers take turns giving and receiving feedback and role-play different scenarios using the GROW framework.
4. Reinforcement in flow of work 
   Send quick reminders about each stage of the framework with a script or tip, once a week. Right before performance reviews, send a scenario-based video on how to incorporate GROW into a performance conversation.

Use an LLM to help you think through the learning architecture of a training program, and how to make it more sustainable for scaling audience. Here’s a prompt to get you started. 

Phase 4: Implement

At this point, it’s time to implement your training plan.

Whenever possible, I recommend running a pilot, ideally for each component of the training program. (You don’t have to run them concurrently, unless you have time for a full run-through.) Pilots give you an opportunity to see where there’s friction in the design and delivery, whether that’s issues with Zoom breakout rooms or instructions in a practice scenario being confusing. 

The challenge of implementation is that you’re managing concurrent workstreams, such as: 

• Logistics: everything from setting up access to confirm all the technical setup works
• Learner: responding to questions, troubleshooting any issues, and monitoring engagement
• Facilitation: whether you're facilitating or someone else, you'll want to think through a guide
• Feedback collection: observing what's working well, and what isn't 

For instance, our manager training program would have you scheduling the kickoff workshop and practice sessions, while also sending out the GROW video. 

At the same time, you’d want to be tracking employee attendance and engagement with the content. Are managers watching (or rewatching) the video? Or dropping off at a specific time? In practice sessions, are they actively listening? Asking more questions than they’re answering? 

You can see the laundry list of tasks starting to add up. 

I could write an opus on the work involved with this phase for training programs.

My recommendation is to find someone who excels at project management. Ask them to help you create a playbook for the program. You know all the little details from adding last-minute registrants to a calendar to making sure a facilitator has host permissions.

And then, wherever possible, see what you can automate with AI (everything from enrollment workflows to automated content delivery). That's the key to scaling training programs sustainably.

Phase 5: Evaluate

We've finally arrived at the last (and most important) phase, evaluation. Remember, this is the difference between a program that's completed and a program that drives behavioral change.

That begins with a rigorous measurement of the training. In Phase 2, I briefly mentioned needing to map out assessments, and then skipped right along. Well, I want to come back to it now.

In L&D, I've seen a variety of assessments being used for training programs. That includes diagnostic assessments (pre-training), formative assessments (during training), and summative assessments (post-training), plus behavioral assessments over time.

There are so many ways to assess how employees show up and engage, whether that's knowledge checks, case studies, or final presentations. And here's what I'll say. I'm not sure which ones you pick for your learning design really matter all that much.

The workplace is an imperfect learning environment. A manager failing a knowledge check in the GROW video, or having to miss a practice day because of a work conflict, doesn't make or break their experience of the training program.

That's why, generally speaking, I'm less concerned with individual assessment results than with business outcomes (there are, of course, exceptions to that). If the manager who failed the knowledge check is able to coach a direct report through developing the experience needed to be promoted into a critical role, then that's what matters.

Because I'm able to show the direct report had increased clarity on their role and exceeded their performance targets, and therefore, I have evidence that the training program served its purpose.

There are so many ways to approach measuring training programs. I go through several in this guide to calculating ROI.