Why scenario questions are different from recall questions
A recall question asks you to state a fact: "What does IaaS stand for?" A scenario question describes a business situation and asks which Azure service, pricing model, or governance tool best applies. These require a different skill — not memorisation, but mapping a described need onto the right Azure capability.
Most AZ-900 exam questions are scenario-based. This means the most effective practice method is working through scenarios repeatedly, not re-reading service descriptions in isolation. If your study routine is mostly reading Microsoft Learn modules passively, you are building recognition but not the applied decision-making the exam actually tests.
Step one: identify the actual requirement being tested
Before looking at any answer options, restate the question in your own words. AZ-900 scenarios often bury the real question inside a paragraph of business context — a company description, a constraint, a stated priority. Strip away the narrative and ask: what specific Azure decision is this question actually testing?
Common question types include "which service should be used," "which pricing model minimises cost," "which redundancy option is required," and "which governance tool enforces this policy." Each type points to a different knowledge area — service selection, cost optimisation, resilience, or governance — and recognising which one you are in narrows your thinking immediately.
Step two: eliminate before you select
With four options, two are usually clearly wrong to anyone with baseline Azure knowledge — they describe a genuinely different service category or contradict a detail explicitly stated in the scenario. Eliminate these first.
This leaves two closer options, and the real decision happens here. The deciding factor is almost always a specific word or phrase in the scenario — "lowest cost" versus "highest availability," "must survive a regional outage" versus "must survive a single server failure," "predictable workload" versus "interruptible batch job." AZ-900 questions are deliberately written so both remaining answers could be reasonable in a vacuum, but only one fits the stated priority.
Step three: know your service-selection decision rules cold
AZ-900 rewards candidates who have internalised simple decision rules rather than memorised service lists. For compute: full OS control needs a Virtual Machine; deploying code without server management points to App Service; event-triggered short-lived code points to Functions. For storage redundancy: LRS survives hardware failure; ZRS survives datacentre failure; GRS survives regional disaster.
Building a mental decision tree for each service category — compute, storage, networking, identity — turns recall into instant pattern-matching under exam time pressure.
Step four: review flagged questions with fresh eyes, not fresh doubt
When using exam mode with a flag-and-review feature, resist changing flagged answers purely because you feel uncertain on a second look. Research on exam-taking behaviour consistently shows first-instinct answers are correct more often than changed answers, for candidates who have genuinely prepared.
Only change a flagged answer if you can articulate a specific new reason on review — a detail you missed on first read, a term you misunderstood. "The other option also seems plausible" is not, on its own, a good enough reason to abandon your original choice.
Put this strategy into practice
Try applying these four steps on our free mock exams — Study Mode shows the reasoning behind every answer.