Why People Fail Azure AZ-104 (and How to Pass)
AZ-104 is one of the tougher Associate exams — and most people who fail it studied hard. They failed because they prepared for the wrong exam: AZ-900 depth, no portal time, and a blind spot on networking. Here are the 8 mistakes that actually sink candidates, why each happens, and the exact fix.
01 The real numbers
Microsoft does not publish official pass rates for AZ-104, so treat every figure as an estimate rather than gospel. What is consistent across community reports is the feeling: AZ-104 is widely regarded as one of the harder Associate-level exams, noticeably tougher than the AZ-900 fundamentals exam that many candidates take first. A meaningful share of people fail on their first attempt — and most of them studied for weeks. They failed because of how they prepared, not how little.
The exam is scored on a scaled 1–1000 range and you need 700 to pass. Because scoring is scaled, 700 is not the same as 70% of questions correct — difficulty is normalised across the item bank, and weighting differs by question type. You will face roughly 40–60 questions in 120 minutes, in a mix of multiple-choice, multiple-response, drag-and-drop, build-list and a case study. Some sittings also include hands-on lab tasks where you perform real actions in a live Azure environment.
02 The 8 reasons people fail
Studying at AZ-900 depth for an Associate exam
The mistake: revising the same fundamentals you used for AZ-900 — what a VM is, what a storage account does — and expecting that to carry an Associate exam.
Why it happens: AZ-900 felt manageable, so people assume AZ-104 is "more of the same". It is not — it jumps from what services exist to how you configure and operate them.
The fix: study at task level. For every service, learn how to deploy, secure, scale and troubleshoot it — not what it is. If your notes read like a glossary, you are still preparing for AZ-900.
Almost no hands-on time in the Azure portal
The mistake: learning Azure entirely from videos and notes, never creating a VNet, a storage account or a VM yourself.
Why it happens: fear of being charged, or the belief that the exam is "theory only". It is not — some sittings include live lab tasks, and almost every question assumes you have seen the blades.
The fix: use a free Azure account and actually build things — peer two VNets, attach an NSG, configure a storage lifecycle rule, assign an RBAC role. The portal muscle memory is what the case study and labs reward.
Treating virtual networking as a minor topic
The mistake: skimming VNet peering, NSGs, routing, VPN/ExpressRoute and load balancers because networking "isn't your thing".
Why it happens: networking is the most conceptually demanding domain, so people defer it — not realising it is one of the most heavily weighted areas on the exam.
The fix: give networking the most study time, not the least. Know NSG rule precedence, the difference between system and user-defined routes, when to peer vs. use a gateway, and Basic vs. Standard Load Balancer cold.
PowerShell, Azure CLI and ARM/Bicep blind spots
The mistake: learning every task only through the portal and ignoring the command-line and template equivalents.
Why it happens: the portal is visual and forgiving, so it is where people stop. The exam, however, freely tests az CLI flags, PowerShell cmdlet names and ARM/Bicep template syntax.
The fix: learn the three ways to do each common task — portal, CLI/PowerShell, and template. Recognise the verbs and parameters; you do not need to write scripts from scratch, but you must read them.
Confusing services and settings with overlapping roles
The mistake: not being able to separate look-alikes — NSG vs. Azure Firewall, RBAC role vs. Azure Policy, Recovery Services vault vs. Backup vault, LRS vs. ZRS vs. GRS.
Why it happens: surface study makes them all sound interchangeable; the distinction only matters when a scenario forces a single correct choice.
The fix: build a personal "vs." sheet for every confusable pair and the one-line rule that separates them. AZ-104 lives in those gaps — especially storage redundancy and access control.
Misreading "least privilege" and "most cost-effective" qualifiers
The mistake: picking a technically correct answer that misses the qualifier — granting Owner when the question wanted least privilege, or choosing a premium tier when it asked for most cost-effective.
Why it happens: time pressure makes people skim. The qualifier is often the last clause and it changes which of two valid answers is correct.
The fix: read the final sentence first and lock onto the qualifier. For RBAC, default to the most scoped built-in role that does the job; for cost, the lowest tier that meets the stated requirement.
Walking into the case study unprepared for its rules
The mistake: hitting the case-study section cold, not realising you cannot return to it once you move past it.
Why it happens: standard questions let you flag and revisit, so people assume the whole exam works that way. The case study does not — and it bundles several questions on one scenario.
The fix: practise full case studies beforehand. Read all the tabs (requirements, existing environment, technical constraints) before answering, and finish every case-study question before you advance.
Booking the exam out of impatience, not readiness
The mistake: scheduling because the study deadline arrived, not because practice scores and hands-on confidence said "ready".
Why it happens: a booked date forces discipline — but it also forces a sitting before the evidence supports it, and a fail costs a 24-hour wait plus another full fee.
The fix: let the readiness signal set the date. A repeatable 80–85%+ on fresh full-length exams and real portal fluency first; the calendar second.
03 Where the marks live, and how to study for them
AZ-104 is built from five skill domains. Microsoft publishes the weighting ranges on the official skills-measured guide — and the gap between candidates who pass and fail is usually where they spent their hours, not how many.
| Skill domain | Weight | How people lose marks here |
|---|---|---|
| Manage identities & governance | 20–25% | Microsoft Entra ID basics, RBAC scope, Azure Policy vs. roles, subscriptions & management groups |
| Deploy & manage compute | 20–25% | VM sizing & availability, scale sets, App Service plans, containers & ACI/AKS basics |
| Implement & manage storage | 15–20% | Redundancy (LRS/ZRS/GRS), access keys vs. SAS vs. Entra, Azure Files vs. Blob, lifecycle rules |
| Implement & manage virtual networking | 15–20% | VNet peering, NSG rules & precedence, routing, VPN/ExpressRoute, load balancing — the classic graveyard |
| Monitor & maintain Azure resources | 10–15% | Azure Monitor, Log Analytics/KQL basics, alerts, and Backup & Recovery Services vaults |
Notice that compute, identity and networking together make up the bulk of the exam. If you are comfortable spinning up a VM but vague on NSG precedence and RBAC scope, you are over-invested in the easy half.
Habits that backfire vs. habits that work
| What fails | What works instead |
|---|---|
| Re-watching videos and re-reading notes | Active recall — answer questions first, then look up what you missed |
| Reading docs about how to peer a VNet | Actually peering one in a free account and breaking it on purpose |
| Learning each task only in the portal | Portal + CLI + PowerShell for the common tasks, so you recognise all three |
| Studying evenly across all domains | Weighting by the blueprint — most time on identity, compute and networking |
| Tracking hours studied | Tracking practice-exam % by domain and attacking the weakest |
04 Exam-day mistakes that cost passes
Plenty of well-prepared people lose AZ-104 in the room, not in the books.
05 Are you actually ready? A pre-exam check
If you cannot honestly tick every box below, you are in the band where people fail. Fix the gaps before you book.
- Repeatable 80–85%+ on at least three fresh full-length, timed practice exams — not the same set scored again.
- You have built it yourself: a peered VNet with an NSG, a storage account with a lifecycle rule, a VM scale set, and an RBAC role assignment.
- You can explain NSG vs. Azure Firewall, RBAC vs. Azure Policy, and LRS vs. ZRS vs. GRS in one line each.
- You recognise the same task done three ways: portal, Azure CLI and PowerShell (and can read a basic ARM/Bicep snippet).
- You instinctively read the qualifier — least privilege, most cost-effective, least admin effort — before choosing.
- You have done a practice case study and know you cannot return to that section once you leave it.
- Your weakest domain — usually virtual networking — is still above 75%, so nothing single-handedly drags you under.
06 FAQ
What is the AZ-104 pass rate?
Microsoft does not publish official pass rates, so any number is an estimate. Community reports consistently put AZ-104 among the tougher Associate exams, with a meaningful share of candidates failing their first attempt. The exam is scored 1–1000 and you need 700 to pass.
Why do so many people fail AZ-104?
The biggest reason is studying at AZ-900 depth for an Associate exam without enough hands-on time in the Azure portal. AZ-104 is scenario- and task-based: virtual networking (VNet peering, NSGs, routing) is heavily weighted and trips up candidates who only memorised definitions. Many also lose marks on PowerShell, CLI and ARM/Bicep gaps, and on misread qualifiers such as "least privilege" or "most cost-effective".
How long do I have to wait to retake AZ-104?
After your first failure you can retake AZ-104 after 24 hours. Every subsequent attempt requires a 14-day wait, and you may take the exam no more than five times in a 12-month period. You pay the full fee (around USD 165, varies by region) on each attempt, so it pays to be genuinely ready before booking.
What practice-test score means I'm ready for AZ-104?
Aim for a repeatable 80–85%+ across multiple full-length, timed practice exams you have not seen before, with no single skill domain dragging you down. Because AZ-104 includes a case study you cannot revisit and may include hands-on lab tasks, you also want genuine portal, CLI and PowerShell fluency — not just a good multiple-choice score.
