B · Business understanding and domain modeling
Data model
Entities, relations, grain, and their mapping to storage.
Classification boundary
Include: Include entity relationships, grain, data quality, storage mapping, and data lifecycle.
Choose a neighboring leaf when: Use feature model for feature definitions and time windows, and C metrics/causality for interpreting business metrics.
This capability may appear in standards across different stages. Explore stage expectations
37 questions
Choose a group based on the evidence you need. Groups follow existing question formats, not difficulty or mandatory prerequisites. Each batch has up to eight questions.
Concepts and boundaries3
Check concepts, assumptions, and conditions of use.
Questions 1–3
- Taking concepts such as orders, contracts, and settlements, how are they mapped into the data model? Which concepts were split into several entities, which were merged, and why?
- In a behavioral event table, what exactly does one row represent? Explain how you settle its grain, and which statistics go wrong when the grain is unclear.
- Orders, payments, and refunds are often stored in one table. How would you distinguish their business identities, lifecycles, and cardinalities? Compare conditions, costs, and validation for separate models and combined storage.
No questions in this section match your target. ·
Judgment and options34
Compare goals, constraints, options, and trade-offs.
Questions 1–8
- Product, finance, and growth teams each report “active users,” but the numbers differ. What would you ask before deciding whether to standardize the definition?
- A user consented to product analytics, then withdrew consent. Derived metrics already reached reports, experiments, and model-training data. How would you model consent state, purpose scope, and propagation?
- A business object moves across teams as owners change. Audit must reconstruct the owner on a past date, while daily reporting needs current ownership. How would you model both queries?
- A user asks to have personal data deleted, while finance must retain transaction records for audit. How would you define the boundary among deletion, anonymization, and traceability?
- A monthly close report was sent to leadership. Days later, a late refund and price correction arrive. The business wants current figures updated, while finance must explain the prior report. How would you model this?
- How does your own business model interact with the industry's or a partner's common model? When fields, definitions, or concepts do not line up, how do you handle it?
- Mobile events may arrive days late, while reports must reflect event time. How would you model and process late, corrected, and duplicate events?
- A critical dataset is reused by several products and reports, but no one can explain its sources, owners, or sensitive fields. How would you improve governance without blocking current use?
No questions in this section match your target. ·
Questions 9–16
- An account moves from sales to customer success mid-month. Monthly metrics must show the outcome and who owned it at each point. How would you define ownership events and reporting rules?
- A fee rule changes today, but a disputed order was created under the old rule. How would you preserve rule versions and transaction facts so finance can reproduce the original calculation?
- A company buys 100 seats used across subsidiaries. Headquarters signs the contract, but some subsidiaries want separate invoices and permission administrators. How would you distinguish customer, contract, organization, account, member, and seat so access and billing stay consistent?
- Customer master data has separate identifiers and states in CRM, billing, and product systems, so downstream reports disagree. How would you establish explainable sources of truth?
- A quality check occasionally produces false alarms and blocks downstream jobs; removing it once let bad data reach settlement. How would you design tiered handling?
- An enterprise customer has multiple organizations, departments, and members, and a member may act for different organizations. When designing customer activity analysis, how would you define the grain and relationships among customer, account, organization, and user?
- Support agents merge duplicate tickets and sometimes split a complex issue into several cases. When modeling service-quality data, how would you define the issue, ticket, contact, and resolution?
- Partner settlement must connect external order IDs, internal orders, payments, and refunds, but systems use different keys and update times. How would you build the mappings and reconciliation process?
No questions in this section match your target. ·
Questions 17–24
- Business units use an “active customer” metric, but some define it by payment, others by usage, and others by likelihood to renew. How would you decide which definitions should be standardized and which should remain distinct concepts?
- Business reports mix orders, order lines, and payment attempts, so metrics disagree across pages. How would you establish fact-table grain and validate aggregation semantics?
- The source database has deleted a customer, but a change data capture (CDC) pipeline and downstream summary views still show the old record. How would you trace the deletion event, update dependent views, and verify cleanup?
- A dashboard's total conversion rate cannot explain which customer segment or process step caused a change; teams want to drill down by channel, plan, and region, but dimensions have different grains. How would you define fact grain and valid aggregation rules?
- Business reporting needs the duration of every state transition, while an operations page needs only the object's current state; a wide table overwrote history during updates. How would you separate event facts, snapshots, and current-state models?
- CRM, billing, and product logs use customer IDs, contract IDs, and anonymous device identifiers; analysts need one customer's lifecycle. How would you model identity mappings, confidence, temporal validity, and correction?
- Several teams maintain customer profiles. When data is wrong, no one knows which system is authoritative, and systems also write values back to one another. How would you define field ownership, source of truth, synchronization direction, and dispute handling?
- Orders contain nested line items, shipping, and tax, while refunds may apply to only some items; joining order headers directly to lines doubles totals in existing reports. How would you model grain and aggregation semantics?
No questions in this section match your target. ·
Questions 25–32
- Sales performance rolls up through a regional hierarchy. Midyear, a region is split; historical reports must remain unchanged, while future reports use the new hierarchy. How would you represent organization versions, effective time, and attribution?
- Some transaction details must be retained for years under policy, while product analytics wants to use them for user segmentation and experiments. How would you model purpose authorization, retention, and access views?
- A user requests personal-data deletion, but the system uses append-only events, search indexes, and offline analytics copies; some events must remain for business audit. How would you distinguish deletable fields, identity mappings, cryptographic erasure, and retention exceptions?
- Revenue was recognized for an order last month, and a partial refund occurs this month. Finance and growth disagree on order count, net revenue, and refund-month treatment. How would you establish consistent data semantics?
- A service rebuilds state from an event log, but the live snapshot differs and responders cannot tell which to repair. How would you define authority and the repair process?
- Payment is authorized but not captured, and goods are not fulfilled. The business report counts an order, while finance records no revenue yet. How would you design cross-domain metric semantics?
- A payment channel charges a fee on partial refunds, and the contract assigns it to the merchant, but reports combine refund amount and fee. How would you model this?
- A user belongs to several organizations. The browser retains prior-workspace cache, and old organization data briefly appears after switching. How would you design the boundary?
No questions in this section match your target. ·
Questions 33–34
- Operations counts revenue when an order is placed, while finance recognizes it after fulfillment and refunds. A leader wants one number. How would you explain the two measurement methods and design shared measures for decisions?
- Several teams manually corrected an aggregate table, and later no one could explain why the number changed. How would you record sources, transformations, versions, owners, and recomputation evidence?
No questions in this section match your target. ·
No questions in this capability match your target. ·