Measurement Is Design: What L&D Still Won’t Admit

L&D doesn’t have a measurement problem.It has a design problem it keeps trying to measure its way out of.

If you can’t measure it, you didn’t design it.

Because if you don’t know what needs to show up in the work, you’re not designing for performance—you’re designing for activity.

I’ve spent my career in this space.

I’ve been in the rooms where we defined what learning should be—embedded in the flow of work, adaptive, real-time, behavior-driven, and connected to actual performance.

We could define it. We could sketch it. We could sell the vision.

And sometimes we could design it.

But most organizations couldn’t execute it—at scale, reliably, in real conditions where performance actually matters.

Not because it was impossible.

Because they never got close enough to the work.

And once you get that close, everything changes.

You can’t stay at the level of frameworks and programs.You have to define what performance actually looks like—clearly, specifically, in real situations.

You have to decide what “good” is—and how it shows up in the work.

Most organizations never go that far.

That’s the truth.

For a long time, I accepted that as a limitation of tools, systems, and environment.

AI changes that.

Not in theory—in practice.

For the first time, we can build learning that adapts, integrates into work, and generates signal as it happens.

That should be a breakthrough.

But watching what’s happening right now, I think we’re about to miss it again.

Because the issue was never the technology.

It was this:

If you can’t measure it, you can’t improve it—because it was never designed to show up in the first place.

And if you can’t design for it, you’re not driving performance—you’re producing activity, engagement, or education for its own sake.

---

Do We Even Want to Measure Business Impact?

There’s a narrative getting louder in L&D:

• We should be more “realistic” about what we can control

• We shouldn’t be accountable for business impact

• Measuring knowledge and skills is enough

Let’s be clear what’s underneath that.

Capability investments exist because organizations believe they will influence performance—quality, productivity, cost, risk, customer outcomes.

Measuring knowledge and skills is necessary.

But stopping there leaves the business asking:

Did anything actually change in how work gets done?

Not everything we call “learning” is intended to change performance in the same way—and that’s part of the issue. Some of it is awareness, compliance, or educational events. In many cases, the function has expanded to include fundamentally different types of purpose under a single label.

When performance is no longer the clear objective, it’s worth asking: is this what the business intended—and what it believes it’s investing in?

But when performance is the goal, we have to define what should change in the work—and design for it.

By grouping fundamentally different types of purpose under the same umbrella of “learning,” we’ve blurred what each is meant to achieve—and made meaningful measurement harder.

---

A Real Constraint: Access to Data

Learning is often held accountable for outcomes it has no access to measure.

The data exists—but it sits across disparate systems, functions, and priorities that weren’t designed to connect. Most organizations didn’t set out to create this gap. It’s the natural result of how the work evolved.

That’s a real constraint. But even when access exists, measurement becomes irrelevant if what was supposed to change wasn’t designed to show up in the work upfront.

I’ve seen this play out repeatedly: A large-scale program launches—high engagement, strong feedback, leadership visibility. Six months later, the business asks:

Are decisions faster? Is quality higher? Are outcomes different?

And the answer is unclear—not because nothing happened, but because those outcomes were never defined or built into the design.

You don’t need perfect attribution to P&L to answer what has changed for the business.

If performance drivers are defined and baselined—cycle time, deal quality, error rates, decision latency—you can observe whether they move.

That’s not perfect.

But it’s real signal.

And more importantly—it’s useful.

---

The Real Problem: We Skip Diagnosis

Learning is almost never the only lever.

Which is exactly why performance diagnosis has to happen before the training lever is pulled.

Too often, L&D has a hammer—so everything looks like a nail.

Learning becomes the default response instead of one option among many.

But the real constraint might be:

• Incentives

• Process design

• Tools and systems

• Role clarity

• Decision rights

• Accountability

• Information flow

Act like a doctor, not a pharmacist. Diagnose before you prescribe.

A meaningful role for L&D is helping the business determine whether capability is actually the constraint.

And this is where I diverge from a common pattern:

You should not be using measurement to figure out after the fact that knowledge and skills weren’t the problem.

You should know that before you design anything.

If you’re doing real performance analysis upfront, you’re not guessing—you’re confirming.

Measurement then becomes what it should be:

Validation that the intervention worked—not discovery that you chose the wrong problem.

Because if we’re finding out after the fact that learning wasn’t enough, the real question is:

Why did we design it that way to begin with?

This becomes even more critical in leadership development.

Take judgment—one of the most cited and least defined capabilities.

When performance matters, the unit isn’t content or skills—it’s the decisions people make in the work. Skills only matter if they change those decisions.

Most programs claim to build better judgment, but never define how it should show up in actual decisions—under real conditions, with real trade-offs.

So what changes?

Are decisions faster—or just more discussed? Is alignment stronger—or just more agreeable? Is rework reduced—or just redistributed?

If the quality, speed, or downstream impact of decisions doesn’t change, then judgment didn’t change.

And if we didn’t define how it should show up in the work, we didn’t design for it.

---

What L&D Actually Owns

L&D doesn’t own business outcomes.

No single function does.

But it is responsible for connecting capability work to the performance drivers the business actually cares about—without over-claiming attribution.

Because the role—when done right—was never about content.

It was about performance.

Individual. Collective. And ultimately business.

It was about understanding the system people operate in—and designing for behavior change inside that system.

Across workflows. Across functions. Across the business.

Rigor isn’t about more measurement. It’s about designing what needs to change—and making it visible.

And if that’s the job:

Measurement isn’t optional.It’s the foundation.

---

What This Looks Like in Practice

These are the moves most teams skip—and why measurement never lands:

1. Start with the business problem

Where are we today vs. where we need to be? What’s blocking performance?

2. Define performance-based objectives in your learning solution design

If you can’t describe what “good” looks like in real conditions, you can’t design—or measure—it. When done well, objectives should ladder up to your leading indicators (more below).

3. Treat learning as a process, not an event

Sequence matters. Modality matters. Measurement happens along the way—not just at the end.

4. Use leading + lagging indicators together

Lagging = outcomes (revenue, margin, KPIs)

Leading = behavior and performance shifts that drive those outcomes

If you only look at lagging indicators, you’re too late.

5. Measure inside the work

Go where the work lives—CRM, workflows, systems. That’s where signal exists.

6. Use multiple measurement vehicles

Surveys alone aren’t enough. Combine assessments, observations, and performance data from the work itself.

7. Design for context

What you measure will vary by business, role, and objective. That’s expected.

8. Establish a baseline

Measure before and after. Show change.

9. Use pilots and comparison groups

Look at differences between those who experienced the intervention and those who didn’t.

10. Anchor to business value

If you can’t state the intended impact, you’re too far from the business.

11. Use measurement to improve design

If the data doesn’t change the solution—or future decisions—it’s just reporting.

---

Final Thought

For years, we said we wanted learning in the flow of work.

Real-time feedback.

Behavior change at scale.

Now we have the tools to actually do it.

But the requirement hasn’t changed.

If you want performance, you have to design for it. And if you design for it, you have to be able to measure it.

Everything else is activity.

People Performance