Counts show all available guides. Filters combine to narrow your results.
Create work & incomePlaybook
Turn your expertise into a practical lesson
Teach one observable skill with a worked example, a real practice task and an honest scoring rubric.
Illustrative resourceNot tool-testedExperienced practitioners making their first workshop or short course
What you’re working towards
A complete beginner lesson on giving useful design feedback, including instruction, a worked example, a submission task, a model answer and an 8-point rubric.
Start with the synthetic worked example or other non-sensitive inputs. Run one prompt at a time in your chosen AI tool. Replace any bracketed placeholders with the requested material, check each reply, and supply the source or earlier outputs that step needs. Paste them again if the tool lacks that context.
Set up the worked example
Open
Open a new conversation in a text AI assistant and a document editor for reviewing the lesson. This sample teaches written design feedback from two supplied screen descriptions. You do not need design software, screenshots, a course platform or a learner account.
Add the sample
In step 1, replace [paste notes] with the complete ‘The starting point’ (Synthetic input), from ‘SYNTHETIC EXPERT NOTES’ through ‘It is not an accredited course.’ Include both the workshop-page example and the separate booking-form exercise. Send one step prompt at a time in the same chat, checking the reply before continuing. Later prompts refer to those briefs and the lesson already in the chat. ‘The intended result’ and ‘Download the complete practical lesson’ are authored examples to compare afterward; do not paste their model answer as evidence that a learner completed the exercise.
Start with the synthetic example, then adapt each prompt to your task.
01
Choose one performance outcome
A topic is too broad to assess. Define the thing a beginner should be able to produce or do after this one lesson.
Your starting prompt
Using the supplied expert notes, define one observable learner outcome, prerequisites, a practice artifact and a completion criterion. Keep the scope to writing one actionable design-feedback note. Do not turn this into a general design course, claim accreditation or invent evidence that the method works.
EXPERT NOTES:
[paste notes]
Expect
One narrow outcome: a beginner writes an actionable design-feedback note. The reply should name starting knowledge, the piece of work to submit and a practice completion rule. It should not expand into a full design course.
Check
The method has four parts: task, observation, possible effect and request. An observation comes from the supplied screen description; a possible effect is uncertain. No research result, accreditation or proven learning benefit was supplied.
If it differs
If the outcome is vague, such as ‘understand design’, ask for the observable feedback note described in the brief. If extra prerequisites or proven effectiveness are claimed, ask which source supports them and remove unsupported claims.
02
Teach the reasoning with a worked example
Explain the decision behind each step, then show the full result. Keep claims grounded in the provided screen description.
Your starting prompt
Write a beginner-friendly lesson explaining task, observation, possible effect and request. Use the workshop-time brief as the worked example. Distinguish observable placement from unmeasured readability or user behavior. Include a complete example note and explain why it follows the method.
Expect
A short lesson explaining the four parts, followed by a complete workshop-time example and why it follows the method. The example is based on a description, not an inspected screen.
Check
The workshop start time is described as 12px light-gray text after six paragraphs. The reply can express concern about finding it, but cannot claim measured contrast failure, lost sales or actual visitor behavior. The request should have one clear focus: finding the time.
If it differs
If the reply says it inspected an image or proved a usability problem, ask it to return to the written description. Restore ‘may’ or another clear uncertainty label for an untested effect, and remove invented measurements.
03
Create a transfer exercise
Use a different fixture so the learner must apply the method rather than copy the example. Supply all facts needed to complete it.
Your starting prompt
Create an exercise from the booking-form brief. Ask for a 60–100-word rewrite of the vague feedback, with one priority and an explicit uncertainty statement. Include the full task instructions and a model submission. Do not add analytics, screen elements or brand rules not in the supplied brief.
Expect
A different practice task using the booking-form brief, with full submission instructions and a model answer of 60–100 words. The learner must rewrite the vague feedback with one priority and an explicit uncertainty statement.
Check
Check that date selection comes before contact entry in the intended task, while the described form shows ‘Your details’ first and puts the date selector after name and email. The model should use those observations and not assert proven abandonment or invent brand rules. Count the model's words in your editor.
If it differs
If the exercise repeats the workshop example, ask it to use the supplied booking-form description instead. If the model is outside 60–100 words or includes unsupported facts, ask for a corrected model that preserves the four-part structure.
04
Build a rubric that matches the task
Score only things taught and observable in the submission. Mark the pass rule as a practice choice, not a validated credential.
Your starting prompt
Make an 8-point analytic rubric with four 0–2 categories: task alignment, evidence, uncertainty and actionability. Define each score concretely. Set a practice completion criterion of 6/8 with no invented facts. Score the model submission and show a source-based reason for each score. If the model fails, revise it.
Expect
Four rubric categories worth 0–2 points each: task alignment, evidence, uncertainty and actionability. A rubric is the set of rules used to review the submitted note. The reply should explain each score and score its model with reasons.
Check
Check the arithmetic: four categories × 2 = 8 possible points; the practice threshold is at least 6/8. Any invented fact still requires revision, even with a high total. Scores must assess what the lesson taught, not confidence, style preference or qualifications.
