The Marketing Practice Lab

The principle

A useful audience begins with a situation that causes someone to seek help. Job titles alone cannot explain urgency. A recent trigger, the current workaround and the cost of leaving the problem unresolved give you something specific to investigate.

Work through it

1. Name one buyer and one recent trigger.
2. List the current workaround and its cost.
3. Write one uncertainty that could invalidate your audience choice.

Check your work

A reader can identify who needs help, when and why.

Watch for: Treating a large demographic as a specific customer.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Evidence has a source and a limit. A customer quote is an observation; your explanation of it is an interpretation. An AI-generated persona is a hypothesis until independent evidence supports it. Keeping these separate makes your next research question clearer.

Work through it

1. Collect three public or authorized observations.
2. Record the source, date and context of each.
3. Label each statement observed, inferred or still unknown.

Check your work

At least one assumption has a realistic way to be checked.

Watch for: Using generated customer quotes as research.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Questions about recent behavior are more useful than asking whether someone likes your idea. Ask what happened, what they tried, who decided and what made the task difficult. Do not pitch while collecting evidence: agreement can reflect politeness.

Work through it

1. Draft five questions about a recent relevant task.
2. Remove leading wording and hypothetical buying questions.
3. Rehearse the interview with a peer or label a solo rehearsal as simulated.

Check your work

The questions could uncover evidence against your idea.

Watch for: Counting a practice interview as market validation.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

An offer connects a customer problem with a deliverable and a boundary. State the input required, the result you can actually deliver, timing and exclusions. A promise about your work is different from a promise about sales or revenue that you cannot control.

Work through it

1. Write the offer in one sentence.
2. Add deliverables, required customer inputs and exclusions.
3. List the proof needed for each claim.

Check your work

A buyer could distinguish the deliverable from a hoped-for business outcome.

Watch for: Promising results that depend on factors outside delivery.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Contribution before acquisition equals revenue less variable delivery, fulfillment, payment and expected refund costs. The amount left must still cover acquisition work, overhead and profit. Use a stated time period; projected lifetime value is not cash already earned.

Work through it

1. Create a labeled fictional or real unit-economics table.
2. Subtract delivery costs and a chosen profit/overhead reserve.
3. Calculate an allowable acquisition cost and mark uncertain inputs.

Check your work

Every number has a source or an assumption label, and the arithmetic reconciles.

Watch for: Treating revenue or ROAS as profit.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Bring the week together into a one-page decision brief. The brief should make the weakest assumption visible. It is a working document for deciding what to learn next, not a polished sales story. Spend extra time checking whether the offer addresses the trigger you chose.

Work through it

1. Combine buyer, trigger, evidence, offer and economics.
2. Ask a reviewer to explain the offer back to you.
3. Revise one confusing claim and choose the next research task.

Check your work

The brief includes an evidence gap and a concrete next action.

Watch for: Perfecting presentation while leaving the central assumption untested.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A review is a decision about what to continue, revise or stop. Compare what you believed on day one with the evidence now available. Lack of evidence is not proof that demand is absent; it tells you which question still needs work.

Work through it

1. Score customer clarity, offer clarity and economics from 0 to 3.
2. Record what changed your mind.
3. Choose one question to carry into measurement.

Check your work

Your decision cites specific evidence and acknowledges its limits.

Watch for: Moving on only because a week has passed.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

Review rubric: 0 = unfamiliar; 1 = can explain; 2 = can do with guidance; 3 = can do, verify and explain a tradeoff. Revisit the lowest consequential skill.

The principle

A funnel is a simplified account of customer behavior, not a law that everyone follows. Map discovery, consideration, commitment and first useful value. A signup can be measurable without being valuable. Include where people return, hesitate or leave.

Work through it

1. Draw five steps from discovery to first useful value.
2. Write the customer question at each step.
3. Pick one business outcome and one customer outcome.

Check your work

The final step describes useful value, not merely a page visit.

Watch for: Optimizing an easy event that has little connection to value.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

An event specification states what happened and how you will know. Define the trigger, allowed parameters, destination and owner. Distinguish clicking submit from a successful submission. Document how failures and duplicates should behave before installing tracking.

Work through it

1. Write specs for three events.
2. Specify the success condition and duplicate prevention.
3. Remove personal information from URLs and event parameters.

Check your work

A tester could reproduce success and failure without guessing.

Watch for: Counting every button click as a completed lead.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Consistent campaign labels let you compare sources without guessing what a name means. Decide a lowercase naming convention and document it. Campaign parameters label traffic; they do not prove that the campaign caused an outcome. Keep personal data out of link parameters.

Work through it

1. Create three draft tagged links.
2. Define source, medium and campaign naming rules.
3. Check destinations and document how internal links will avoid overwriting acquisition context.

Check your work

Another person would assign the same names to the same campaign.

Watch for: Using a different spelling for the same channel in every link.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A tag firing is only one part of verification. Follow a controlled test from interaction to the reporting destination and, where applicable, the lead or order record. Check expected consent behavior and rejected submissions. Keep test records clearly labeled and exclude them from performance conclusions.

Work through it

1. Use a sandbox or write a test protocol for success, failure and retry.
2. Record expected versus observed events.
3. Check one event in a debugging view or mark verification pending.

Check your work

The log shows both what passed and what remains unverified.

Watch for: Reporting instrumentation as verified without observing the destination.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A baseline records what is measurable before an intervention. Use the same date range and definitions across the report. Missing data is not zero. Small samples deserve counts alongside rates because a large percentage change can represent one additional person.

Work through it

1. Choose a reporting window and three metrics.
2. Record numerator, denominator and data source.
3. Mark unavailable values unknown rather than filling them with guesses.

Check your work

A reader can reproduce every reported rate from its counts.

Watch for: Mixing time windows or silently replacing missing values.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Use a controlled journey to find where the measurement story breaks. The goal is traceability, not a perfect dashboard. Document the evidence for each handoff, including consent, duplication, confirmation and the first-value event. If access is unavailable, produce the exact verification plan and label it pending.

Work through it

1. Run or specify one end-to-end test.
2. Create an event-to-record evidence table.
3. Prioritize the first measurement defect to repair.

Check your work

A decision maker can distinguish observed events from planned instrumentation.

Watch for: Treating screenshots of a configuration screen as proof of collected data.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Before explaining performance, check whether the data can support the explanation. A tracking defect can resemble a marketing problem. Separate verified measures, partial measures and unavailable measures. Use the next week to research intent without pretending incomplete tracking is reliable.

Work through it

1. Classify each metric by confidence.
2. Identify a decision you cannot yet make.
3. Write the smallest verification step that would change that.

Check your work

You can name the largest reporting blind spot and its consequence.

Watch for: Building a confident narrative from unverified data.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

Review rubric: 0 = unfamiliar; 1 = can explain; 2 = can do with guidance; 3 = can do, verify and explain a tradeoff. Revisit the lowest consequential skill.

The principle

Keyword research starts with the words people use to describe a task. Collect phrases from authorized interviews, support questions and public discussions. Preserve the source. Tools can expand this vocabulary, but generated terms and estimated volumes still require review.

Work through it

1. Collect ten phrases tied to the chosen problem.
2. Record where each phrase came from.
3. Separate direct phrases from tool-generated suggestions.

Check your work

The list represents customer situations rather than only product jargon.

Watch for: Assuming an AI suggestion demonstrates search demand.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A query may signal learning, comparison, purchase or a different task entirely. The same word can mean different things in different contexts. Read the current search results to inspect likely intent and suitable destinations. Volume is useful only after relevance is established.

Work through it

1. Classify twelve queries by likely intent.
2. Inspect results for three ambiguous phrases.
3. Assign each query a destination or an exclusion reason.

Check your work

Every proposed paid keyword has a defensible connection to the offer.

Watch for: Choosing high-volume terms that attract the wrong task.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A theme connects a coherent customer need with a message and destination. Grouping is not just sorting by repeated words. Separate queries when they require materially different promises, pages or buying stages. Keep the first plan small enough to inspect manually.

Work through it

1. Group the query list into three themes.
2. Write one promise and one destination per theme.
3. Flag phrases that do not fit any supported offer.

Check your work

Each theme could be explained as one customer need.

Watch for: Creating groups so broad that no specific message fits.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A targeting keyword is not the same as the actual search term someone uses. Matching can include meaning and intent, so exact match does not mean only the identical wording. Exclusions and search-term review help control relevance, but careless negatives can block useful demand.

Work through it

1. Read the official matching guide linked below.
2. Draft a match-type rationale for each theme.
3. Propose five negatives and test whether any could exclude suitable customers.

Check your work

You can explain both an unwanted match and a useful match you might accidentally block.

Watch for: Assuming a match-type label eliminates the need for review.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Prioritize relevance, ability to serve the need, likely economics and evidence quality. Search-volume and bid estimates are planning inputs, not promises of traffic or cost. A small relevant theme may be more useful than a large one you cannot satisfy.

Work through it

1. Score each theme for fit and evidence.
2. Create low, middle and high cost assumptions.
3. Choose one theme and explain why another should wait.

Check your work

The priority has a business rationale beyond search volume.

Watch for: Treating an estimated CPC as a guaranteed purchase price.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

An intent map connects the query, customer situation, message, destination, exclusion and measurement goal. Include at least one rejected opportunity. The rejected item shows that you can make a judgment about fit rather than simply collecting keywords.

Work through it

1. Build a 20-query map from the week’s work.
2. Review five query-to-page matches aloud.
3. Select the first draft campaign theme and its main risk.

Check your work

Someone else can trace the query through to a useful customer action.

Watch for: Keeping irrelevant keywords to make the list look more complete.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Explain the reasoning without relying on platform jargon. Why does this customer use this phrase, why does the page help, and what would show that your interpretation was wrong? Keep uncertain phrases in a research backlog rather than automatically buying them.

Work through it

1. Write a short defense of your top three queries.
2. Name one contrary interpretation for each.
3. Choose one phrase to exclude or research further.

Check your work

The decision includes a plausible alternative explanation.

Watch for: Confusing confidence in your writing with evidence of intent.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

Review rubric: 0 = unfamiliar; 1 = can explain; 2 = can do with guidance; 3 = can do, verify and explain a tradeoff. Revisit the lowest consequential skill.

The principle

Campaign structure should make the business question easier to answer. Separate materially different goals, markets or budget responsibilities when necessary. Start with a small draft that maps directly to the brief. Complexity is not a measure of expertise.

Work through it

1. Choose one objective and conversion definition.
2. Sketch campaign and ad-group structure.
3. Explain the reason for each separation.

Check your work

Each group has a coherent intent, offer and destination.

Watch for: Copying a complex account structure without understanding its purpose.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Responsive search assets may appear in different combinations. Each should make sense alone and with the others. Write clear benefits, relevant details and a truthful next step. Check current format limits in official documentation rather than relying on memory.

Work through it

1. Draft six headlines and three descriptions.
2. Check combinations for repetition or conflicting promises.
3. Verify character limits and evidence behind each claim.

Check your work

Every asset is relevant, readable and consistent with the destination.

Watch for: Using unverified superlatives or fabricated results to improve the ad.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Locations, languages, schedules and exclusions affect who may encounter an ad. The right setting depends on who you can actually serve and when. Document choices in plain language, including platform defaults that deserve review. Keep all work in an offline plan or paused draft.

Work through it

1. Specify the service area and audience constraints.
2. List settings to verify before any launch.
3. Write one scenario in which targeting could waste budget.

Check your work

The proposed audience fits delivery capacity and the offer’s limits.

Watch for: Leaving defaults unexplained because they were preselected.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A budget model tests whether plausible performance could fit your economics. Work backward from allowable customer acquisition cost through qualified-lead and conversion assumptions. A stopping rule prevents open-ended experimentation. Include enough room to observe the relevant buying cycle.

Work through it

1. Build a low/base/high conversion model.
2. Set a hypothetical maximum total loss and review date.
3. State the evidence required before a real launch.

Check your work

The model distinguishes assumptions, media costs and other acquisition costs.

Watch for: Treating a daily budget setting as the entire risk-control plan.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Bidding automation uses the goals and signals supplied to it. A system can efficiently seek an event that is poorly aligned with customer value. Compare options against measurement quality, conversion delay and business constraints. Record which recommendations you accept, reject or need to investigate.

Work through it

1. Describe the desired optimization signal.
2. Identify the consequence of a false conversion.
3. Write a rationale for the draft bidding approach using current official guidance.

Check your work

The selected goal measures something worth acquiring.

Watch for: Accepting every automated recommendation as a business strategy.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A rehearsal should expose missing dependencies before money is committed. Review targeting, assets, landing page, measurement, follow-up and budget ownership together. A launch-ready checklist is not permission to launch; the exercise ends with a documented decision.

Work through it

1. Assemble the complete paused or offline campaign packet.
2. Walk a reviewer through the customer journey.
3. Record pass, fail or pending for each launch dependency.

Check your work

The packet ends with an explicit draft-only status and unresolved issues.

Watch for: Launching to discover that the form or follow-up was never tested.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A strong campaign review explains what was chosen, what was excluded and what evidence would change the plan. Account scores and cosmetic completeness do not replace business judgment. Keep the draft version so future changes can be compared with a stable starting point.

Work through it

1. Summarize the campaign in five sentences.
2. Name the two biggest risks.
3. Choose whether to revise, keep researching or retain the draft for later.

Check your work

A reviewer can identify the tradeoff you made and why.

Watch for: Calling an offline campaign a successful advertising case study.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

Review rubric: 0 = unfamiliar; 1 = can explain; 2 = can do with guidance; 3 = can do, verify and explain a tradeoff. Revisit the lowest consequential skill.

The principle

The landing page should continue the expectation created by the query and message. If the visitor must reinterpret the promise, the journey becomes harder. Check the buyer, result, proof, commitment and next step. A coherent page serves one primary task clearly.

Work through it

1. Compare the ad promise with the first screen.
2. Highlight any mismatched audience or result.
3. Draft a headline and explanation that preserve the promise.

Check your work

The visitor can tell whether the offer is relevant without guessing.

Watch for: Improving ad appeal by promising more than the page delivers.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

An objection may reflect missing information, a real mismatch or a delivery limitation. Address it honestly. Explain scope, process, effort, timing and costs where known. Specific proof is more useful than decorative trust badges or unsupported claims.

Work through it

1. List five likely buyer questions.
2. Match each with evidence or an honest limitation.
3. Rewrite two vague claims into concrete explanations.

Check your work

The page answers a real decision question with supportable information.

Watch for: Inventing testimonials or hiding exclusions in small print.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Usability is about whether someone can complete the intended task. Observe navigation, reading, form entry, feedback and recovery on a small screen. Use keyboard access and visible focus as part of the check. Do not infer usability only from how attractive a screenshot looks.

Work through it

1. Choose a task and success condition.
2. Test at a narrow viewport and with a keyboard.
3. Record one observed obstacle and its severity.

Check your work

The proposed change responds to an observed problem.

Watch for: Changing layout based only on personal taste.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Ask for information that is needed to fulfill the next action. Explain why sensitive or unfamiliar fields are requested. Show clear errors and a meaningful confirmation. The customer’s journey continues after submission, so the next promise needs an owner.

Work through it

1. Review each form field for purpose.
2. Draft error and confirmation messages.
3. Specify the follow-up owner, timing and fallback.

Check your work

A person knows what happens after submitting and what to do if it fails.

Watch for: Collecting extra fields simply because a CRM supports them.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A useful hypothesis explains a mechanism: a change should affect a particular behavior for a reason. Define a primary measure and a guardrail. Low traffic may support qualitative task checks before an A/B test; a small temporary lead is not reliable evidence of a winner.

