You have 3 weeks to elicit requirements for a customer onboarding redesign involving 12 stakeholders across sales, compliance, operations, and engineering. How would you plan the elicitation?

I would sequence discovery by risk and dependency rather than schedule 12 disconnected conversations. - In days 1 and 2, I would map each stakeholder to the onboarding decisions, evidence, and systems they own, then mark compliance and operations as mandatory inputs. - I would review the current forms, policy documents, funnel data, and 20 recent applications before interviews so questions test known gaps instead of asking people to restate the process. - I would run role-specific interviews in week 1, a cross-functional playback in week 2, and reserve week 3 for unresolved rules, acceptance criteria, and sponsor approval.

A procurement portal must be scoped in 10 working days with 10 stakeholders whose influence and daily involvement vary widely. How would you use a power-interest map to focus your analysis?

I would use the map to set the depth and cadence of engagement, not to exclude low-power users. - The procurement director and finance controller have high power, so I would validate objectives, budget rules, and final decisions with them twice during the 10 days. - Buyers and approvers have high interest but less formal power, so I would observe 6 real purchase requests and test detailed workflows with them. - I would record an owner and contact frequency for every quadrant, then revisit the map when a stakeholder gains decision authority or a hidden dependency appears.

Fourteen people from product, legal, finance, support, and engineering are contributing to a billing feature due for design approval in 2 weeks. How would you create a useful RACI?

I would facilitate a RACI around 8 to 12 decisions and deliverables, not assign authority myself or map every project task. - For pricing rules, refunds, invoice content, data retention, and release acceptance, I would propose one Accountable candidate and ask the sponsor or process-governance owner to confirm that authority; the people doing the analysis or implementation would be Responsible. - I would review overloaded rows, especially any with two Accountable people or all 14 marked Consulted, because those patterns predict delayed decisions. - I would test the matrix in a 45-minute session using two real decisions; if the gap is decision rights rather than delivery responsibility, I would use DACI or RAPID instead and record its decision process.

You have 15 business days to analyze a call-center returns process used by 80 agents, 6 supervisors, and 3 policy owners. How would you choose between workshops, interviews, and observation?

I would combine the techniques because each exposes a different part of the process. - I would observe 8 agents handling at least 24 calls to capture workarounds, screen switching, and steps that policy owners may not know about. - I would interview the 3 policy owners separately to clarify exceptions and controls that agents cannot decide in a group workshop. - I would use a workshop with agents and supervisors to validate the as-is map and prioritize gaps, then confirm disputed details against recordings and system timestamps before day 15.

A partner onboarding product has 11 stakeholders and must freeze its core requirements in 3 weeks, but nobody can say who decides identity checks, commercial terms, or support policy. What would you do?

I would establish decision rights before asking the group to approve detailed requirements. - I would create a decision register for identity checks, contract eligibility, service levels, support ownership, and exceptions, naming one decision owner and a due date for each. - I would ask the sponsor to resolve unowned or multiply owned decisions within 48 hours, presenting the delivery impact of leaving each one open. - Every workshop conclusion would be recorded as a decision, assumption, or open question, so silence is never mistaken for approval at the 3-week baseline.

You must elicit a shared order-management workflow from 15 stakeholders in 5 time zones within 4 weeks, and no meeting slot reaches more than 9 people. How would you work remotely?

I would design an asynchronous-first cycle with short regional sessions and one controlled source of truth. - I would send a 10-minute process walkthrough and focused questions 48 hours before each session, then collect comments directly on the same BPMN model in Miro or Confluence. - I would run two mirrored workshops, publish one decision log within 24 hours, and give absent stakeholders a 2-business-day review window. - For the 5 cross-region decisions, I would propose candidate decision owners for sponsor confirmation, then schedule a final 60-minute overlap session rather than reopen every local detail.

A regulated lending workflow serving 30,000 applications per month needs requirements ready in 4 weeks, with input from 9 stakeholders and mandatory audit evidence. How would you adapt elicitation?

I would treat regulations and control evidence as primary requirement sources alongside stakeholder interviews. - In week 1, I would map each policy clause to the application step, data captured, responsible role, retention period, and evidence an auditor must be able to retrieve. - I would validate interpretations with compliance and legal, while operations demonstrates 10 real anonymized cases including approvals, declines, and manual reviews. - I would make every regulatory rule traceable to a requirement and test, and require named compliance approval before the 4-week baseline.

