Microsoft Power Platform Fundamentals Exam Prep

Practice progress

Practise PL-900 product and concept decisions

Unofficial, independently authored practice for Microsoft Power Platform Fundamentals. No affiliation or endorsement is implied.

Target: English PL-900, skills measured 2026-07-24; sources verified 22 September 2026. 60 original single-choice scenarios cover all 37 listed objective bullets at least once. This curated bank is not exhaustive preparation.

The official English guide directly confirms July 24, 2026. Its five ranges are 5–10%, then 20–25% for each remaining area.

Practice uses 8%, 23%, 23%, 23%, 23%: authored integer weights within the published ranges, not exact Microsoft question-distribution claims. Whole-question counts use largest remainder with ties in displayed domain order.

These untimed rounds do not reproduce official exam difficulty or the full exam experience. Microsoft's proctored assessment may include interactive components; this bank uses one correct answer per question. No practice score predicts passing.

Official PL-900 details Official July study guide

Skills-measured date:

Only 2026-07-24 is available. An unavailable or stale edition blocks the round and saved-attempt restoration.

Planned exam date (optional):

Use YYYY-MM-DD, or leave blank for today. Dates from 2026-10-22 require source re-verification under this practice bank’s 30-day review policy. Practice admission uses UTC calendar days.

Practice mode:

Questions follow the disclosed practice weights within the official ranges, rounded by largest remainder. Short rounds may omit a domain. The selected practice level limits the available counts.

Domain:

Practice difficulty:

Foundational concepts practise recognition and recall; applied reasoning uses scenarios. These authored levels are not calibrated to official exam difficulty.

Feedback:

Question count:

Replay seed:

Keep the seed to replay the same questions and option order.

Answers and the replay seed are saved in the page URL. Sharing that URL shares the attempt. No account, exam credentials or certification details are needed.

{{ progressPercent }}%

{{ answeredCount }} of {{ questionCount }} answered

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

Study: {{ studySource(currentQuestion.id).title }}

Answer recorded. Corrections appear after the round.

Review this practice round

Readiness review

Unofficial original PL-900 practice, skills measured 2026-07-24. This score describes only the authored questions in this round; it is not an official scaled score or a passing prediction.

This untimed single-choice practice does not reproduce the full official exam experience. Microsoft's proctored assessment may include interactive components.

Use the documentation-linked corrections to choose your next study topics. Unsampled objectives remain untested. Repeat the missed items, then use hands-on exercises and Microsoft's official preparation resources.

Missed answer review

{{ item.prompt }}

Your answer: {{ item.selected_answer_label }}

Correct answer: {{ item.correct_answer_label }}

{{ item.explanation }}

Study: {{ studySource(item.question_id).title }}

Domain performance

Unanswered or unsampled domains are not evidence of weakness.

{{ row.label }} · {{ row.weight_percent }}% practice weight: {{ row.question_count }} sampled; accuracy {{ row.accuracy_percent === null ? 'not sampled' : row.accuracy_percent.toFixed(1) + '%' }}; weighted contribution {{ row.contribution_points === null ? 'unavailable' : row.contribution_points.toFixed(1) + ' percentage points' }}.

Not tested this round: {{ untestedDomains.join("; ") }}.

Loading domain-performance chart…

Answer ledger

{{ ledgerContextSummary }}
Question and domain	Answers	Explanation and study source	Copy
{{ row.question_id }} · {{ row.category_label }} · {{ itemDetails(row.question_id).objective }} · {{ difficultyLabel(row.question_id) }}{{ row.prompt }}	Your answer: {{ row.selected_answer_label }} Correct answer: {{ row.correct_answer_label }} {{ row.outcome_label }}	{{ row.explanation }} {{ studySource(row.question_id).title }}

A leave-request process needs more than a form. Employees submit dates, a manager approves or rejects the request, and the organization keeps a reliable record of the outcome. Microsoft Power Platform combines applications, business data, automation, websites and conversational agents so those responsibilities can work together without being mistaken for the same capability.

Much of this work uses low-code development, where makers configure screens, data and rules rather than write every part from scratch. The appropriate starting point depends on who will use the solution and what they need to do.

Power Platform approaches for different user experiences
Requirement	Typical approach	Important distinction
A tailored phone interface for entering information	Power Apps canvas app	The maker controls the screen layout and interactions.
Consistent case management over related records	Power Apps model-driven app	The experience is organized around Dataverse tables, forms and views.
A business website for external applicants	Power Pages	A website provides access to selected business data under configured permissions.
A conversation that answers questions and requests actions	Copilot Studio agent	Knowledge supplies information; configured tools perform operations.

