Exam preparation

{{ summaryValue }}

{{ summaryLine }}

Practise Google Cloud business decisions

Unofficial, independently authored practice for the Cloud Digital Leader standard exam. Guide launched 12 August 2026; verified 11 September 2026. No affiliation with or endorsement by Google Cloud.

48 curated questions, eight per domain. Uniform selection, one answer per question, no timer. This is not a full exam simulation or a pass prediction. The standard exam has 50–60 multiple-choice and multiple-select questions in 90 minutes; no prerequisites. Recommended: experience collaborating with technical professionals.

Official domain weights: transformation 18%, data 18%, AI 18%, infrastructure 18%, security 18%, operations 10%. These are scope context, not practice sampling quotas. Read the current exam guide.

Optional. Leave blank for the standard seed, or enter up to 128 characters to reproduce a round.

{{ seedNotice }}

{{ examSetupValidation }}

{{ answeredCount }} of {{ questionCount }} answered · {{ progressLabel }}%

{{ progressStatusLine }}

Question {{ examCursor + 1 }} of {{ questionCount }}

{{ currentQuestion.stem }}

Answer recorded. Corrections appear after completion.

Review this practice round

{{ resultLead }}

This score describes only these authored questions. It is not an official scaled score or a prediction of passing. Review the full current guide, including topics absent from this curated bank, and use official training for broader preparation.

{{ readinessReviewTitle }}

{{ readinessReviewLine }}

{{ card.label }}
{{ card.value }}
{{ card.note }}

Missed answer review

  1. {{ item.prompt }}

    Your answer: {{ item.selected_answer_label }}

    Correct answer: {{ item.correct_answer_label }}

    {{ item.explanation }}

    Source: {{ item.source.label }} · {{ item.source.version }} · {{ item.source.license }}

Domain performance

Which sampled domains need more study? Compare correct and missed answers in this round. Unsampled domains are omitted, not scored as zero.

Loading domain-performance chart…

{{ examChartStatus }}

{{ examChartExportStatus }}

Answer ledger

Question ID Prompt Topic Your answer Correct answer Outcome Explanation Source Copy
{{ row.question_id }} {{ row.prompt }} {{ row.category_label }} {{ row.selected_answer_label }} {{ row.correct_answer_label }} {{ row.outcome_label }} {{ row.explanation }} {{ row.source ? `${row.source.label} · ${row.source.version} · ${row.source.license}` : '—' }}

{{ examLedgerExportStatus }}

AWS Cloud Practitioner Exam Prep Practise AWS Cloud Practitioner CLF-C02 concepts with original questions, immediate explanations and focused reviews of missed answers.
Microsoft Azure Fundamentals Exam Prep Practise AZ-900 with 36 original questions and sourced explanations that help you review mistakes across the three exam domains.
Cloud Egress Cost Calculator Estimate cloud egress spend from measured traffic and a marginal rate card with verified reductions, peak-month stress testing, and budget variance.
Cloud Tag Coverage Analyzer Audit cloud inventory tags against required keys and allowed values, measure whole-resource coverage, and rank the gaps that need cleanup.

For a retailer moving its ordering system to the cloud, reliable checkout and faster delivery of new features are useful measures of success. The number of cloud products in the design is not. Cloud computing supplies processing, storage, networking and software on demand, but the business benefit depends on choosing services that meet the workload's requirements and managing the responsibilities that remain.

Cloud Digital Leader focuses on this connection between cloud capabilities and organizational goals. It covers foundational Google Cloud knowledge for people who discuss technology choices with business and technical colleagues. There are no prerequisites for the standard exam, although experience collaborating with technical professionals is recommended. The emphasis is on understanding use cases and tradeoffs rather than treating product names as answers by themselves.

A ticket seller with brief demand spikes benefits from elasticity, which means adding and releasing resources as demand changes. Metered consumption also changes spending from an upfront hardware purchase toward ongoing operating expense. This can reduce unused capacity, but it does not guarantee a lower total cost. Migration work, staffing, software, networking and continuing usage still belong in the comparison.

The service model determines how much of the computing environment the customer operates. Greater control can be necessary for an older application, while a more managed service can reduce routine administration.

Service model What the customer receives Responsibility that remains
Infrastructure as a service (IaaS) Compute, storage and networking resources, including virtual machines. Managing the guest operating system, applications, data and access.
Platform as a service (PaaS) A managed environment for developing and running applications. Managing application code, data and application configuration.
Software as a service (SaaS) A finished application operated by the provider. Managing users, account access and appropriate handling of business data.

Data requirements introduce another important distinction. Recording an individual sale is a transactional task, while analyzing years of sales is an analytical task. A data warehouse organizes data for analysis. A data lake can retain varied data in its original forms for later processing. Neither removes the need for data governance, which establishes ownership, shared definitions, quality expectations and access rules. Two departments using different definitions of an active customer can still disagree after moving their data to the cloud.

