Weekly Breakdown · CBDA · Guide to Business Data Analytics
Welcome to Weekly Breakdown, a new issue every Monday from certificate.tips. Each issue takes one certification exam, builds one realistic case study and runs it through five or six exam-style questions. Try each question before you read the breakdown: we explain why the right answer is right, why the other options are wrong, and which section of the official guide says so. All questions are original and based on the official guides. We do not publish exam dumps.
This issue has three cases and 15 questions, one case each for the Analyze Data, Use Results to Influence Business Decision-Making and Guide Organizational-Level Strategy domains. Issue 1 covered the other three domains. The question of the week, who decides how the data is analyzed, is Question 1.
Case 1. Northfield Clinics: who will miss the appointment?
Northfield Clinics runs 14 outpatient clinics and books about 310,000 appointments a year. 11% of patients do not show up, and every empty slot costs a doctor's time that another patient could have used.
The research question is agreed: which factors are linked to missed appointments, and which bookings are at risk? A data scientist will build the model. You are the business data analyst on the team. Guide sections behind this case: 2.3.1 Develop Data Analysis Plan, 2.3.2 Prepare Data, 2.3.3 Explore Data and 2.3.5 Assess the Analytics and System Approach Taken.
- Week 1. The operations director wants the analyst to choose the model.
- Week 2. 3% of appointment records have no booking channel.
- Week 3. A clinic manager doubts a proposed measure of patient engagement.
- Week 4. 2% of waiting times are exactly 999 minutes.
- Week 5. The data looks fine, but the predictions do not help clinic managers.
Question 1. Week 1: who decides how the data is analyzed
The operations director tells the analyst: you know the clinics best, so you decide which model the data scientist will use to predict no-shows. According to the Guide, how should the data analysis plan be developed?
- A. The analyst selects the model, since the analyst owns the business requirements.
- B. The analyst gives domain input or drafts the plan, and the data scientist decides the approach.
- C. The data scientist writes the plan alone, since planning the analysis is purely technical work.
- D. The operations director approves the model choice before anyone reviews the data.
Answer: B.
Section 2.3.1 says the business analysis professional provides insights into the plan or may draft the initial plan for review by the data scientist, but it is the data scientist who possesses deep technical expertise to decide how the data analysis will be conducted. The analyst makes sure enough business domain information reaches the data scientist and understands the models well enough to explain to stakeholders why a model is chosen. A is wrong: the Guide gives the decision on the approach to the data scientist. C is wrong: the plan needs the business domain information that the analyst provides. D is wrong: a sponsor does not choose the model, and the plan covers techniques, models, data sources and cleaning.
Read more: CBDA Analyze Data Questions: Preparation, Exploration, Modeling
Question 2. Week 2: records with no booking channel
While preparing the data, the data scientist finds that 3% of appointment records have no booking channel. She plans to drop these records because they are statistically insignificant. What should the analyst do?
- A. Agree, since a 3% gap cannot change the results of a predictive model.
- B. Fill the field with the most common booking channel to keep every record.
- C. Investigate with stakeholders and the recording process what the gap means first.
- D. Ask the data engineer to remove the booking channel field from the data set.
Answer: C.
Section 2.3.2 describes this case: a data scientist may choose to ignore observations with missing information because they may be statistically insignificant, but from a business standpoint further investigation may be required. Such scenarios are best handled by analysts with facilitation, collaboration and elicitation skills, through stakeholder collaboration and investigation of the recording process. All 3% could come from one channel, which would change the analysis. A is wrong: the statistical view alone can miss the business meaning of the gap. B is wrong: imputing a value may conflict with business rules, as the Guide notes for mean or median imputation. D is wrong: removing the field throws away data before its business meaning is known.
Question 3. Week 3: is this a good measure of engagement
During exploration the team plans to use the share of reminder text messages that patients open as the measure of patient engagement. A clinic manager doubts that opening a message shows engagement. Which data quality checkpoint is the manager raising?
- A. Data integrity
- B. Data validity
- C. Data reliability
- D. Data bias
Answer: B.
Section 2.3.3 lists four checkpoints the data scientist uses to assess data quality during exploration. Data validity asks whether the data is truly representative of an underlying construct, and the Guide's example asks whether win ratio is a good measure of monthly sales performance. The manager asks the same about opened messages and engagement. A is wrong: integrity asks whether the data can be trusted, for example whether it is structurally correct. C is wrong: reliability asks whether the same results come back when data is collected more than once. D is wrong: bias asks whether the data portrays an accurate picture, for example a sample that is not representative.
Question 4. Week 4: waiting times of 999 minutes
Exploration shows that 2% of recorded waiting times are exactly 999 minutes. The data scientist calls them outliers but is not sure whether to exclude them. What is the analyst's MOST useful contribution?
