Most of us in learning and development have seen some version of this: a new training video goes out to a few hundred learners, the completion rate climbs past 90 percent within a couple of weeks and everyone feels good about the launch.
Yet a month or two later, the same mistakes are still showing up on the floor, in the call queue or the audit report.
After over 20 years of producing video for corporate, health care and manufacturing teams, I’ve found that you need to dig deeper beyond the video itself to identify the problem. The lighting and narrative can both be excellent, but the trouble usually starts before production.
Often, it boils down to the difference between a video built around explaining a topic rather than helping employees perform the work.
The good news is that closing this gap doesn’t usually require a bigger budget. It simply requires a few distinct conversations before anyone picks up a camera, a modification to the video structure and a willingness to measure something less intuitive than views.
Define the job behavior first
Most training video requests arrive as topics: “We need something on the new expense system.” “Can we do a safety video for the warehouse?” “Leadership wants a module on giving feedback.” Those are reasonable starting points, but a topic doesn’t tell you what a learner should be doing differently once the video is over, and that change is why the organization is making this investment.
So, before we write the script, we ask the chief learning officer (and ideally a manager who sees the work every day) to describe the behavior in plain, observable terms. We usually start with these questions: What should the learner do, in what situation and to what standard? Where do learners typically go wrong today, and what does that mistake cost? And how would a manager know, just by watching, that someone got it right?
The answers often sound far less exciting, but they give actionable insight into the tasks. “Warehouse safety” becomes something like this: when a conveyor jams, the associate locks out the line and confirms it’s de-energized before reaching in. “Giving feedback” becomes: within a day of a missed deadline, the manager describes what happened and its impact on the team, then asks for the employee’s view before agreeing on next steps.
The difference is clear; capturing specific descriptions of each task and associated pain point means viewers of this training video can now practice and be observed later on the job.
This step also works as a helpful filter. Sometimes it becomes clear that learners already know what to do and something else is getting in the way, such as a confusing screen, a missing tool or a process that inadvertently rewards a shortcut. It’s much better to learn these lessons in a pre-production planning meeting than after the production budget has been spent.
Remember that subject matter experts (with the very best intentions) usually want to include everything they know. Defining the behavior first gives everyone a fair way to decide what makes the cut. Anything that doesn’t help someone perform the exact standard operating procedure can typically live in a job aid document instead.
Show the task, then let learners practice it
Once the behavior is defined, the next decision is how to teach it on screen. The most common approach, and in our experience the least effective, is the narrated explanation: a presenter or voice-over describing the steps while general footage of the workplace plays underneath. It feels efficient, but hearing someone talk about a task is a very different experience from watching the task being done.
A meta-analysis by Paul Taylor, Darlene Russ-Eft and Daniel Chan, published in the Journal of Applied Psychology, reviewed 117 studies of training that demonstrates the desired behavior and then has the learners practice it. They found that the knowledge gains diminished over time, while the effects on skills and job behavior held steady or increased. Transfer to the job was strongest when learners saw both positive and negative examples, practiced with scenarios they generated themselves, set goals and when their managers were trained as well.
Several of those findings translate directly into production decisions, and we have found these to make the biggest difference:
- Film the task in the real environment, performed by someone who does the job. On our shoots, for example, we often ask the most experienced operator to walk us through the process before we film anything, and it’s surprising how often they perform a step that never made it into the script. If it aligns with legal and compliance, that step is worth capturing.
- Show the common mistake and how to recover from it alongside the correct method and show why it goes wrong. A short “here’s what usually happens when things go wrong” moment gives learners something they’ll recognize on the job, which lines up with the findings on positive and negative examples.
- Frame the work from the worker’s point of view—or a POV or over-the-shoulder shot—whenever you can. A camera placed over the operator’s shoulder shows what they see and where their hands go, which a wide shot of the room cannot.
With the demonstration doing most of the teaching, narration reinforces what the eyes see: the reason for a step, the sound or feel to watch for or even the consequence of skipping a process. Be sure to include captions and a transcript as well, so the video works for every learner.
Karl Szpunar, Novall Khan and Daniel Schacter at Harvard studied what happened when a recorded lecture was divided into short segments with brief quizzes in between. Undergraduates who were tested between segments reported mind-wandering on about 19 percent of attention checks, compared with roughly 40 percent in the groups that weren’t tested, and they scored higher on the final test. This principle carries over to corporate L&D: When learners know they’ll be asked to use what they’re seeing, they watch with a purpose.
In practice, that means breaking the video at the decision points in the task and pausing after each one with a realistic question, such as “The line has stopped, but the indicator light is still on. What do you do next?” It also means planning the hands-on piece (a supervised attempt, role-play or simulation) from the start. And, given what the Taylor research found about managers, it helps to let managers know what to look for.
Measure what happens after
Completion rates and view counts are easy to collect, so they tend to become the scoreboard. Philip Guo, Juho Kim and Rob Rubin’s study of 6.9 million video-watching sessions on edX used watch time as its main engagement measure and found that shorter videos held attention much better. That’s useful to know, although the task should still decide the length—a procedure with eight critical steps needs room to show all eight. But watch time measures attention, which is only the first step toward changing on-the-job behavior.
Will Thalheimer’s Learning-Transfer Evaluation Model, or LTEM, offers a clear path for what comes next. It separates knowledge from decision-making competence and task competence, and then looks at transfer to the job and the effects of that transfer. For a training video, we have found that a practical measurement plan often follows the same progression:
Comprehension: Embed realistic situation scenario questions between segments that ask learners to make decisions. Recalling a fact from the video is a much weaker signal than choosing the right action.
Task competence: Soon after training, have learners perform the task while a manager or peer uses a short checklist. This is where defining observable behavior at the start really pays off.
Application on the job: At 30, 60 and 90 days, managers should look for the behavior in the actual work and check the business measure you agreed on at the start. For tasks learners perform only occasionally, a short refresher scenario helps, and someone should own updating the video whenever the procedure changes.
This connects closely to a point Cheryl Jackson and Peggy Durbin made recently in Chief Learning Officer: Observable behavior is a leading signal that tends to move before lagging business metrics do. Choose that measure carefully, because a number like “fewer reported incidents” can improve for the wrong reasons.
There’s an important reality to acknowledge here as well. A single video rarely deserves full credit (or full blame) for what happens on the job, because managers, tools, workload and incentives all play a part. That’s why the manager’s role matters so much, and why agreeing on the behavior and the metrics before production begins gives everyone a fair baseline and a shared definition of success.
Before the next request
Most of this comes down to asking better questions at the beginning of a project. Here are a few worth keeping handy the next time a video request lands on your desk:
- What should learners do differently after watching, and in what situation?
- Where do people go wrong today, and what does that cost the business?
- Can we film the real task, with the people who actually do it?
- Where will learners pause to make a decision or practice?
- How will a manager know it worked, and when will we check?
Video is still one of the most useful popular tools an L&D team has, especially for tasks people need to see before they can do them well. When it’s designed around the job instead of the viewing experience, it usually earns its place in the learning plan.
If you’re looking for a place to start, try these questions on an SOP already in your content library. You might be surprised by the answers.
