Why start with the exam itself?
Before we touch Cloud Run or Pub/Sub, you need to know what you're preparing for. I've watched strong engineers fail certification exams. They knew the products. What they hadn't practised was reading a long scenario under time pressure, finding the one requirement that decides it, and picking the best answer when three of them would work.
This level covers three things: how the exam is set up, how its four sections are weighted, and a repeatable way to take questions apart.
Housekeeping. The Professional Cloud Developer exam guide belongs to Google. It isn't reproduced in this course. It's free on Google's certification page for the exam, so download it and keep it open while you study. This course is independent of Google and not endorsed by it, and no course can promise you a pass. What it can do is make your preparation efficient.
The exam at a glance
| Item | Detail |
|---|---|
| Length | 2 hours |
| Questions | 50–60 |
| Format | Multiple choice (one answer) and multiple select (several answers) |
| Fee | $200 (plus tax where applicable) |
| Languages | English and Japanese |
| Delivery | Online-proctored (remote) or onsite at a testing centre |
| Prerequisites | None. Google recommends 3+ years of industry experience, including 1+ year designing and managing solutions on Google Cloud |
Work out your time budget straight away: 120 minutes ÷ 60 questions = about 2 minutes per question. Long scenario questions will take longer, so answer the quick recognition questions fast to bank time for them.
Gotcha: multiple-select questions say how many answers to pick ("Choose two"). Read that instruction before you read the options.
The four sections and their weights
The guide describes a Professional Cloud Developer as someone who builds scalable, secure applications "by using Google-recommended tools and best practices". That includes generative AI APIs and AI-powered development tools such as coding assistants and automated debugging agents. The content is split into four sections:
- Designing highly scalable, secure, and reliable cloud-native applications (~32%)
- Building and testing applications (~23%)
- Configuring cloud-native applications for deployment (~24%)
- Integrating applications with Google Cloud services (~21%)
Section 1 is the biggest by a clear margin. Sections 2–4 are roughly equal. Split your study time the same way. Here's the arithmetic as a tiny script:
# Split a study budget in proportion to the exam guide's section weights
weights = {
"1 Design (scalable, secure, reliable)": 0.32,
"2 Build and test": 0.23,
"3 Configure for deployment": 0.24,
"4 Integrate with Google Cloud services": 0.21,
}
total_hours = 60 # your realistic budget
for section, w in weights.items():
print(f"{section:<42} {w*total_hours:5.1f} h")
1 Design (scalable, secure, reliable) 19.2 h
2 Build and test 13.8 h
3 Configure for deployment 14.4 h
4 Integrate with Google Cloud services 12.6 h
On a 60-question exam, ~32% is around 19 questions about design. Section 1 is also the broadest: platforms, APIs, security, storage and consistency. Give it the extra hours.
Three kinds of question, three ways to eliminate
Most questions fall into one of three types. Spotting the type tells you how to cut it down.
1. Recognition. "Which service stores and rotates application secrets?" This is a straight fact. Method: rule out options that are real products doing a different job. Here, Cloud Storage stores objects, not managed secrets.
2. Product contrast. Two or more options could plausibly work, and the question turns on one difference between them, for example strong versus eventual consistency, or zonal versus regional. Method: name the deciding difference, then keep only the option on the right side of it.
3. Scenario best answer. You get a paragraph about a company and its constraints, and often every option would technically work. Method: pick out the requirement words, rank them, and remove options that break the top-ranked one.
Requirement words: the signals in the scenario
These words decide most best-answer questions. Learn to treat each one as a filter:
| Requirement word | What it rules out |
|---|---|
| minimal operational overhead, fully managed | Anything you patch, scale or babysit yourself (self-managed VMs, hand-built clusters) |
| lowest latency | Cross-region round trips, synchronous chains of calls |
| cost-effective, minimise cost | Always-on capacity for spiky or idle workloads |
| least privilege | Broad roles (Owner/Editor), shared keys, a default identity used for everything |
| Google-recommended | Workarounds, even ones that function |
Worked example
A team's Cloud Run service needs a database password. The security team requires least privilege and regular rotation, and the team wants minimal operational overhead. What should they do?
- A. Put the password in an environment variable in the deployment YAML
- B. Store it in Secret Manager and grant the service's dedicated service account access to that one secret
- C. Store it in a Cloud Storage bucket readable by the project's Editor role
- D. Run a self-hosted secrets vault on Compute Engine
Working through it:
Type: scenario best answer
Signals: least privilege | rotation | minimal operational overhead
A -> fails rotation (secret baked into config, visible to anyone who reads it)
C -> fails least privilege (Editor is a broad role)
D -> works, but fails minimal operational overhead (you run the vault)
B -> meets all three -> ANSWER
D is the classic distractor because it works. The exam punishes "works" when a managed, Google-recommended option also satisfies every requirement.
Practical habits for exam day
- Read the last sentence first. It tells you what's being asked ("what should you do first?" is a different question from "what's the best long-term design?").
- Underline qualifiers: first, most, least, without changing code.
- Flag and move on if you're past about 3 minutes. Come back once you've banked time on the rest.
- Don't bring in requirements the question didn't state. If cost isn't mentioned, don't choose an option just because it's cheaper.
What the exam asks about this
The exam won't ask you about its own format. But every question uses the skills from this level. Expect long scenarios where two or three options work and the deciding factor is a requirement word: managed, least privilege, latency, cost. If you build the elimination habit now, every later level becomes easier to turn into points.
Key takeaways
- The exam is 2 hours long, has 50–60 multiple-choice and multiple-select questions, and costs $200. Allow about 2 minutes per question.
- Weights: Design ~32%, Build and test ~23%, Configure for deployment ~24%, Integrate ~21%. Plan study hours to match.
- The exam guide is Google's and is free on its certification page. This course is independent, not endorsed by Google, and cannot promise a pass.
- Recognition questions: rule out products that do a different job. Product-contrast questions: find the deciding difference. Scenario questions: rank the requirement words and filter on them.
- The option that "works" is often the distractor. Choose the managed, least-privilege, Google-recommended one that meets every stated requirement.