- A. Assess the data collection method to see whether a business rule explains the value.
- B. Replace each 999 with the median waiting time to keep the distribution intact.
- C. Keep the values, because every record that exists must be used in the model.
- D. Run more statistical tests until the outliers stop affecting the predictions.
Answer: A.
Section 2.3.3 gives a similar example from house price data: some prices were outliers, and the analyst helped by assessing the data collection method to identify missed business rules. The values turned out to be set artificially high to mark prices that could not be disclosed, which supported excluding them. A value of 999 may be the same kind of code, for example for a check-in time that was never recorded. B is wrong: imputing hides the code, and Section 2.3.2 warns that imputation may conflict with business rules. C is wrong: the Guide says outliers (noise) may signal that data needs to be excluded. D is wrong: more tests do not reveal the business rule behind the value.
Question 5. Week 5: the predictions do not help
The data passed exploration. After two model runs, the predictions still do not help clinic managers decide which patients to call before their appointment. What does the Guide say happens next?
- A. Repeat data exploration and data analysis, adapting the approach until it answers the question.
- B. Present the results as they are, since the data already passed exploration.
- C. Move on to recommending actions with the best model available so far.
- D. Close the initiative and report the analytics investment as a failure.
Answer: A.
Section 2.3.5 says that even when the results of data exploration are acceptable, the results of data analysis may fail to answer the questions being asked. In these scenarios data exploration and data analysis are repeated, and the iteration continues until the data scientist is comfortable with the data sources. Assessing the approach is performed iteratively with Explore Data and Analyze Data. B is wrong: passing exploration does not mean the analysis answers the question. C is wrong: Section 2.5.1 says recommendations come only after checking that the analysis answered the research question. D is wrong: the Guide expects the approach to adapt, not to stop.
Case 2. Fernway Market: from a model to a change
Fernway Market is an online grocery that handles about 1.2 million orders a quarter. Refund claims for missing or damaged items grew 35% in a year, and some of them look suspicious.
A proof of concept now flags suspicious refund claims. The analysis is done, and the team has to turn it into a business decision. Guide sections behind this case: 2.5.1 Recommend Actions, 2.5.2 Develop Implementation Plan, 2.5.3 Manage Change and 2.5.4 Select Techniques.
- Day 1. The data scientist reports an acceptable F1 score.
- Day 2. The head of customer service wants the model live next week.
- Day 3. The recommendation is approved and needs an implementation plan.
- Day 4. The COO says the work is finished because the model works.
- Day 5. The analytics lead wants to keep what the team learned.
Question 6. Day 1: an acceptable F1 score
The data scientist reports that the proof of concept flags suspicious refund claims with an acceptable F1 score. The head of customer service asks the analyst to start drafting change proposals. What should the analyst do FIRST?
- A. Evaluate whether the outcome answered the research question and addressed the business need.
- B. Rank the solution options by cost and send the top one to the decision-makers.
- C. Ask the data scientist to add more models to raise the F1 score further.
- D. Draft the implementation plan, since the model results are already acceptable.
Answer: A.
Section 2.5.1 says that before an analyst can recommend changes to address the business need, an evaluation is conducted to determine the success of the analysis: did the outcome answer the research question, and how well did it address the business need? If it did not, the cycle repeats with a new research question. Only then does the work move to solution options. B is wrong: rating and ranking options comes after the evaluation. C is wrong: the score is already acceptable, and model tuning is not the next business step. D is wrong: the implementation plan in Section 2.5.2 follows an agreed recommendation.
Read more: CBDA Business Decision Making Quiz: Recommendations and Change
Question 7. Day 2: live next week
The proof of concept used static data from last year. The head of customer service wants the model live in the refund workflow next week. Which question is MOST important for the analyst to raise before recommending deployment?
- A. What is the return on investment in production, including live data costs, against the savings?
- B. Can the F1 score be raised further by adding gradient boosting to the model mix?
- C. Which chart will show the F1 score most clearly to the executive team?
- D. Should the data scientist retrain the model on a smaller sample of claims?
Answer: A.
Section 2.5.1 has a close example, a pricing anomaly proof of concept built on static data. Analytics results alone are not sufficient to deploy the solution. Before recommending it, analysts consider the financial return on investment in production, including deployment costs such as live data streams against the savings, alternative solutions, process changes and how the deployed solution will behave. B is wrong: model tuning is technical work, not an assessment against business needs. C is wrong: presenting the score does not tell whether deployment is worth it. D is wrong: retraining on less data does not address deployment at all.