Artificial intelligence (AI) adds choices about what to predict, generate or automate. Machine learning (ML) finds patterns in data; supervised learning uses examples with known outcomes, such as customers who renewed or canceled. Generative AI produces content, while an agentic workflow coordinates steps and software tools toward a goal. These distinctions matter because transcribing a call, predicting cancellation and arranging a proposed customer resolution are different jobs. More data or more automation does not remove the need to evaluate results, protect access and consider effects on people.

A successful backup does not establish that a service can be restored when needed. Recovery exercises test that process against service requirements. Likewise, encryption of stored data does not fix overly broad permissions or insecure application behavior. Cloud knowledge includes recognizing what a service provides, what it does not provide and which responsibilities still need an owner.

The standard Cloud Digital Leader exam allows 90 minutes for 50 to 60 multiple-choice and multiple-select questions. Multiple-select questions can require more than one choice. Preparation should include the full relevant exam guide and the ability to explain unfamiliar scenarios, not just recall familiar answers. A practice score does not establish deployment, security or operational competence.

How to Use This Tool:

This is independently authored, unofficial practice with no affiliation with or endorsement by Google Cloud. The starting setup is a 12-question round from all six domains with feedback after each answer. Every round is untimed.

  1. Choose a Practice domain and Question count. The combined question pool offers 6, 12, 24 or 48 questions. Each individual domain contains eight questions and offers rounds of four or eight, with eight as its default.
  2. Select Feedback. Learn · feedback after each answer reveals the correction and explanation after your choice. Review · feedback after completion records your choice without revealing correctness until the round is complete.
  3. Enter an optional Replay seed under Advanced when you need a reproducible round. Leading and trailing spaces are ignored. Leave it blank for the standard seed, not a newly randomized seed. The limit is 128 characters; an overlong value is ignored with a notice and the standard seed is used.
  4. Select Start exam practice, then choose one of the four answers for each question. Choosing an answer submits and locks it immediately. Use Next question to continue and View results after the last answer. There is no answer-editing or question-skipping control.
    If progress is unavailable, use Discard attempt and change setup to start a clean round. This discards the current answers rather than converting an incomplete attempt into a score.
  5. Review Missed answer review and the Answer ledger. Use Retry missed for the questions answered incorrectly, Replay same seed to repeat the round or Try a new seed for another selection and ordering. Use Change setup to choose a different domain or round length.
    If only the Domain performance chart is unavailable, use Retry chart. The answer ledger remains available for reviewing the round.

Interpreting Results:

During a round, the headline counts questions answered. After completion, it counts answers correct. The progress bar measures completion, so reaching 100% means every question has an answer, not that every answer is right.

Missed answer review identifies the choice you made, the keyed correct answer and its explanation. The Answer ledger also includes questions you answered correctly. Review those explanations when a correct choice came from guessing or recognizing a familiar option rather than understanding the requirement.

The Domain performance chart compares correct and missed answer counts in the sampled domains. A longer bar can simply mean that more questions from that domain appeared. Unsampled domains are omitted, not assigned zero. A few correct answers in one domain do not establish knowledge of all its exam objectives.

Perfect round means no questions were missed in this attempt. There is no readiness band, official scaled score or passing prediction. Question difficulty is not calibrated to the official exam, and the practice does not reproduce its multiple-select format or time pressure. Use the missed explanations to guide study, then compare your preparation with the full current exam guide and official training, including topics outside this question bank.

Technical Details:

A single-choice practice score measures agreement with an authored answer key. Each scenario has one keyed answer and three question-specific alternatives. All questions carry the same point value, so the result is a raw count rather than a difficulty-adjusted or provider-scaled assessment.

Counting correct business decisions

A correct answer earns one point and an incorrect answer earns zero. There is no negative marking or partial credit. Let C be the number correct, A the number answered and N the total questions in the round. Possible points equal N, and scores range from zero to N points.

Score points = C Accuracy (%) = 100 × CN Progress (%) = 100 × AN

Both percentages use the full round length. An unanswered question remains in the denominator but does not count as an incorrect or missed answer. Counts are exact, and raw accuracy retains the unrounded value of 100 × C/N. Displayed progress rounds to the nearest whole percent, with a half percent rounded up. The final headline presents correct answers out of all round questions.

Completion requires A = N. A completed round with C < N shows Start with the missed answers; C = N gives Perfect round. No passing threshold is applied. Both feedback modes and missed-question retries use the same scoring rules.

Exam outline and curated coverage

The practice targets the English Cloud Digital Leader standard exam guide launched on 12 August 2026. Question references carry a verification date of 11 September 2026. These dates identify the intended scope; the questions do not automatically update when provider guidance or product details change.

The guide's approximate domain shares describe the official exam outline. The authored question totals below describe this practice bank and are not official sampling quotas.

Domain Approximate exam share Authored questions Examples of decisions covered
Digital transformation 18% 8 Elastic capacity, business outcomes, hybrid and multicloud choices, service responsibilities.
Data transformation 18% 8 Governance, analytical versus transactional services, retained objects and streaming data.
Artificial intelligence 18% 8 Supervised prediction, representative data, responsible AI and agent workflows.
Infrastructure and applications 18% 8 Migration approaches, virtual machines, managed containers and application interfaces.
Trust and security 18% 8 Permissions, security findings, model-content protection and compliance boundaries.
Cloud operations 10% 8 Budget notifications, cost ownership, interruption-tolerant compute and recovery.

