In this conversational interview, Christopher Parsons and a guest (Speaker B) critique the current state of Learning & Development (L&D) in organizations. They argue that L&D teams focus too heavily on producing content—PDFs, PowerPoints, and knowledge quizzes—rather than on practice that leads to behavior change. The discussion highlights the lack of meaningful measurement beyond course completion, noting that organizations rarely assess whether training actually moves business metrics such as sales cycle time or productivity. They reference learning‑science research (e.g., Pooja Agarwal) to illustrate that the traditional 80% content / 20% practice split is misguided; instead, practice and high‑level scenario‑based assessment should dominate. The speakers differentiate low‑level recall questions from high‑level decision‑making scenarios, providing an architectural example to show how a factual question can be transformed into an applied problem. They also discuss practical ways to convert standard multiple‑choice items into mini‑scenarios that give targeted feedback, emphasizing that this is not rocket science but requires deliberate design. Finally, they anticipate future pressure from finance leaders demanding ROI evidence for training spend, urging L&D to proactively adopt impact‑focused metrics and design.
Chapters
Do you think learning development teams try and take on too much in terms of scope? It's kind of implicit in what you just said is that maybe they're doing too much that maybe could be culled back for some less effective ones.
I'm not— they think they're doing too much, but I think they're doing too much of the wrong stuff.
Okay.
So I think they're doing too much content development. Back to how we perceive school. If I give people new information, they're gonna change their behavior. And that actually is, I believe, and I don't really have good evidence. I don't think there's contrary evidence, but I just don't really have good story. But we used to believe that we were formal logical reasoning beings.
Right. And we are feeling machines that happen to think, right?
Right, right. And if we, so then if that's true, we could give people new information that would change their behavior.
Yeah.
Oh, logical. It's, this is not right. This is right. Okay, I'm gonna change. But we're not. We're complex beings, as you say, feeling machines with capability to think, not thinking machines with the capability to feel. And, but it's not even explicit, but I believe it's implicit in so much of what we see in organizations. Yearly reviews, that's just been proven to be worthless. Why do we continue to do that? Even twice yearly reviews isn't frequently enough. People really need regular feedback. And the list goes on, bullet point and whatever training. Just giving people information isn't going to change their behavior. And yet too often, here's the content, PDF, PowerPoint, add a knowledge quiz and you're done. That's— and people now want to do that faster and faster. And— Why do you think—
I mean, we're in 2026. Like, I've heard what you just said For a few decades now. Like, why is it taking— why is this still a problem?
Why has education been recognized as ineffective for decades and it's still true? I think there are multiple reasons. L&D isn't very good at communicating and has been quite comfortable just taking orders and producing courses. Nobody cares. It does— they don't measure, so they don't know that they're not effective.
And they don't measure beyond did somebody like the course, right? Did they complete the course? That's where it's at.
Well, that— it's nice if they even do that. Most people do that, but not everybody.
But what they don't measure is did this person move the needle? Did their sales cycle shorten? Did they do their door schedule faster? Did they make less? Yeah.
Even can they now do it? Are they still doing it in the workplace? And is that leading to the change we need? That's the story we should have and we don't.
You mean, can they do it more than 10 minutes after the training?
Right.
You know, that's— yeah, yeah, yeah.
Oh, and there's a reliable result that if you just do training, 3 days later it's gone if you don't reactivate it. Working with a startup, Elevator 9, that's trying to develop a solution and using learning science and technology to, to follow up on face-to-face courses so that learning does persist and actually transitions from Not just can you recite the information or solve mini scenarios, but what's happening in the workplace and are you ultimately seeing any impact change?
Hmm.
So we don't measure, we have bad ideas about what good course design is, and our stakeholders, if it look— think that if it looks like school, it must be effective. So they're not really asking for this data. I'm still waiting for that to change, by the way. Eventually CFO's gonna go, we're spending money on training, What evidence do you have that any of this is doing any impact? And I would rather L&D be changing to prepare for that rather than faced with that as a sudden and surprising outcome.
We have bad ideas about good course design. What is good course design?
Well, as I mentioned, practice. I believe strongly that we've got our ratio wrong. We tend to have about 80% content and 20% practice. And most of that's knowledge test. And there's research that shows— Pooja Agarwal showed that we used to have this belief that we need to make sure they have the information before we test them to see if they can use it.
Hmm.
And so there were low-level questions and high-level questions. And she clearly— the final test is, can they do high-level things? But used to believe that you need to test knowledge and then do this. Lots of people just test the knowledge. That's the problem, as I was mentioning, is a problem. But if you test the knowledge and do the high-level questions, you should transfer to here. But it turns out if you just ask high-level questions, you still get that transfer to ability to do. You don't need the low-level knowledge questions because it turns out here, You have to retrieve the knowledge and apply it. Here, here you're just retrieving the knowledge, but ultimately you're gonna have to retrieve the knowledge and apply it.
Can you give me an example like of what like a high-level, low-level would be in this scenario?
Yeah, let me see if I can make an architecture one. So level question would be, what are the requirements for a weight-bearing beam in terms of the current code?
Okay.
And you recite an answer.
That's just a fact.
The current code says that a weight-bearing beam has to be this long and sustain this much weight.
Mm-hmm.
That doesn't mean you can design. What you then need to do is say, here's a building that has this much weight. How many columns does it need? I'm making this up 'cause I'm not an architect, but that difference in question is applying the knowledge about that weight to then say, what do people really need to make decisions on? What are the decisions they need to make in the real world? Not as, not what are the knowledge they need, but what are the decisions they make? And just asking those questions requires you to retrieve the knowledge and apply it the same as just asking the knowledge retrieval questions. So you don't need to ask those knowledge retrieval questions.
Is it, is partly the reason that people do these courses this way, I'm sure there's a lot of underlying, but I'll ask if one of them is, it's easier to test on the low-level stuff because that could be multiple choice and fill it in where the other one has to be an interactive exercise. Someone has to evaluate whether or not their design that they came with would work. Do you think that, is that part of this?
Yes, and I do think it's part of it. It turns out I recently ran for the second time, actually the third, a workshop on writing mini scenarios. You can take multiple choice questions and turn them into scenarios. So instead of asking, you ask that question about how many columns do you need, and then you have several ways people reliably go wrong, like too many, The right number and too few. And you have specific feedback. That might seem like enough. You were probably assuming X, but you forgot to add in Y. Or yeah, this many more would make sure, but you're not maximizing the space because you're having too many columns. And the right answer says yes, the formula says this, and that means you've rightly addressed it. So you can write better multiple choice questions.
Mm-hmm. Hmm.
Even with your standard multiple choice question tools, you can still do better jobs of writing mini-scenarios, but you have to know that's important. And it turns out it's not rocket science, but people can reliably go wrong. One of the things I found the first time I was trying to train people to do this is they would write decent scenarios, but then the alternatives would still be categorizing things instead of making decisions.
Hmm.
And so that's why I created the workshop and ran the workshop now twice. And it's challenging to get people to get that message, but it's valuable. And so it's worthwhile to recognize and start thinking about. So the most important thing is to start making realistic practice and having a fair amount of it because that's what makes it work.