Question 8. Day 3: building the implementation plan
Decision-makers approve the recommendation: flag suspicious claims for manual review inside the refund workflow. The analyst now has to break this high-level change into lower-level tasks for the implementation plan. Which technique BEST fits?
- A. Functional decomposition into a work breakdown structure or story maps
- B. SWOT analysis of the strengths and weaknesses of the refund team
- C. Five Forces analysis of new entrants in the online grocery market
- D. A data dictionary that defines each field of the refund claims table
Answer: A.
Section 2.5.2 says that when developing an implementation plan, analysts break down the work to implement the proposed changes. Functional decomposition drills high-level tasks down into lower-level tasks and activities, often as a work breakdown structure or story maps. The plan also holds resources, high-level estimates from the stakeholders responsible for the tasks, and the dependencies between them. B is wrong: SWOT describes strengths, weaknesses, opportunities and threats, not tasks. C is wrong: Five Forces analyzes market forces, a strategy technique in Section 2.6.4. D is wrong: a data dictionary standardizes data definitions, not work.
Question 9. Day 4: is the work finished
The COO says the initiative is finished because the model works. The analyst disagrees. According to the Guide, where does the organization realize the value from its analytics efforts?
- A. When the change is implemented in the policies and procedures of the organization
- B. When the model reaches the target F1 score on the test data
- C. When the results are presented to the executive team
- D. When business stakeholders approve the data sources
Answer: A.
Section 2.5.3 says the change manager oversees the transformation of the analysis results into implemented policies and procedures, and that implementing change is the end goal and where the organization realizes the value from its analytics efforts. Analysts are well suited to this role because they keep continuity between the analytics work and implementation. B is wrong: model performance is a technical milestone, not realized value. C is wrong: Section 2.5 says reports are useful only when some action is taken. D is wrong: approving data sources is part of validating data in Section 2.2.4.
Question 10. Day 5: keeping what the team learned
Before the team moves on, the analytics lead wants to record what could be improved in the analytics approach, data resources and analytics capabilities for future initiatives. Which technique does the Guide list for this?
- A. Lessons learned
- B. Balanced scorecard
- C. Process analysis
- D. Risk analysis and management
Answer: A.
Section 2.5.4 lists lessons learned for compiling the learnings from the analytics initiative with respect to improvements that can be made to the analytics approach, data resources, analytics capabilities and organizational capabilities that can be leveraged for future initiatives. B is wrong: the balanced scorecard assesses the organizational impact of recommended actions. C is wrong: process analysis looks at the impact of recommendations on future-state processes. D is wrong: risk analysis covers risks, constraints and dependencies of implementing a change.
Case 3. Kestrel Parcel: becoming analytics-driven
Kestrel Parcel delivers parcels in three countries and has about 4,000 employees. Each of its 5 business units hired its own analysts, built its own reports and keeps its own data.
The CEO wants Kestrel to move from analytics-aware to analytics-driven, and asks the analyst to help shape the organization-level strategy. Guide sections behind this case: 2.6.1 Organizational Strategy, 2.6.2 Talent Strategy and 2.6.3 Data Strategy.
- Meeting 1. A proposal mixes country hubs with analysts inside the units.
- Meeting 2. The CEO plans to read one analytics summary a year.
- Meeting 3. A small initiative has budget for only two people.
- Meeting 4. Two price studies on the same customers cannot be compared.
- Meeting 5. Analysts spend days finding out what each table contains.
Question 11. Meeting 1: hubs and embedded analysts
Kestrel operates in three countries. The proposal: an analytics hub in each country with a central structure, plus analysts who stay inside the business units. Which organizational model is this?
- A. Centralized model, like an analytics Centre of Excellence
- B. Decentralized model, with analytics teams inside each unit
- C. Hybrid model, a mix of centralized and decentralized teams
- D. Functional model, with analytics placed under the IT office
Answer: C.
Section 2.6.1 describes three models. The hybrid model is a mix of centralized and decentralized analytics teams, for example a hub and spoke model with geographic separation for hubs and a centralized structure within a geography. Kestrel's proposal has both country hubs and embedded analysts. A is wrong: a centralized model is one analytics unit that supports the other business units. B is wrong: a decentralized model only embeds teams in units, with no central hubs. D is wrong: organizing by function, such as IT, is another choice the Guide mentions, but it does not describe this mix.
Read more: CBDA Analytics Strategy Quiz: Operating Models, Talent, Culture
Question 12. Meeting 2: one summary a year
The CEO supports the change but plans to leave analytics to the teams and read one summary a year. What does the Guide name as a necessity for analytics engagements to succeed?
- A. Executive leadership support, with a clear and direct channel to the analytics teams
- B. A single data lake that holds all raw data before any analysis begins
- C. Outsourcing the analytics work to a vendor with a proven track record
- D. A separate analytics budget that is approved for at least three years
Answer: A.