In the leave example, Microsoft Dataverse can hold employees and requests in related tables. Rows represent individual records, while columns hold values such as dates or approval status. A relationship connects each request to its employee instead of relying on repeated names. Dataverse also provides security and reusable business logic, so it is more than a place to store a spreadsheet-shaped list. A form presents one record for viewing or editing; a view presents a selected list of records.

Power Automate handles the work that follows a submission. A trigger starts a flow and actions retrieve information, request approval or send a notification. Automated cloud flows respond to events, instant cloud flows start on demand, and scheduled cloud flows run on a recurrence. Connectors expose supported service operations. Desktop flows address a different need by interacting with application interfaces, including legacy software without a suitable supported service operation.

An agent that explains the leave policy has not necessarily gained the ability to submit a request. Informational answers use knowledge sources; a submission needs a tool with the required inputs and permissions. Topics organize controlled conversation paths, such as asking for dates and handling an exception. This distinction prevents a convincing conversational response from being mistaken for a completed transaction.

Access control remains necessary whichever interface is chosen. Sharing an app or hiding a Delete button does not establish the underlying data permissions. Dataverse security roles control permitted operations, and privileges from multiple assigned roles are cumulative. Data policies address another concern by restricting supported connector use and combinations. A working form therefore does not, by itself, prove that records are protected or that data can be sent to another service.

Artificial intelligence (AI) can help draft data models, apps and flows, but generated work still needs review and testing. Plan designer helps turn requirements into a proposed solution, while code apps offer a code-first approach using managed Power Platform capabilities. PL-900, Microsoft Power Platform Fundamentals, introduces these product roles and solution choices. Recognizing them is a useful foundation for hands-on work, not evidence that a production application is secure, accessible or ready to deploy.

How to Use This Tool:

This is unofficial, independently authored, untimed practice with no Microsoft affiliation or endorsement. The initial setup selects 10 questions from All skill areas, Mixed practice and feedback After each answer.

Keep Skills-measured date at 2026-07-24. Enter your Planned exam date in YYYY-MM-DD format, or leave it blank to use today in Coordinated Universal Time (UTC). Planned dates from 2026-10-22 require source re-verification, and practice also pauses when the current UTC date reaches that boundary.
For an invalid or conflicting saved link, Use verified July target and reset attempt clears the attempt and restores the default setup. Saved answers are not migrated, and resetting does not extend the source-review deadline.
Choose Focused practice for a selected domain or an unweighted mix of all skill areas. Choose Practice-weighted across skill areas for the disclosed domain proportions. Set Practice difficulty to mixed, foundational or applied questions. These are authored study levels, not calibrated official exam difficulty.
Select a Question count from the available menu. Focused all-area practice can include all 60 questions, or all 30 at one difficulty. Weighted practice permits up to 50 mixed questions or 25 at one difficulty; individual-domain choices have smaller limits.
Choose After each answer to read corrections as you go, or After the round to defer them. Under Advanced, retain the Replay seed to reproduce a selection with the same setup and unchanged bank. The seed must contain 1 to 128 characters and cannot be blank.
Select Start exam practice, then choose one of the four answers. Selecting an answer immediately records and locks it. Continue with Next question, or View results after the last answer. Reset and change setup discards the current attempt.
Read the corrections in Missed answer review and the complete Answer ledger. Use Retry missed for the questions answered incorrectly, Replay same round to repeat the selection, or New shuffled round to change the seed. A new shuffle can repeat familiar questions.

Interpreting Results:

During a round, Practice progress counts answers submitted, not answers correct. At completion, the headline changes to correct answers out of all questions. A 10/10 result produces Perfect round, but this means only that the ten selected authored answers were recognized.

The Readiness review is a study review, not a readiness rating. There is no practice pass threshold, official scaled score, certificate or prediction of passing. Microsoft's proctored assessment may include interactive components, which these untimed single-choice rounds do not reproduce. Use the missed explanations, objective labels and linked study sources to identify distinctions to revisit, then check that understanding through hands-on exercises and official preparation resources.

Domain performance uses answer counts for focused rounds and practice percentage points for weighted rounds. Weighted accuracy can differ from the ordinary fraction correct. Answered-domain coverage indicates which domain weights are represented by answers; even 100% coverage does not mean that every objective or subtopic was tested. An unsampled domain is untested, not a demonstrated weakness.

Compare rounds with similar scope, difficulty and feedback timing. A missed-only retry concentrates on known questions and is not comparable to a fresh weighted round. If the domain chart fails to load, Retry chart can attempt it again; the answer ledger and domain-performance CSV remain available.