If it differs
If the points do not total 8, ask for four 0–2 categories. If an unsupported assertion passes, quote it and ask the AI to apply the no-invented-facts rule, revise the model and show the new category reasons.
05
Check teachability before packaging
Read it as a novice and as a subject expert. Only learner trials can reveal whether the sequence actually teaches the skill.
Your starting prompt
Audit the lesson, worked example, exercise and rubric as one package. Check that all assessed skills were taught, the exercise supplies every required fact, the model obeys the word range and the rubric can identify an unsupported claim. Return the complete clean lesson plus a short plan for one consenting learner trial. Label all unexecuted teaching and tool tests.
Expect
One complete lesson containing instruction, worked example, distinct exercise, model answer and rubric, plus a separate plan for a consenting learner trial. You may copy the checked package into your document editor.
Check
Check the model's word count again and confirm the exercise supplies all facts the rubric demands. A proposed trial is not a completed trial, and AI review is not evidence that beginners learned the skill. The sample package remains untested teaching material.
If it differs
If the audit invents learner feedback or claims tests passed, ask it to mark those tests unrun. If a criterion needs facts the exercise never supplied, ask for the inconsistency to be identified and the exercise or criterion corrected within the source method before anyone teaches it.
Optional: improve the result
Improve it using a real learner trial
With permission, watch one learner attempt the task and retain only the information needed to revise the lesson. A small trial can identify friction, not prove effectiveness.
Advanced prompt
Compare this lesson with anonymized observations from a consenting learner trial. Separate unclear instructions, missing prerequisite knowledge and rubric disagreements. Propose at most three specific revisions, each linked to an observed difficulty. Keep version history and preserve any unresolved expert disagreement. Do not infer broad learning outcomes from a small trial.
LESSON:
[paste]
ANONYMIZED OBSERVATIONS:
[paste]
Optional: use your own input+
Make a lesson from your own safe working method
Start a new text chat with a non-sensitive method you can explain and material you have permission to use. Choose a narrow, low-risk skill that can be practised in writing. Supply two different text briefs: one for the worked example and one for practice. Fill the placeholders below. After checking the outcome, send these messages one at a time: “Write the beginner lesson for my agreed outcome. Explain my supplied method, then give a complete worked example using my example brief. Mark unsupported effects and missing evidence.” “Create a separate practice exercise using my practice brief. Include all needed source facts, submission instructions and a model answer within my requested length. Use only taught steps.” “Build a practice rubric using my agreed criteria and score limits. Define each score, calculate the total and show source-based reasons for the model's score. Revise unsupported model claims regardless of the total.” “Audit the whole package: taught method, supplied facts, model length, criteria and point arithmetic. Return the complete clean lesson and a proposed consenting-learner trial, with unrun checks labeled.” Review the subject matter yourself before teaching. The original advanced extension stays sample-scoped unless you replace its inputs with this lesson and actual permitted, anonymized trial observations; never invent a trial result.
Your own-input prompt
Help me create one practical beginner lesson from my supplied working method. First propose one observable learner outcome, starting knowledge and a concrete practice submission. Use my requested length and review rules, or ask about missing details. Do not write the full lesson yet. Keep the scope to a safe written practice task. In later messages we will write the instruction and worked example, create a distinct exercise and model, build a matching rubric and audit the package. Do not invent expertise, research support, credentials, measured effects or learner trials. Label any completion threshold as a practice choice, not a validated qualification. Return text only; do not publish, enroll anyone or contact learners.
MY METHOD AND ITS LIMITS
[Paste the permitted process steps, explanations, exceptions and limits on what is known.]
LEARNER AND ONE TARGET SKILL
[Describe the beginner's starting knowledge and the single piece of written work they should produce.]
WORKED-EXAMPLE BRIEF
[Paste the first non-sensitive scenario with all facts needed to demonstrate the method.]
SEPARATE PRACTICE BRIEF
[Paste a different scenario with all facts needed to apply the same method. Mark unknowns.]
SUBMISSION AND REVIEW RULES
[Give the submission word range or other length limit, the criteria to assess, score limits and any completion threshold; or ask the AI to propose these for your review. Require unsupported facts to be revised whatever the score.]
Check your own result
The outcome, examples, requested length and criteria fit your own method; the sample's screen details and scoring numbers are not carried over automatically.
The exercise differs from the worked example but tests only taught steps and supplies every required fact.
Check the model's actual length, the score arithmetic and each score's reason. Unsupported claims require revision regardless of total.
The lesson distinguishes known facts from uncertain effects and does not claim accreditation, effectiveness or a completed learner trial.
A knowledgeable person reviews the package before teaching; any later trial evidence is genuine, permitted and clearly limited.
Show, then try
See the shape of the result.
This input is synthetic. The output is a written illustration of the intended result, not a verified AI-tool run.
01 / The starting point
Synthetic input
SYNTHETIC EXPERT NOTES
Expert role: a fictional design reviewer teaching new project coordinators. These are a supplied working method, not research findings or a certification standard.
Method: identify the screen's intended task; describe one visible observation without guessing a cause; explain how it might affect that task; request one concrete revision or test. Label opinions as opinions. Keep one priority per feedback item. Do not say a layout caused lost sales unless there is evidence.
Audience: beginners giving their first structured design review.
Course outcome: write one evidence-bounded feedback note a designer can act on.
Worked-example brief: a fictional workshop page should help a first-time visitor find the start time. The supplied screen description says the time appears in 12px light-gray text after six paragraphs. No user testing or contrast measurement is available.
Exercise brief: a fictional desktop booking form should help visitors choose a date before entering contact details. Screen description: the first visible heading is “Your details”; the date selector appears after the name and email fields. A reviewer wrote: “This looks dated. Make it pop. People will definitely abandon this.” No usability test, analytics or brand rule was supplied.
Length target: one short self-paced lesson with a worked example, a practice submission and a rubric. It is not an accredited course.
Copies text only. For image or file tasks, use the source files named in the steps too.
MICRO-COURSE: FEEDBACK A DESIGNER CAN USE
Lesson 1 of a proposed practical course. Synthetic teaching material; the instructional workflow has not been trialed with learners.
LEARNER OUTCOME
Given a screen brief and a description, write one feedback note that states the task, names observable evidence, identifies a possible consequence without overstating it, and requests one useful next step.
Materials: the two supplied briefs, paper or a document editor. No design software or account is needed.
Suggested sequence: read the lesson, inspect the worked example, submit the exercise, then use the rubric. Duration is learner-dependent.
LESSON
A designer needs to know what problem your feedback is trying to solve. “Make it pop” leaves the problem and the change undefined. Start with the intended task: what should a visitor be able to do here?
Next, separate observation from interpretation. “The time appears after six paragraphs” is supported by the supplied description. “Nobody sees the time” is a claim about visitors and needs evidence. It may be a reasonable concern, but label it as a possibility.
Link that observation to the task. The point is not that every important detail must be first; the point is whether this placement supports the particular job in the brief. You can recommend a revision, or ask for a small test if the best answer is uncertain. Make the request specific enough that someone can respond to it.
Use this four-part structure:
1. Task: what the screen should help someone do.
2. Observation: what you can actually point to.
3. Possible effect: why it might matter; label uncertainty.
4. Request: one revision or one test.
Avoid diagnosing the designer's intent, inventing analytics or rewriting the entire design in one note. If your point is taste rather than task performance, say so and lower its priority unless a brand requirement supports it.
WORKED EXAMPLE
Task: help a first-time visitor find the workshop start time.
Observation: the supplied description places the time in 12px light-gray text after six paragraphs.
Possible effect: this placement may make the time harder to find during a quick scan. Its actual readability and findability have not been measured.
Request: try placing the time beside the workshop title, then ask a first-time reader to locate it without prompting.
Why this works: the recommendation names a specific change and a check, while keeping the possible effect separate from an observed result. It does not claim a measured contrast failure.
YOUR EXERCISE
Rewrite this feedback about the supplied booking form: “This looks dated. Make it pop. People will definitely abandon this.”
Submit 60–100 words using the four-part structure. Use only the exercise brief. Prioritize date selection. Include one revision or test, and explicitly state what has not been measured. Do not invent a new visual brand or claim abandonment data.
MODEL SUBMISSION
“The brief asks visitors to choose a date before entering contact details. In the supplied screen description, the first heading is ‘Your details’, while the date selector appears after the name and email fields. That order may lead people to enter details before checking a date, although we have no usability or abandonment evidence. Please try moving the date selector above the contact fields, then ask a first-time visitor to choose a date without prompting. My impression that it looks dated is a separate style preference.”
RUBRIC / 8 POINTS
Task alignment, 0–2: 2 names date selection as the first task; 1 vaguely mentions booking; 0 omits or contradicts the brief.
Evidence, 0–2: 2 accurately cites both the first heading and field order; 1 cites one accurate detail; 0 invents a screen fact.
Uncertainty, 0–2: 2 marks the possible effect as a hypothesis and states missing testing/data; 1 does only one; 0 asserts proven abandonment.
Actionability, 0–2: 2 requests one concrete change or test tied to date selection; 1 asks for a vague improvement; 0 has no usable request.
Practice completion criterion: at least 6/8, with no invented fact or measured result. If an unsupported assertion appears, revise it even if the point total is high. This is a practice rubric, not a validated assessment.
SELF-CHECK
Underline observations. Circle hypotheses. Box the requested next step. If a phrase fits none of those roles and does not name the task, ask whether it is helping the reviewer act.
TRANSFER TASK
Apply the same structure to one non-sensitive screen description from your own work. Include the actual goal, then ask another person whether they can identify the single requested next step. Do not publish the review or share private screens without permission.
These checks use the worked example and its particular facts. With your own material, check your actual names, numbers, sources and rules; do not reuse the sample’s values as the answer.
Unsaved checks stay in this tab and reset when you reload. If you choose Save guide, its current example checkmarks are kept in this browser when storage is available.
!
A few things to watch for.
Turning a practitioner’s process into an enormous syllabus before testing one useful lesson.
Assessing unstated criteria, or allowing a plausible-sounding invented result to earn a pass.