Eight questions in a domain do not cover every subobjective. For example, the broader guide includes AlloyDB, Managed Service for Apache Spark and Google Threat Intelligence, none of which has a dedicated scored question in this bank. Recognition of a product among answer alternatives is also different from having practiced its own use case.

How business constraints change the answer

Compute choices depend on the control required. Compute Engine provides virtual machines (VMs) when an application needs direct administration of its guest operating system. Cloud Run supports managed execution of a containerized service without requiring the team to operate a Kubernetes cluster. Google Kubernetes Engine (GKE) supplies managed Kubernetes when its orchestration and application programming interfaces (APIs) are part of the requirement. A container requirement alone does not make those services interchangeable.

For data, the distinction is the work being performed. BigQuery supports large-scale analytical queries using Structured Query Language (SQL), while Cloud SQL can retain an existing MySQL application on a managed relational database. Choosing an analytics service simply because a question mentions SQL overlooks compatibility and transactional requirements. Similarly, retaining raw data does not by itself provide governed business metrics or the processing needed to make those metrics useful.

AI choices also have different scopes. Gemini Enterprise Agent Platform supports developing, deploying and governing agents and model-based solutions. It is not merely a place to run a container. An agent can coordinate actions, but each connected system still needs appropriate permissions. Model Armor screens prompts and responses for configured risks such as prompt injection and sensitive content; this does not replace authorization or make every generated answer trustworthy.

Access decisions distinguish authentication, which establishes identity, from authorization, which determines permitted actions. Identity and Access Management (IAM) controls permissions on resources. An analyst who needs to read one dataset does not need organization-wide administration. Restricting both the actions and their resource scope is the practical meaning of least privilege.

Operational choices depend on interruption tolerance and measurable service needs. Spot VMs offer lower-cost spare capacity that can be stopped or deleted when reclaimed, so restart and checkpoint behavior matter. A service-level indicator (SLI) is a measured property, such as the proportion of successful requests. A service-level objective (SLO) is the target for that measure. A service-level agreement (SLA) adds agreed consequences; an internal target does not automatically become such an agreement.

Selection, replay and score comparisons

Questions are selected uniformly without replacement from the chosen pool. No question repeats within one round, but there is no guarantee that a shorter mixed round includes every domain or the same number from each. The 48-question round includes all eight questions from each domain, which differs from the official outline's approximate shares. Neither selection nor scoring applies those official weights.

Reproducibility depends on keeping the domain, question count, feedback mode, replay seed and question scope unchanged, with the same question bank. Question order and the order of the four answer choices are seeded. A new seed can produce overlapping questions; it does not create new material. When all 48 questions are selected, another seed still uses those same 48 questions and can vary only their presentation order.

A missed-question retry uses only the questions answered incorrectly in the preceding round, even when that produces a one-question attempt. It is a new attempt rather than a correction to the earlier score. Results from different domain selections, round lengths or retries are not directly comparable, and immediate-feedback practice includes teaching that deferred-feedback practice withholds until completion.

Privacy Notes:

Answers are scored in your browser, but the active attempt is also encoded in the page address for restoration and sharing. That address carries the seed, attempt details and recorded responses without encryption. Anyone receiving it can potentially restore the attempt and inspect those responses.

Saved addresses can remain in browser history, and loading them can send their contents to the site. Avoid personal or confidential information in the replay seed and treat an attempt link as an answer record, not an anonymous link to an empty exercise. Exported answer ledgers also contain your responses. Returning to setup clears the active attempt from the current address, not copies already shared, downloaded or kept in history.

Page resources, chart or document features, and references you open can require network access. Browser scoring does not mean that the complete experience is guaranteed to work offline or that answer-bearing addresses never leave the device.

If Saved attempt unavailable appears, the saved round could not be restored, for example because its data is invalid or no longer compatible. Start a clean round from setup. If a warning says the attempt could not be saved to the address, your current answers remain on the page, but sharing or reloading will not restore them.

Worked Examples:

Finishing a round with deferred feedback

Choose a 12-question mixed round with Review · feedback after completion. After six answers, the progress display shows 6 of 12 answered and 50%. It does not reveal how many are correct. Suppose nine answers are correct when all 12 are finished. After View results, the headline is 9/12, equivalent to 75% accuracy, and three questions need review. Completion is 100% because all questions were answered; the ledger identifies which three choices to revisit.

Mistaking a budget notification for a spending limit

An eight-question Cloud operations round in Learn includes an alerts-only Cloud Billing budget with email thresholds and no automated response. Selecting the answer that promises all services will shut down at the threshold is incorrect. The explanation distinguishes a notification from an automatic control. If the other seven answers are correct, the final score is 7/8 with one missed question, and Retry missed creates a one-question round. The important qualifier is alerts-only: separately configured spend cap budgets, where supported, are a different case. Correcting this misconception on a retry does not change the earlier 7/8 result.

References: