Activation, Onboarding and Conversion Optimization

Introduction

Growth is not created merely by acquiring more visitors. It is created when people understand a product, experience meaningful value, adopt a useful behavior, and continue using the product in a way that supports sustainable economics. Activation, onboarding, and conversion optimization therefore sit at the centre of growth hacking. They connect marketing promises with product experience.

Activation is the process through which a user reaches an early, meaningful indication of value. Onboarding is the designed sequence that helps a new user understand and reach that value. Conversion optimization is the systematic improvement of the proportion of users who complete a desired action, such as registering, starting a trial, making a purchase, inviting a colleague, or upgrading to a paid plan.

These activities should not be treated as isolated interface techniques. They involve customer research, behavioral science, product analytics, experimentation, service design, copywriting, privacy governance, and commercial judgment. A fast sign-up flow that produces low-quality users is not necessarily successful. Similarly, a high trial-to-paid conversion rate may be misleading if customers cancel quickly or require excessive support.

This chapter develops a practical and analytical approach to mapping customer journeys, identifying friction, improving time-to-value, using personalization responsibly, and running controlled tests.

Learning Objectives

By the end of this chapter, the learner should be able to:

1. Distinguish acquisition, activation, conversion, retention, and revenue outcomes.
2. Map customer journeys and identify cognitive, emotional, technical, and procedural friction.
3. Design onboarding around user intent and time-to-value rather than feature exposure.
4. Select meaningful activation events and construct reliable conversion funnels.
5. Apply behavioral design principles without using manipulation or dark patterns.
6. Design controlled experiments and interpret their results cautiously.
7. Use segmentation and personalization while respecting privacy, fairness, and user autonomy.
8. Evaluate onboarding initiatives using both immediate and downstream metrics.

1. The Strategic Role of Activation and Conversion

A customer journey is the sequence of experiences through which a prospective customer moves from recognizing a problem to obtaining value and deciding whether to continue. A simplified growth journey is:

Reach → Consideration → Sign-up → Onboarding → Activation → Retention → Revenue → Advocacy

The stages are analytically useful but not always linear. A customer may compare alternatives after signing up, return to research after encountering a technical obstacle, or invite another user before personally completing setup.

1.1 Important distinctions

A registration is observable, but it may not be activation. For a project-management product, activation might mean creating a project, adding a task, and inviting a collaborator. For a language-learning application, it might mean completing an initial lesson and returning for a second session. The correct activation event must reflect a behavior plausibly connected to later value.

1.2 The AARRR logic and its limitations

The commonly used AARRR sequence—acquisition, activation, retention, referral, and revenue—helps teams organize growth questions. However, it is not a universal law and should not be treated as a rigid funnel. Referral may occur before long-term retention; revenue may precede repeated use in a transactional business. Teams should adapt the sequence to their business model and customer decision process.

The central managerial principle is to optimize the whole value chain rather than a single percentage. A design that increases sign-ups but reduces activation may worsen performance. A design that increases immediate purchase but increases refunds may also destroy value.

2. Customer-Journey Mapping and Friction Diagnosis

2.1 Journey maps as analytical instruments

A customer-journey map records what users are trying to accomplish, what they do, what they experience, and what the organization does in response. A useful map contains at least the following layers:

• User goal: the outcome the person wants, not merely the feature they click.
• Stage: the point in the journey.
• Action: observable behavior.
• Touchpoint: website, mobile application, email, sales call, support channel, or external integration.
• Thought and emotion: uncertainty, confidence, urgency, anxiety, or motivation.
• Friction: anything that increases effort, confusion, risk, delay, or distrust.
• Evidence: analytics, interviews, support tickets, session recordings, surveys, or usability tests.
• Opportunity: a specific improvement hypothesis.

A journey map should be based on evidence rather than internal assumptions. Analytics can show where users stop; interviews and usability testing help explain why.

2.2 Types of friction

Friction is not always bad. A confirmation step can prevent an expensive mistake, and identity verification may be necessary for safety or regulation. The relevant question is whether friction is justified by the value or risk it addresses.

Common categories include:

1. Cognitive friction: unclear language, too many choices, unfamiliar concepts, or inconsistent navigation.
2. Procedural friction: unnecessary fields, repeated data entry, complicated eligibility rules, or excessive steps.
3. Technical friction: slow loading, broken links, errors, poor mobile performance, or failed integrations.
4. Trust friction: unclear pricing, hidden conditions, weak security communication, or lack of social proof.
5. Emotional friction: fear of making an irreversible mistake, embarrassment, perceived surveillance, or lack of confidence.
6. Value friction: users cannot see why a step matters or cannot experience value until too much work has been completed.

Friction should be prioritized by severity, frequency, affected segment, business impact, and feasibility of correction. A minor issue affecting thousands of users may deserve more attention than a severe issue affecting only a very small segment, although ethical and legal risks may change that priority.

2.3 A practical diagnosis sequence

A disciplined diagnosis can follow these questions:

1. What is the user trying to accomplish?
2. What behavior indicates progress?
3. At which step does abandonment increase?
4. Is the problem lack of motivation, lack of ability, or lack of a prompt?
5. What evidence supports the diagnosis?
6. Could the proposed solution create harm, bias, or downstream quality problems?
7. How will success and unintended consequences be measured?

This sequence prevents teams from treating every drop-off as a copywriting problem.

3. Activation and Time-to-Value

3.1 Defining activation

Activation should be defined behaviorally. “User understands the product” is difficult to measure; “user imports one data set and generates a report” is observable. A useful activation event has three characteristics:

• It represents a meaningful user outcome.
• It occurs early enough to influence onboarding decisions.
• It is associated, based on evidence, with later retention or commercial value.

Correlation alone does not prove that the event causes retention. Highly motivated users may both activate and retain. Consequently, activation analysis should be complemented by qualitative research and, where feasible, controlled experiments.

3.2 Time-to-value

Time-to-value is the elapsed time between a defined starting point and the first meaningful outcome. The starting point must be specified—for example, first landing-page visit, account creation, payment, or first login. The endpoint must also be defined operationally.

For user i:

where t_value,i is the timestamp of the first meaningful value event and t_start,i is the timestamp of the chosen starting event. Both timestamps must use the same time zone and clock convention. Time-to-value is usually measured in minutes, hours, or days.

Because time-to-value is often skewed by a small number of very slow users, teams should inspect the median and relevant percentiles rather than relying only on the arithmetic mean. A median of two hours and a 90th percentile of four days suggests a different problem from a median of two days and a 90th percentile of three days.

3.3 Designing for early value

Effective onboarding usually follows a value-first sequence:

1. Establish the user’s intended outcome.
2. Remove or defer tasks that do not contribute to that outcome.
3. Provide a credible first success.
4. Explain the next useful action.
5. Introduce complexity progressively.
6. Confirm progress and provide recovery when errors occur.

This approach is sometimes called progressive disclosure: show what is necessary for the current task and reveal advanced capability when it becomes relevant. It does not mean hiding important information. Pricing, permissions, data use, cancellation terms, and material constraints should remain clear.

4. Behavioral Design in Onboarding

Behavioral design applies knowledge about attention, motivation, memory, decision-making, and habit formation to the design of user experiences. It should support informed user goals rather than exploit predictable weaknesses.

4.1 Motivation, ability, and prompts

A practical behavioral model proposes that behavior is more likely when motivation, ability, and a timely prompt converge. The model is diagnostic rather than a precise predictive equation. If users do not complete a step, a team should ask whether they do not want to, cannot easily do it, or were not prompted at an appropriate time.

• Increase motivation by clarifying relevance and expected benefit.
• Increase ability by reducing effort, simplifying language, providing defaults, and offering examples.
• Improve the prompt by making it timely, visible, specific, and connected to the user’s goal.

A prompt cannot compensate indefinitely for a product that lacks value or requires unreasonable effort.

4.2 Choice architecture

Choice architecture concerns how options are organized and presented. Useful practices include:

• sensible defaults that are easy to change;
• grouping related choices;
• clear labels rather than internal product terminology;
• displaying the recommended path when it genuinely fits most users;
• preserving user control and showing consequences of important choices.

A default becomes unethical when it is designed to obtain consent that users would not reasonably understand, or when refusal is deliberately made difficult. “Confirmshaming,” disguised advertising, forced continuity, and confusing opt-out paths are examples of dark patterns.

4.3 Cognitive load and error recovery

New users have limited familiarity with the product’s concepts. Interfaces should reduce unnecessary working-memory demands through recognition rather than recall, meaningful grouping, visible system status, and immediate error messages. Error messages should explain what happened, why it matters, and what the user can do next.

Recovery is part of onboarding. A failed import should not leave the user with a blank screen and no explanation. Saving progress, allowing backtracking, and providing support pathways can preserve trust even when the initial process fails.

5. Funnel Measurement and Conversion Metrics

A funnel represents a sequence of defined events. Let n_j be the number of eligible users reaching stage j. The stage conversion rate from stage j to stage j+1 is:

This formula assumes that n_j is nonzero and that both counts refer to a consistent cohort and eligibility rule. The overall conversion from stage 1 to stage k is:

If users can enter the journey at different points, a simple sequential funnel may misrepresent behavior. Cohort definitions and event order must therefore be explicit.

Worked example

An online learning platform observes the following weekly cohort:

The landing-to-account conversion is 2,000 ÷ 10,000 = 20%. The account-to-lesson-start rate is 1,200 ÷ 2,000 = 60%. The lesson-start-to-completion rate is 720 ÷ 1,200 = 60%. The overall landing-to-paid conversion is 144 ÷ 10,000 = 1.44%.

The largest percentage loss is not automatically the best opportunity. The account-to-lesson stage loses 800 users, while the lesson-completion-to-paid stage loses 576 users. Teams should combine volume with user value, problem severity, feasibility, and evidence about causes.

5.1 Beyond conversion rate

Important companion metrics include:

• Absolute lift: p_B − p_A, where p_A and p_B are conversion rates for control and treatment.
• Relative lift: (p_B − p_A) ÷ p_A, provided p_A > 0.
• Completion time: time required to complete a task.
• Activation rate: proportion reaching the defined activation event.
• Retention: proportion returning or continuing to obtain value at a specified later time.
• Quality metrics: refund rate, support contacts, error rate, fraud rate, and downstream revenue.

A conversion improvement should be judged against guardrail metrics. For example, increasing checkout completion while increasing chargebacks is not an unqualified success.

6. Controlled Testing and Experimentation

6.1 The logic of an experiment

A controlled experiment compares outcomes for users exposed to different versions while attempting to make the groups comparable. A basic design includes:

1. A clearly stated hypothesis.
2. A defined target population.
3. A control condition and one or more treatment conditions.
4. Random assignment where operationally and ethically feasible.
5. A primary outcome specified before analysis.
6. A sufficient observation period to capture the behavior.
7. Guardrail metrics and a decision rule.

Example hypothesis: “For new self-serve users, replacing a feature tour with a goal-based setup checklist will increase activation because it reduces irrelevant cognitive load.” The hypothesis identifies population, intervention, outcome, and mechanism.

6.2 Statistical interpretation

For a binary outcome, the observed conversion rate in group g is:

where x_g is the number of conversions and n_g is the number of eligible users in group g. The difference between treatment and control is ∆ = p̂_T − p̂_C.

A difference may arise from the intervention or from random variation. Statistical inference requires assumptions about assignment, independence, measurement, and the analysis method. A p-value is not the probability that the hypothesis is true, and statistical significance does not establish practical importance. Teams should consider effect size, uncertainty intervals, business value, and potential harms.

For low-risk product decisions, a simple proportion comparison may be appropriate. For clustered users, repeated observations, unequal exposure, or time-to-event outcomes, more appropriate methods may be required. Analysts should avoid repeatedly checking results and stopping as soon as a favorable result appears, because this inflates false-positive risk. Sequential or Bayesian designs can be used, but they must be planned and interpreted according to their assumptions.

6.3 Experiment quality

Common threats include:

• Selection bias: treatment groups differ systematically.
• Contamination: users encounter both versions.
• Novelty effects: behavior changes temporarily because the design is new.
• Seasonality: results depend on academic calendars, holidays, or campaigns.
• Instrumentation error: events are missing, duplicated, or defined inconsistently.
• Interference: one user’s treatment affects another user, common in collaboration products.
• Multiple testing: many outcomes are examined until one appears favorable.

A test should not be considered successful solely because the primary metric improves. Its result must be interpreted in the context of retention, revenue quality, user trust, and operational cost.

7. Personalization and Segmentation

Personalization adapts content, sequence, defaults, or assistance to relevant differences among users. Segmentation groups users according to characteristics such as job role, use case, company size, acquisition source, device, or stated goal.

Useful personalization asks the user directly about intent and uses the answer to reduce irrelevant work. For example, a design application might ask whether the user wants to create a presentation, social post, or report, then provide an appropriate starting template.

Personalization should satisfy four tests:

1. Relevance: the difference improves the user’s task.
2. Transparency: the user can understand or control important adaptations.
3. Data proportionality: only necessary data is collected.
4. Fairness: the experience does not disadvantage protected or vulnerable groups without a legitimate reason.

Behavioral targeting can produce discriminatory outcomes when proxy variables reproduce sensitive characteristics. Teams should audit performance across meaningful segments and provide accessible alternatives. Consent, purpose limitation, retention, security, and rights to access or deletion may be required under applicable data-protection law.

8. A Professional Optimization Process

A repeatable operating process can be organized as follows.

Step 1: Define the customer and desired outcome

Identify the target customer, job to be done, context, constraints, and value proposition. Avoid optimizing for “all users” when different segments have different success conditions.

Step 2: Define the activation event

Select a meaningful behavior and document its event name, properties, eligibility, time window, and relationship to later value. Maintain a measurement dictionary so product, marketing, and analytics teams use consistent definitions.

Step 3: Map the current journey

Combine quantitative funnel data with interviews, usability sessions, support records, and technical monitoring. Document both successful and failed paths.

Step 4: Form a prioritized diagnosis

Rank friction points by user harm, number affected, strategic value, expected impact, confidence in the evidence, and implementation effort. A lightweight scoring model can support discussion, but it should not replace judgment.

Step 5: Generate interventions

Possible interventions include reducing fields, improving information hierarchy, offering templates, importing data, delaying advanced settings, adding contextual guidance, improving performance, clarifying pricing, or providing human assistance.

Step 6: Choose the evaluation method

Use usability testing for comprehension and task problems, qualitative interviews for motivations and anxieties, telemetry for behavioral patterns, and controlled experiments for causal evidence when feasible. Not every change requires an A/B test; a safety-critical defect should be corrected immediately.

Step 7: Test and monitor

Predefine the primary metric, guardrails, audience, duration, and stopping rule. Monitor instrumentation and customer feedback. Do not expose vulnerable users to avoidable risk merely to obtain experimental evidence.

Step 8: Institutionalize learning

Record the hypothesis, result, segment effects, limitations, and decision. A failed experiment can improve understanding if the test was valid and the learning is documented.

9. Illustrative Case: Onboarding a Small-Business Cash-Flow Platform

Background

The following case is fictional and illustrative. “FlowPilot” is a subscription platform that helps small businesses monitor cash flow. Visitors can create an account, connect a bank feed, categorize transactions, and view a 30-day cash forecast.

Problem

FlowPilot’s marketing campaigns generated many registrations, but only 28% of new accounts connected a bank feed within seven days. Interviews suggested that users were uncertain about security, did not know why connection was necessary, and encountered a long setup form before seeing any value. The team initially proposed adding more feature explanations.

Analysis

The team mapped the journey and found four major friction points:

• The landing page emphasized “powerful analytics” rather than the immediate cash-flow question.
• The onboarding form requested business details that were not required for the first forecast.
• The bank-connection screen displayed technical provider language and weak security reassurance.
• Users received no preview of what the completed forecast would look like.

The activation event was defined as “a new account connects at least one eligible bank feed and views its first forecast within seven days.” This definition was more meaningful than account creation because it represented initial product value.

Approach

The team developed three changes:

1. A goal-selection screen asking whether the user wanted to monitor upcoming bills, estimate cash runway, or identify unusual spending.
2. A shortened setup flow that deferred nonessential business-profile questions.
3. A transparent connection screen explaining permissions, encryption practices, data use, and the next step, accompanied by a sample forecast.

A controlled test compared the existing flow with the revised flow among new self-serve users. The primary metric was activation within seven days. Guardrails included support contacts, connection failure rate, cancellation within thirty days, and permission-related complaints.

Outcome

In the illustrative results, activation increased from 28% to 35%, an absolute increase of 7 percentage points. However, the team did not declare success immediately. Analysis showed that the revised flow improved activation for users seeking cash-runway information but had little effect on users primarily interested in expense categorization. The company therefore introduced segment-specific entry points while retaining a common transparent privacy explanation.

The team also discovered that one bank provider had a higher technical failure rate. Product and engineering work on the integration produced a larger improvement than further changes to interface wording.

Learning points

1. The best onboarding intervention may be technical reliability rather than persuasive copy.
2. Activation must represent customer value, not merely activity.
3. Segment differences can reveal why an average effect is incomplete.
4. Privacy communication should be informative, not a substitute for sound data governance.
5. Guardrail metrics prevent teams from celebrating a narrow improvement that creates downstream harm.

10. Technology and AI in Optimization

Product analytics platforms, event pipelines, feature-flag systems, usability-testing tools, and experimentation platforms can support optimization. AI can assist with clustering feedback, summarizing support tickets, generating interface alternatives, and identifying unusual funnel patterns. These uses require human validation because automated systems may misclassify intent, reproduce historical bias, or infer sensitive attributes.

AI-generated copy and personalization should be tested for accuracy, accessibility, cultural appropriateness, and unintended pressure. Organizations should document data sources, access controls, model limitations, and escalation procedures. An AI system should not independently make high-impact decisions about eligibility, pricing, credit, employment, or access to essential services without appropriate governance and legal review.

11. Challenges and Limitations

Optimization is constrained by several recurring problems:

• Local maxima: improving one step can conceal a larger problem elsewhere.
• Metric fixation: teams optimize what is easy to count rather than what matters.
• Short-termism: immediate conversion may be prioritized over trust and retention.
• Segment averaging: aggregate results hide differences among users.
• False certainty: observational correlations are treated as causal evidence.
• Operational constraints: support, engineering capacity, compliance, and partner systems limit feasible changes.
• Accessibility neglect: improvements for typical users may exclude users with disabilities.
• Ethical tension: persuasive design can become coercive when user autonomy is weakened.

The appropriate response is not to abandon measurement but to use multiple forms of evidence, make assumptions explicit, and include long-term and human outcomes in the decision process.

12. Best-Practice Checklist

Before launching an activation or conversion initiative, ask:

• Is the target user and desired outcome clearly defined?
• Does the activation event represent meaningful value?
• Have user research and behavioral data been combined?
• Is unnecessary effort removed without hiding important information?
• Are pricing, permissions, privacy, and cancellation terms clear?
• Is the design accessible across relevant devices and abilities?
• Are defaults reversible and choices understandable?
• Are primary and guardrail metrics specified in advance?
• Is the experiment method appropriate to the decision and risk?
• Will downstream retention, revenue quality, complaints, and fairness be monitored?
• Has the organization documented what was learned?

Chapter Summary

Activation is the point at which a user experiences meaningful product value, while onboarding is the designed process that helps the user reach that point. Conversion optimization should be understood as a system rather than a search for isolated percentage improvements. Customer-journey maps connect user goals, actions, emotions, touchpoints, and evidence. Friction may be cognitive, procedural, technical, trust-related, emotional, or value-related; effective diagnosis distinguishes necessary safeguards from avoidable obstacles.

Time-to-value provides a useful lens for assessing how quickly users obtain meaningful outcomes. Funnel metrics quantify progression, but they must be interpreted with cohort definitions, quality measures, and downstream effects. Behavioral design can improve motivation, ability, prompts, choice architecture, and error recovery, provided it preserves autonomy and avoids dark patterns.

Controlled experiments support causal learning when assignment, measurement, timing, and analysis are sound. Statistical significance is not the same as practical importance, and every test should include guardrails. Personalization can reduce irrelevant effort when it is relevant, transparent, proportionate, and fair. AI and analytics tools can extend research and optimization, but they require human oversight, privacy protection, and bias monitoring.

The professional objective is not simply to make more users click. It is to help appropriate users achieve legitimate goals efficiently, confidently, and sustainably while protecting trust and user welfare.

Key Terms

Activation: A meaningful early behavior indicating that a user has experienced product value.

Activation event: A specifically defined, observable behavior used to measure activation.

A/B test: A controlled comparison between a control experience and a treatment experience.

Behavioral design: Design informed by how people perceive, decide, and act.

Choice architecture: The organization and presentation of options that influences decisions.

Conversion rate: The proportion of eligible users completing a defined target action.

Cohort: A group of users sharing a defined entry condition or time period.

Conversion funnel: A sequence of stages representing progression toward a target outcome.

Dark pattern: A design practice that manipulates or obstructs users against their interests.

Guardrail metric: A measure used to detect undesirable consequences of an optimization.

Onboarding: The process of helping a new user understand and obtain initial value.

Personalization: Adaptation of an experience to relevant user characteristics or intent.

Progressive disclosure: Revealing information or functionality as it becomes relevant.

Retention: Continued valuable use after an initial interaction or activation.

Segmentation: Dividing users into groups with relevant shared characteristics.

Time-to-value: Elapsed time from a defined start event to a defined meaningful-value event.

Treatment: The version or intervention evaluated in an experiment.

User friction: Effort, uncertainty, delay, risk, or distrust that impedes progress.

Value proposition: The clear statement of the problem solved and benefit offered to a target customer.

Review Questions

Short Answer Questions

1. What distinguishes activation from registration?
2. Identify four categories of friction in a customer journey.
3. Why should time-to-value have clearly defined start and endpoint events?
4. What is the purpose of a guardrail metric?
5. State two ethical requirements for responsible personalization.

Long Answer Questions

1. Explain how a customer-journey map can combine qualitative and quantitative evidence to diagnose onboarding problems.
2. Discuss the role of motivation, ability, and prompts in behavioral design, including the limits of this model.
3. Explain how a business should select an activation event and evaluate whether it is associated with long-term value.
4. Describe the design requirements of a credible controlled experiment for conversion optimization.
5. Critically evaluate the tension between short-term conversion gains and long-term trust, retention, and customer welfare.

Analytical / Application Questions

1. A software product has high registration but low activation. Design a diagnostic plan that distinguishes motivation, ability, technical, and trust problems, and specify the evidence you would collect.
2. A treatment increases checkout conversion from 8% to 9% but also increases refunds and support complaints. Explain how you would evaluate whether the treatment should be rolled out.
3. Design an onboarding experiment for a professional learning platform. Define the target segment, activation event, hypothesis, primary metric, guardrails, and possible threats to validity.

Multiple Choice Questions with Answers

Question 1

Which behavior is the strongest example of an activation event for a project-management product?

A. Viewing the home page

B. Creating an account

C. Creating a project, adding a task, and inviting a collaborator

D. Receiving a marketing email

Correct Answer: C. Creating a project, adding a task, and inviting a collaborator

Question 2

If 2,000 users reach a funnel stage and 1,200 proceed to the next stage, what is the stage conversion rate?

A. 40%

B. 50%

C. 60%

D. 66.7%

Correct Answer: C. 60%

Question 3

What does time-to-value measure?

A. The cost of acquiring a customer

B. Elapsed time from a defined start event to a meaningful value event

C. The number of features used in a session

D. The duration of a marketing campaign

Correct Answer: B. Elapsed time from a defined start event to a meaningful value event

Question 4

Which is an example of trust friction?

A. A page loads slowly

B. A user cannot understand why a permission is requested

C. A user sees a relevant template

D. A user completes a short form

Correct Answer: B. A user cannot understand why a permission is requested

Question 5

What is the primary purpose of random assignment in a controlled experiment?

A. To guarantee a positive result

B. To make treatment and control groups more comparable on average

C. To eliminate the need for a primary metric

D. To increase the price paid by users

Correct Answer: B. To make treatment and control groups more comparable on average

Question 6

Which statement about statistical significance is correct?

A. It proves that the business impact is large

B. It proves that the hypothesis is true

C. It helps assess whether an observed difference is unlikely under a specified null model

D. It eliminates the need to examine guardrail metrics

Correct Answer: C. It helps assess whether an observed difference is unlikely under a specified null model

Question 7

Which practice best represents progressive disclosure?

A. Displaying every advanced feature during first login

B. Hiding pricing and permissions until after payment

C. Showing functionality when it becomes relevant to the user’s current task

D. Removing all help content from the product

Correct Answer: C. Showing functionality when it becomes relevant to the user’s current task

Question 8

Which is an appropriate principle for responsible personalization?

A. Collect all available personal data

B. Make important adaptations invisible to users

C. Use only data relevant to the user’s task and provide appropriate transparency

D. Optimize only for the average user

Correct Answer: C. Use only data relevant to the user’s task and provide appropriate transparency

Question 9

A landing page has 10,000 visitors and 144 paid customers. What is the overall landing-to-paid conversion rate?

A. 0.144%

B. 1.44%

C. 14.4%

D. 69.4%

Correct Answer: B. 1.44%

Question 10

Why are guardrail metrics used in experimentation?

A. To replace the primary outcome

B. To detect harmful or undesirable consequences of an intervention

C. To guarantee statistical significance

D. To increase sample size automatically

Correct Answer: B. To detect harmful or undesirable consequences of an intervention