Layered analysis: how to answer IAPP exam questions that stack facts
Many IAPP exam questions are layered. To reach the answer, you have to work through several facts in order, such as scope, risk category and penalty. Each layer is something you either remember or you don't. Name the layers, answer them one at a time, and when you miss a question, find the layer that stopped you.
IAPP exam questions are a lot like onions.
To answer them correctly, you have to peel back multiple layers. Each layer is a fact you either know or you don't.
I've coached more than 12,000 students across 120+ countries, and this idea helps with almost every scenario question. Try the question below first.
Try this AIGP-style question first
This is an illustrative question I wrote in the style of the AIGP exam. It is not an actual IAPP exam question.
Meridian Workforce Analytics, a Chicago-based software company with no offices, subsidiaries or legal establishment in the European Union, licenses its "ShiftSense" platform to a logistics company operating warehouses in Rotterdam. ShiftSense analyzes real-time webcam and microphone feeds from warehouse workstations to infer each employee's emotional state during their shift. It flags workers it scores as "frustrated" or "disengaged" so that floor supervisors in Rotterdam can intervene. A Dutch market surveillance authority opens an investigation. If Meridian is found to be in violation of the EU AI Act, what is the maximum administrative fine it faces?
A. Up to €35 million or 7% of total worldwide annual turnover, whichever is higher B. Up to €20 million or 4% of total worldwide annual turnover, whichever is higher C. Up to €15 million or 3% of total worldwide annual turnover, whichever is higher D. No penalty. Meridian does not fall within the scope of the EU AI Act because it has no establishment in the European Union.
What does layered analysis mean on an IAPP exam?
A layered question hides several smaller questions inside one stem. You answer them in order, and each depends on a fact you've memorized.
This is where memorization earns its keep. On Bloom's Taxonomy, the two foundational levels are remember and understand. If you haven't committed the right facts to memory, you can't carry out the analysis, and your odds of answering correctly drop sharply. (I go deeper on this in my post on Bloom's Taxonomy and the IAPP exam.)
The five layers in the Meridian question
Here are the five layers in the question above:
- Extraterritoriality. Does a US company with no EU establishment fall within the scope of the EU AI Act?
- The risk categories of the EU AI Act. Most courses, mine included, teach these as four tiers: prohibited, high-risk, limited (transparency) risk and minimal risk.
- Examples of systems that fall within each category.
- The types of non-compliance that can trigger a monetary penalty.
- The specific penalty attached to each type of violation.
All five layers are factual. Know them, and the analysis falls into place.
Peeling it back, one layer at a time
Layer one: scope. Under Article 2(1) of the EU AI Act, the Act applies to providers placing AI systems on the EU market, wherever they are established. It also covers providers and deployers in a third country when the system's output is used in the EU. Meridian licenses ShiftSense into Rotterdam, and the output is used there. No European establishment required. That rules out D.
Layers two and three: risk category. Article 5(1)(f) prohibits AI systems that infer the emotions of a natural person in the workplace or in education institutions, unless the system is intended for medical or safety reasons. ShiftSense infers warehouse workers' emotions so supervisors can intervene. That's a prohibited practice. The system should never have been deployed.
Layers four and five: violation and penalty. Article 99 sets three fine levels. Breaching an Article 5 prohibition carries the highest: up to €35 million or 7% of total worldwide annual turnover for the preceding financial year, whichever is higher. Breaches of most other operator obligations carry up to €15 million or 3%. Supplying incorrect or misleading information to authorities carries up to €7.5 million or 1%.
The answer is A.
Notice where B came from. €20 million or 4% is the GDPR's higher fine tier in Article 83(5), a familiar number from a neighboring law. C is a real AI Act number for the wrong kind of violation.
One nuance worth knowing for the real exam: for SMEs and start-ups, Article 99(6) caps each fine at whichever amount is lower, not higher. The scenario doesn't tell you Meridian's size, so the general rule applies.
Which layer stopped you?
If you missed it, ask yourself which layer stopped you. That's not a failure. That's a study plan.
Here's how I'd turn a miss into a study task:
| Layer you missed | What to study next |
|---|---|
| Scope | Article 2 of the AI Act, and how it compares with GDPR territorial scope |
| Risk category | The list of prohibited practices in Article 5 |
| Examples | Sort ten real-world systems into categories without notes |
| Violation type | Which obligations sit with each operator role |
| Penalty | The three fine levels in Article 99, side by side with the GDPR's two |
How to practice layered analysis
Take any practice question you get wrong and run three steps.
- List the layers. Write down every fact the question needed, in the order you'd need it.
- Mark each layer. Did you know it cold, half know it, or not know it at all?
- Study the weakest layer first. Then retry the question a few days later.
This pairs well with the two-part reading framework: paraphrasing the question helps you see the layers before you answer. Then check the last step for qualifiers like "first" or "best".
Layering isn't unique to the AIGP. A CIPP/E question might stack territorial scope, role and lawful basis. A CIPM question might stack program stage and the next best action. The move stays the same.
You can't peel back layers you never learned. Commit the facts to memory first. The analysis will follow.
Frequently asked questions
Are IAPP exam questions mostly memorization or analysis?
Both. Most scenario questions ask you to analyze, but the analysis rests on facts you have memorized. If one fact is missing, the whole chain breaks, which is why I tell students to build the memory layer first.
What should I do if I don't know one of the layers during the exam?
Use the layers you do know to eliminate answers. In the Meridian example, knowing scope alone removes D. Then pick from what's left and move on.
Is emotion recognition always prohibited under the EU AI Act?
No. Article 5(1)(f) prohibits it in the workplace and in education institutions, with an exception for medical or safety reasons. Outside those settings, emotion recognition systems are treated differently, so read the scenario's setting carefully.
Keep reading
- Bloom's Taxonomy and the IAPP exam: why the remember and understand levels come first.
- How to read IAPP exam questions carefully: a framework for spotting each layer before you answer.
- Partially true distractors on the IAPP exam: why answers like B above look so convincing.
Succeeding on IAPP exams takes the right resources and the right strategy. If you're preparing for the AIGP and want help building every layer, my AIGP Certification Masterclass walks you through the material and the exam strategy. Take a look whenever you're ready. I'd be glad to help.
Preparing for an IAPP Exam?

- Free
- 53 minutes
- 10 techniques
