The Analysis Phase in Instructional Design Is Lying to You

We keep designing for the learner leadership describes. Not the one who actually exists.

I want to tell you something that took me years to say out loud.

The model most of us were trained on has a problem.

Not a small one.

A foundational one.

And it lives in the very first phase.

---

ADDIE isn’t broken.

Analyze. Design. Develop. Implement. Evaluate.

It’s still the backbone of instructional design worldwide. Over 40% of L&D practitioners globally still build from it. The framework is solid.

But here’s what nobody says in the certification courses:

In real-life application, stages get skipped. Phases get merged.

And the one that suffers most, every single time, is Analysis.

Not because we don’t value it.

Because most of us were taught to run it in a way that gets us close to the right answer.

Just not close enough to the right human.

---

Here’s the pressure that makes this urgent

According to ATD’s 2025 State of the Industry report, employees received an average of 13.7 formal learning hours in 2024.

In 2020 that number was 35.

That’s a 60% drop in four years.

At the same time, more than 90% of organizations are currently facing a major or minor leadership skills gap.

Less time.
Bigger gap.
Same broken Analysis phase running the show.

When every hour of learning time is this scarce, designing the wrong thing isn’t just inefficient.

It’s a decision you can’t afford.

What traditional analysis actually looks like

Survey managers.
Review performance data.
Interview a subject matter expert.
Document the gap.
Build what leadership says the problem is.

It’s efficient.
It gets sign-off.
It produces a needs assessment that looks rigorous.

And it almost always misses what’s actually happening.

Because the people closest to the work, doing it under real pressure, with real constraints, are rarely the primary source of insight.

We design for the learner leadership describes.
Not the one who actually exists.

That gap is where training disappears.

What UX research figured out that L&D hasn’t fully adopted

The discipline behind every product and app you use daily operates on one principle:

You cannot design something useful for a person you haven’t actually observed.

According to the Nielsen Norman Group, contextual inquiry is an ethnographic field study involving in-depth observation and interviews, conducted in the user’s natural environment.

Not a conference room.
Not filtered through a survey.
Not interpreted by a manager.

In the place where the actual work happens.

Unlike questionnaires, contextual inquiry is more interactive than usability testing and more revealing than standard interviews.

We ask what people think they need.
UX researchers watch what people actually do.

Those are two completely different datasets.

One of them produces training that changes behavior.

What I found when I started doing this differently

After the rollout failure I wrote about last week, I rebuilt my entire discovery process using UX methodology.

I stopped starting with learning objectives.

I went into the field. Watched people work. Stopped asking “what do you need to learn?” and started asking “walk me through the last time you faced this situation.”

What those conversations revealed was not what leadership told me.
Not what surveys showed.
Not what subject matter experts described.

The real friction lived in undocumented moments.

Workarounds people invented because the official process didn’t match reality.
Decision points where training gave knowledge but not judgment.

And judgment was the only thing that mattered when the moment arrived.

You can’t find that in a needs assessment.

You can only find it by getting close enough to the human to see it.

The UX-Informed Analysis Framework

Before your next learning project, run these four steps before touching a single learning objective:

1. Contextual Observation
Watch the work happen in real conditions. No agenda. You’re looking for hesitations, shortcuts, and the moments the process breaks down in practice.

2. Friction Mapping
Find where performance actually breaks, not where managers think it does. Every workaround is a signal the official process failed someone. Design for that.

3. Assumption Auditing
Write down every assumption your team has made about the learner. Verify at least three before building anything. You will be wrong about at least one.

4. Decision-Point Interviews
Replace “what do you need to know?” with “what decision are you trying to make and what makes it hard?” This one shift changes everything you build.

One more thing worth naming

AI is changing how fast we can build.

Content that took weeks now takes hours.
Personalization at scale is becoming possible.
Translation and localization are faster than ever.

But none of that changes what happens when the Analysis phase produces the wrong brief.

Faster production of the wrong design is still the wrong design.

The Analysis phase isn’t just the first step.

It’s the most important investment you can make in the entire project.

Not Analyze.

Understand.

— Ali🧡