Technical Details:

Ordinary question accuracy and domain-weighted accuracy measure different things. Ordinary accuracy gives each question the same influence. Domain weighting gives each skill area a fixed share, even when rounding assigns unequal numbers of questions to the areas.

Question accuracy and completion

Each correct answer earns one point and each incorrect answer earns zero. There is no partial credit or deduction. Let C be the number correct, A the number answered and N the total questions in the round.

\begin{array}{l} Points = C \\ Accuracy (%) = 100 \times \frac{C}{N} \\ Progress (%) = 100 \times \frac{A}{N} \end{array}

Possible points equal N, giving a point range of 0 to N and an accuracy range of 0% to 100%. Unanswered questions remain in the accuracy denominator without becoming incorrect answers. Completion requires A to equal N. After completion, C equal to N produces Perfect round; any smaller C produces Start with the missed answers. Neither message applies an exam pass standard.

English objective edition and practice scope

The target is the English PL-900 skills measured as of July 24, 2026. The 60-question bank has 30 foundational and 30 applied questions. The bank reviewed on September 22, 2026 declares selected mappings to 37 listed July objective bullets, but representation in the bank does not establish exhaustive coverage of each objective or its real-world variations.

The official guide supplies percentage ranges. The exact integer weights below are independently chosen practice weights within those ranges, not Microsoft's exact live question distribution.

PL-900 July 2026 domain ranges, authored practice weights and available question counts
Skill area	Official range	Practice weight	All questions	At each difficulty
Business value	5% to 10%	8%	8	4
Environment and Dataverse	20% to 25%	23%	14	7
Power Apps	20% to 25%	23%	12	6
Power Automate	20% to 25%	23%	12	6
Copilot Studio agents	20% to 25%	23%	14	7

The July edition includes AI-assisted app and flow creation, Plan designer, code apps and the Power Apps vibe experience. Agent topics include conversation paths, knowledge, action tools, publishing, monitoring and evaluations. Model Context Protocol (MCP) connects agents to exposed tools and resources; it does not remove authorization requirements. The documented Copilot Studio connection also requires generative orchestration to be enabled so the agent can select suitable capabilities for a request. Microsoft Agent 365 addresses organization-level agent visibility, governance and security rather than replacing answer-quality testing.

Whole-question allocation

Weighted rounds use largest-remainder allocation. For domain i, w_i is its practice weight expressed as 8 or 23, q_i is its ideal fractional allocation, and b_i is that allocation rounded down.

\begin{array}{l} q_{i} = \frac{N \times w_{i}}{100} \\ b_{i} = ⌊ q_{i} ⌋ \\ R = N - \sum_{i = 1}^{5} b_{i} \end{array}

R is the number of questions still to assign. One extra question goes to each of the R largest fractional remainders. Equal remainders are resolved in the table's displayed domain order. For 10 questions, the ideal counts are 0.8, 2.3, 2.3, 2.3 and 2.3. Rounding down leaves two questions, which go to Business value and Environment and Dataverse. The final allocation is therefore 1, 3, 2, 2 and 2.

Questions are selected without replacement within the eligible pools. Short rounds can omit a domain, and a difficulty filter reduces each pool. Focused rounds use uniform selection within their chosen pool without domain quotas. A missed-only retry uses its selected missed questions rather than reapplying the weighted proportions. The weighted count limits prevent a quota from demanding more eligible questions than a domain contains.

Domain contributions and weighted accuracy

Within domain i, C_i counts correct answers and A_i counts submitted answers. Domain accuracy D_i uses submitted answers as its denominator. Its contribution K_i is measured in percentage points.

\begin{array}{l} D_{i} = 100 \times \frac{C_{i}}{A_{i}} \\ K_{i} = w_{i} \times \frac{C_{i}}{A_{i}} \\ W = \sum_{i = 1}^{5} K_{i} \end{array}

A domain with no submitted answers has unavailable accuracy and contribution, not zero ability. The overall weighted accuracy W is available only for a completed round drawn using domain quotas, with at least one answer in every domain. It is unavailable for a missed-only retry. Missing domains are not discarded and the remaining weights are not rescaled to manufacture an overall result.

Answered-domain coverage H sums the practice weights for domains with at least one submitted answer.

H = \sum_{i : A_{i} > 0} w_{i}

Raw question accuracy and progress remain unrounded, and visible progress uses that raw percentage directly. Weighted accuracy, displayed domain accuracies and contributions are presented to one decimal place. The aggregate uses the underlying contributions rather than adding the rounded display values, so displayed contributions can differ slightly from the displayed total when summed.