Work through it

1. Name one observed conversion obstacle.
2. Write a change, expected mechanism and primary metric.
3. Choose a suitable evaluation method and state its limits.

Check your work

The test can teach you something even if the hoped-for outcome does not occur.

Watch for: Testing several unrelated changes and attributing the result to one.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Preserve the original and make one focused revision. Explain what observation motivated the change and how you will evaluate it. Use a draft, screenshot annotation or local prototype; public publication is not required to learn the skill.

Work through it

1. Save the starting page or annotated description.
2. Create one revised section and confirmation path.
3. Run a task check and document remaining confusion.

Check your work

The artifact shows the original problem, the revision and the review evidence.

Watch for: Claiming a conversion lift from a design revision that has not been measured.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Review the customer’s ability to understand and complete the task, not only the visual finish. Separate observed improvement in a task check from measured business impact. Carry unresolved questions into creative work so the message does not outrun the experience.

Work through it

1. Write what became clearer and how you know.
2. List the business effect that remains unmeasured.
3. Choose one next observation to collect.

Check your work

The conclusion stays within the limits of the evaluation.

Watch for: Equating a reviewer’s preference with increased sales.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

Review rubric: 0 = unfamiliar; 1 = can explain; 2 = can do with guidance; 3 = can do, verify and explain a tradeoff. Revisit the lowest consequential skill.

The principle

Creative can introduce a problem, demonstrate a process, explain an offer or answer an objection. Choose one job for each concept. Paid social often reaches people who were not actively searching, so the opening needs to make relevance understandable without exaggeration.

Work through it

1. Choose one customer situation and creative job.
2. Write an opening that names the situation.
3. Define the action appropriate to the person’s readiness.

Check your work

The concept has a specific reason to exist in the customer journey.

Watch for: Trying to teach, entertain, close and upsell in one short asset.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A creative angle is a different reason to care, not just a different color. Compare a problem demonstration, a process explanation and a concrete offer. Keep the underlying audience and promise stable enough to understand what the comparison would teach you.

Work through it

1. Draft three short concept cards.
2. Give each a distinct reason to care.
3. Specify what evidence each concept needs.

Check your work

The concepts test different messages while respecting the same offer.

Watch for: Calling three nearly identical designs a strategic creative test.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A demonstration should show the product or process accurately. Outline the initial problem, the action and the visible result. Label prototypes and illustrative outputs. A polished video cannot substitute for a capability that does not exist.

Work through it

1. Write a six-frame storyboard.
2. Describe the visual evidence needed in each frame.
3. Add captions and identify any simulated material.

Check your work

The sequence is understandable without sound and makes no unsupported claim.

Watch for: Using generated imagery as proof of real client results.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Review the message, visual accuracy, rights, readability and destination together. Check platform rules for the intended format before launch. AI-assisted assets still need human review, especially faces, hands, product details and any implied performance claim.

Work through it

1. Review one draft against five concrete checks.
2. Rewrite an unclear call to action.
3. Check the image at phone size and provide useful alternative text.

Check your work

The asset is legible, honest and connected to a working next step.

Watch for: Letting visual polish conceal inaccurate product details.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A comparison needs a question, consistent definitions and enough observation time. Decide what you would measure at attention, click and useful-action stages. A cheap click can still be poor demand. Use labeled sample data when no live campaign is authorized.

Work through it

1. Choose two concepts and one primary question.
2. Define a useful-action metric and a quality guardrail.
3. Create a hypothetical results table and explain what would remain uncertain.

Check your work

The decision would consider downstream quality as well as click performance.

Watch for: Declaring a winner using only engagement or a tiny early sample.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Combine the brief, concepts, storyboard, draft asset and review checklist. Include the assumptions required before a paid test could be sensible. The packet should be usable by someone else without a long explanation and should remain clearly labeled as practice.

Work through it

1. Assemble three concepts and one finished draft.
2. Attach claim support and review notes.
3. Write the proposed comparison and stop conditions.

Check your work

A reviewer can tell what is ready, what is illustrative and what still needs approval.

Watch for: Presenting a proposed test as a completed campaign result.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Look for specificity and usefulness rather than novelty alone. Compare the three angles against the customer situation and offer. Keep an explanation of why you rejected a concept; learning includes recognizing ideas that would attract the wrong people.

Work through it

1. Rank the concepts with reasons.
2. Identify one likely misunderstanding.
3. Revise the leading concept and record the next evaluation.

Check your work

The chosen idea is supported by a customer need, not only a trend.

Watch for: Assuming a viral format guarantees qualified demand.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

Review rubric: 0 = unfamiliar; 1 = can explain; 2 = can do with guidance; 3 = can do, verify and explain a tradeoff. Revisit the lowest consequential skill.

The principle

An email should have a clear purpose that fits what the recipient requested. Distinguish transactional information from optional marketing. Document signup expectations, sender identity and an accessible unsubscribe path. This exercise drafts the journey without sending messages.

Work through it

1. Write the signup promise in plain language.
2. Map the first three email purposes.
3. Specify consent records and the way to stop marketing messages.

Check your work

Each planned message has a reason the recipient would expect it.

Watch for: Treating any collected email address as permission for all marketing.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

The first message should fulfill the signup promise and make one useful action easy. Explain what was requested, deliver it and clarify what comes next. A welcome email cannot repair an unusable product or substitute for access to the promised resource.

Work through it

1. Draft subject, opening and one primary action.
2. Check that the promised resource can be accessed.
3. Write a helpful fallback if the link fails.

Check your work

The recipient could complete one useful task from the message.

Watch for: Filling the welcome with unrelated offers before delivering the promise.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Qualification connects the inquiry with the need, service fit, readiness and capacity to deliver. Write an operational definition that two people can apply consistently. Avoid using a vague score as a substitute for the evidence behind it. Keep unnecessary personal information out of learning notes.

Work through it

1. Write three fit criteria and two disqualifiers.
2. Classify five fictional inquiries.
3. Explain an ambiguous case and the question needed to resolve it.

Check your work

Another reviewer reaches similar classifications from the same evidence.

Watch for: Assuming every form submission has the same business value.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Follow-up should resolve a question or help someone reach value. Assign ownership, timing and a stopping condition. Describe what happens when someone replies, declines or unsubscribes. Automation needs an exit path as well as a trigger.

Work through it

1. Map a three-step draft sequence.
2. Add reply, unsubscribe and poor-fit exits.
3. Identify the human owner and a failed-delivery fallback.

Check your work

The sequence can stop promptly when it is no longer appropriate.

Watch for: Continuing automated messages after a human conversation has changed the context.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Compare sources using a consistent maturation window and definitions. Count qualified opportunities, customers and delivery effort alongside lead volume. A more expensive lead may be more useful, but do not infer profitability without contribution and retention evidence.

Work through it

1. Build a sample cohort table for two sources.
2. Calculate cost per lead and per customer.
3. Name what additional data is needed to compare contribution.

Check your work

The conclusion distinguishes lead cost, acquisition cost and actual profit.

Watch for: Adding overlapping platform-attributed revenue as though every claim were unique.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Test the journey without enrolling real contacts in a marketing sequence. Use a sandbox, a preview or a written test script. Check delivery content, links, expectations, segmentation and exits. A configuration screen is not proof of end-to-end delivery.

Work through it

1. Assemble the signup promise and draft sequence.
2. Run a preview or document a sandbox test.
3. Record passed, failed and unverified handoffs.

Check your work

The record shows where a real recipient could become confused or stuck.

Watch for: Calling delivery verified when only the template was saved.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Use the week’s evidence to decide which source and message deserve another test. Include customer fit, first useful action and operational effort. Keep the unknowns visible. More names in a list are not automatically a healthier acquisition system.

Work through it

1. Summarize the strongest source-quality signal.
2. Record one follow-up defect and its owner.
3. Decide what to revise before any live sequence.

Check your work

The decision connects the communication to customer progress.

Watch for: Optimizing subscriber count while ignoring whether people receive value.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

Review rubric: 0 = unfamiliar; 1 = can explain; 2 = can do with guidance; 3 = can do, verify and explain a tradeoff. Revisit the lowest consequential skill.

The principle

Useful search content answers a real question for a specific audience. Inspect intent, existing results and what firsthand information you can add. Choose a page that fits the customer journey. A collection of keywords does not by itself make a helpful article.

Work through it

1. Choose one relevant query and intended reader.
2. Outline a direct answer and supporting evidence.
3. Define the useful next action and an internal link.

Check your work

The outline offers something useful beyond a rephrasing of existing pages.

Watch for: Publishing thin near-duplicate pages for every wording of a query.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

A useful page must be reachable and understandable. Check status code, indexability, canonical URL, crawlable links and meaningful page text. A sitemap helps discovery but does not guarantee indexing. Keep structured data consistent with what a visitor can actually see.

Work through it

1. Inspect one page’s response and indexability.
2. Check title, canonical and internal links.
3. Record one technical defect or a verified clean result.

Check your work

The check separates what you observed from what a search engine may later do.

Watch for: Assuming a sitemap submission guarantees ranking.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Answer-focused writing starts with the answer, then explains context, evidence and limitations. Give readers descriptive headings, named sources and useful examples. AI-search visibility is an outcome to observe; formatting or schema cannot guarantee a citation in an AI answer.

Work through it

1. Rewrite a section with a direct answer first.
2. Add a primary source and a limitation.
3. Check that any structured data matches the visible page.

Check your work

A reader can understand the answer without losing its qualifications.

Watch for: Selling a special markup trick as guaranteed AI visibility.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

An AI answer is a variable observation influenced by the prompt, model, date and context. Use a fixed prompt set and record citations and errors. A mention is not proof of traffic or revenue. Treat a small sample as a diagnostic, not a market-share estimate.

Work through it

1. Write five customer discovery prompts.
2. Record tool, date, prompt and returned citations.
3. Identify one factual error or evidence gap worth addressing.

Check your work

Another person could repeat the observation and compare its limits.

Watch for: Claiming universal visibility from one favorable answer.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

An integrated plan connects the customer, offer, message, destination, measurement and follow-up. Each channel needs a job. Choose a primary experiment rather than launching everything simultaneously. State the dependencies that must be true before scaling activity or spending.

Work through it

1. Map one primary channel and one supporting channel.
2. Assign owner, artifact, metric and review date.
3. Write the condition for keeping, revising or stopping the experiment.

Check your work

The plan shows how the customer reaches value and how you will learn.

Watch for: Confusing a long list of channels with an executable strategy.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Your capstone is a record of reasoning: customer brief, event map, intent map, campaign draft, page revision, creative packet, email journey and acquisition plan. Label practice data and unresolved verification. Explain a decision you changed, not only the work you completed.

Work through it

1. Assemble the eight weekly artifacts.
2. Write a one-page case study with evidence and limits.
3. Ask a reviewer to challenge the weakest assumption.

Check your work

The portfolio is truthful about what was drafted, observed and actually measured.

Watch for: Inventing results to make a practice project look commercial.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

The principle

Completing lessons establishes practice, not permanent expertise. Pick the most consequential remaining uncertainty and design one bounded experiment. Continue the rhythm of question, artifact, observation and decision. Revisit fundamentals when the customer, offer or channel changes.

Work through it

1. Score your skills and identify the weakest consequential gap.
2. Write the next experiment with a limit and review date.
3. Export your notes and schedule your own next practice session.

Check your work

Your next experiment follows from evidence and has a clear decision rule.

Watch for: Starting a new course to avoid using what you have already learned.

Evidence to keep: Your artifact or a description of it, its source/date, what you observed, and what remains uncertain. A draft or a pending verification is a valid result when labeled honestly.

Review rubric: 0 = unfamiliar; 1 = can explain; 2 = can do with guidance; 3 = can do, verify and explain a tradeoff. Revisit the lowest consequential skill.