Section 2.6.1 says analytics teams are most effective when they are cross-functional and that executive leadership support is a necessity for analytics engagements to succeed. A clear and direct information channel between leadership and the analytics teams to review engagements and their outcomes ensures a shared understanding. One summary a year is not such a channel. B is wrong: data lakes appear in Section 2.6.2 as one of many tools, not as a condition for success. C is wrong: outsourcing requirements are one factor in how teams are organized, not a necessity. D is wrong: the Guide does not name a multi-year budget as the condition.
Question 13. Meeting 3: budget for two people
A small pricing initiative has budget for two people: a data engineer and one more role. The second person must bring knowledge of the business, interpret the data and turn the results into data stories. Which role does the Guide say can cover all three?
- A. A data or solution architect with strong design skills
- B. A business analysis professional with enough analytics and business knowledge
- C. An experience designer who builds prototypes for stakeholders
- D. A data scientist who creates and evaluates analytics models
Answer: B.
Section 2.6.2 says the roles have overlapping skills and can be merged based on the size of the initiative, industry, budget and capabilities. Its example: a business analysis professional with sufficient analytics and business knowledge can play the roles of SME, data analyst and data journalist, and support data architecture, data engineering and experience design. A is wrong: the architect designs data systems to capture and store data. C is wrong: the experience designer builds prototypes and visual enhancements. D is wrong: the data scientist builds models, and the Guide does not give this role the merged example.
Question 14. Meeting 4: two samples, no comparison
Two business units ran price studies on the same customers. One used a random sample, the other a quota sample, and the results cannot be compared. Kestrel adopts a policy that sampling methods are shared across analytics projects. Which pillar of the talent strategy is this?
- A. Establishing the right team structure for analytics initiatives
- B. Creating an eco-system for learning and development
- C. Establishing best practices for analytics initiatives
- D. Recruiting and retaining analytics professionals
Answer: C.
Section 2.6.2 names three pillars of a talent strategy: team structure, learning and development, and best practices. Establishing best practices involves identifying a standard set of tools and techniques, and the Guide's own example is a policy to ensure sampling methods between different analytics projects are shared across teams. A is wrong: team structure is about which roles work on an initiative. B is wrong: learning and development upgrades skills in business, platforms and data communication. D is wrong: recruitment and retention sit next to the three pillars, not among them.
Question 15. Meeting 5: what is in this table
Analysts in every unit spend days finding out what each table contains, where it came from and when it was last updated. Which planning consideration of an organization-level data strategy addresses this MOST directly?
- A. Data governance
- B. Data architecture
- C. Data security
- D. Metadata management
Answer: D.
Section 2.6.3 lists four planning considerations. Metadata management is the administration of information that is maintained about the data assets an organization collects and manages, which is what the analysts are missing: what a table holds, its origin and its freshness. A is wrong: data governance is the rules and policies that manage data assets to ensure high-quality data. B is wrong: data architecture is the models and standards for how data is collected, stored and integrated. C is wrong: data security protects data from a privacy and confidentiality perspective.
What this case teaches
- The data scientist decides how the analysis is done, and the analyst brings the business context: what a gap in the data means, which business rule sits behind an outlier, and whether the results answer the question.
- A good model is not yet a decision. Check that the analysis answered the research question, test the deployment case, plan the work and manage the change, because value comes from implemented change.
- Analytics-driven organizations choose a team model, keep executive leadership close, merge roles to fit the initiative, share best practices and manage metadata.
Practice the full exam
Our CBDA practice exam has 75 scenario questions across all six domains of the Guide, and it is free right now. Last verified: November 2026.
FAQ
What is Weekly Breakdown?
A weekly series from certificate.tips. Every Monday we take one certification exam, build realistic cases and break down exam-style questions on them with references to the official guide.
Are these real CBDA exam questions?
No. All 15 questions are original and written from the text of the IIBA Guide to Business Data Analytics. We do not publish exam dumps.
Which Guide sections does this issue cover?
Sections 2.3.1, 2.3.2, 2.3.3 and 2.3.5 in case 1, Sections 2.5.1, 2.5.2, 2.5.3 and 2.5.4 in case 2, and Sections 2.6.1, 2.6.2 and 2.6.3 in case 3.
Sources
- IIBA: Guide to Business Data Analytics
- IIBA: Certification in Business Data Analytics (CBDA)
- IIBA: CBDA Certification Handbook
- UCI Machine Learning Repository, the source of the house price data in the Guide's exploration example
- scikit-learn: f1_score, how the F1 score combines precision and recall
The case and company are fictional. Not affiliated with Scrum.org or IIBA.