Dated target rules

The local source-review policy starts from September 22, 2026 and requires another review after 30 days. October 22 is an exclusive practice boundary, not a Microsoft exam retirement or certification expiry date.

Accepted edition and date boundaries for PL-900 practice
Requirement	Accepted value	Outside this condition
Skills-measured edition	Exactly 2026-07-24	No earlier or alternative edition is substituted.
Current UTC date	On or after 2026-07-24 and before 2026-10-22	Starting, restoring or continuing practice is blocked.
Planned exam date	A real date in the same interval, or blank to use the current UTC date	Invalid dates, earlier dates and dates on or after 2026-10-22 are rejected.

Saved attempts must also match the supported bank and round format without conflicting setup information. Incompatible saved answers are rejected rather than silently transferred to different questions. Resetting the attempt or entering an earlier planned date cannot extend the boundary checked against the current UTC date. This target is English-specific; a translated exam's update schedule must not be assumed to match it.

Availability Notes:

Product availability and exam scope are separate. A feature can remain a named July objective while its authoring experience, regional access or preview guidance changes.

The Vibe questions preserve September 22, 2026 guidance identifying the experience as preview and not intended for production. That guidance recommended app building in Copilot Cowork and described a Copilot Studio rollout as forthcoming. The rollout statement describes that documentation snapshot, not current availability. The Vibe overview lists English-only use in United States, Australia, Asia and India environments, requires administrator enablement, and excludes default environments.
The referenced natural-language desktop-flow creation capability is separately a preview, not intended for production, with United States environment and work or school account requirements. Product questions and action summaries have a different availability scope. A generated recording placeholder still needs recorded or manually configured interface actions; leaving it unchanged does not automate the task.
A generated plan is not a finished application. The documented canvas-app path creates editable data-connected screens, while the model-driven path opens an app designer with tables added. Review the resulting data, behavior and permissions before use. Code-first development and agent integrations likewise remain subject to applicable access, administration and licensing requirements.

Privacy Notes:

Answer evaluation happens in the browser and does not require an account, exam credentials or certification details. However, the page URL stores the replay seed, attempt and responses without encryption. Sharing the URL shares the attempt; it can remain in browser history and reach the site when the page is reloaded.

If saving fails, the prior saved URL attempt remains unchanged. The current in-page responses may not be recoverable on reload, so do not rely on that address to preserve them. Downloaded answer ledgers also contain responses. Page resources, chart or document support, and opened study references can use the network, so browser scoring is not a guarantee of fully offline use or that attempt information never leaves the device. Avoid putting personal or confidential information in the replay seed.

Worked Examples:

The same fraction correct can carry different weights

Select Practice-weighted across skill areas, Mixed practice and five questions. The allocation contains one question from each domain. If only the Business value answer is incorrect, the headline is 4/5, ordinary accuracy is 80%, and Practice-weighted accuracy is 92.0% because the four correct domains each contribute 23 percentage points. If instead only the Copilot Studio answer is incorrect, the headline remains 4/5 but weighted accuracy is 77.0%, comprising 8 plus three contributions of 23. The difference reflects the authored weights, not a different official passing probability; review the missed concept in either case.

A perfect one-question round is still a narrow sample

Select the weighted mode with one question. The largest-remainder tie rule assigns it to Environment and Dataverse. A correct answer produces 1/1 and Perfect round, with domain accuracy of 100.0%, a contribution of 23.0 percentage points and answered-domain coverage of 23%. Overall weighted accuracy remains unavailable because the other four domains have no answers. A longer round can sample the missing areas; repeating the single question cannot establish understanding of them.

References:

Study guide for Exam PL-900: Microsoft Power Platform Fundamentals, Microsoft Learn.
Microsoft Certified: Power Platform Fundamentals, Microsoft Learn.
What are model-driven apps in Power Apps?, Microsoft Learn.
What is Microsoft Dataverse?, Microsoft Learn.
What is Power Pages?, Microsoft Learn.
Overview of cloud flows, Microsoft Learn.
Connectors overview, Microsoft Learn.
Security roles and privileges for Dataverse, Microsoft Learn.
Data policies, Microsoft Learn.
Copilot Studio overview, Microsoft Learn.
Extend your agent with Model Context Protocol, Microsoft Learn.
Overview of Microsoft Agent 365, Microsoft Learn.
Overview of plans, Microsoft Learn.
Build a solution from a plan, Microsoft Learn.
Power Apps code apps overview, Microsoft Learn.
Overview of the new Power Apps vibe experience (preview), Microsoft Learn.
Get started with Copilot in Power Automate for desktop, Microsoft Learn.