A warehouse picking redesign has 8 managers in discovery but affects 120 pickers on 3 shifts, and findings are due in 12 days. How would you avoid capturing only management's view?

I would sample frontline evidence across shifts and compare it with management assumptions. - I would observe 3 pickers per shift through complete picks and interview them at the workstation, including one new hire and one experienced picker on each shift. - I would compare observed travel time, scan failures, and exception handling with the standard operating procedure and 4 weeks of system data. - I would play back discrepancies to managers and pickers together, labeling whether each requirement comes from policy, measured behavior, or a proposed improvement.

After 3 weeks of discovery, a claims product has 120 draft requirements from 13 stakeholders and must enter solution design in 5 days. How would you assess completeness?

I would assess completeness against the product scope and models, not by counting documents or approvals. - I would walk every actor, trigger, main flow, alternate flow, business rule, data item, interface, permission, and measurable NFR through a coverage checklist. - I would trace the 120 requirements to 6 business objectives and 18 process steps, flagging objectives or steps with no supporting requirement or acceptance test. - I would run scenario reviews for a normal claim and the 5 highest-risk exceptions, then classify remaining gaps as design blockers, explicit assumptions, or later-scope items before day 5.

A vendor will build a 6-month case-management solution, and you have 3 weeks to provide requirements covering 4 departments and 2 external interfaces. Would you use a BRD, an FRD, or both?

I would use a concise BRD for the agreed business scope and a linked functional specification for build detail. - The BRD would fix the 5 objectives, in-scope departments, target outcomes, policy constraints, and exclusions so the vendor cannot interpret features as the business goal. - The FRD would describe workflows, rules, data, roles, interface behavior, and measurable NFRs at the level needed for estimation and acceptance. - I would avoid duplicating text by giving every functional requirement an ID linked back to its BRD objective and forward to acceptance scenarios.

An omnichannel returns feature has 4 user roles, 7 exception paths, and a design deadline in 10 working days. Would you model it with user stories, use cases, or both?

I would use a use case to control the end-to-end interactions and stories to slice delivery. - The use case would capture actors, preconditions, the normal return, 7 alternate paths, and postconditions across store, web, warehouse, and payment systems. - I would derive stories by valuable increments such as standard refund, exchange, damaged item, and manager override, each with acceptance examples. - I would keep the use case as the shared behavior model and link stories to its flows, avoiding two independently maintained versions of the same rule.

A discount engine has 6 customer segments, 4 basket thresholds, and 3 promotion types, and engineering needs unambiguous rules in 7 days. How would you model them?

I would use a decision table to expose every meaningful condition combination and its outcome. - Rows would cover segment, basket threshold, promotion eligibility, stacking permission, and date validity, with actions for discount amount and rejection reason. - Based on the agreed stacking rule, I would choose an explicit DMN hit policy: Unique when exactly one rule may match, Priority when one outcome outranks another, or Collect when eligible promotions may stack; I would then test gaps and overlaps against that policy. - Product, finance, and QA would review examples at the boundaries, such as exactly $100 and one cent below, before the table becomes the implementation reference.

A loan application can move through 9 statuses with automated and manual transitions, and the workflow must be specified in 2 weeks. How would you build a state model?

I would define states by stable business meaning and make every transition explicit. - For each of the 9 states, I would document allowed entry events, permitted next states, responsible actor, data changes, and whether the customer can still edit the application. - I would challenge ambiguous statuses such as pending by separating waiting for customer, waiting for bureau, and manual review when their rules or timers differ. - I would test the model with a standard approval, a decline, a withdrawal, and two resubmission paths, then link transitions to acceptance scenarios.

Three source systems use different names and formats for 35 customer fields, and a unified profile specification is due in 8 working days. How would you create the data dictionary?

I would make the dictionary the agreed business definition and mapping for each field. - Each of the 35 entries would include business name, precise definition, source, type and format, allowed values, nullability, owner, sensitivity, and transformation rule. - I would resolve terms such as customer, active, and primary address with examples because naming differences often hide different populations or time rules. - I would validate 20 sampled records through the mappings and record unresolved fields as assumptions with owners and dates rather than silently choosing one source.

A booking product must exchange customer, inventory, payment, and confirmation data with 4 systems, and interface requirements are due in 10 days. What would you specify?

I would specify the business exchange and failure expectations for every interface before discussing implementation details. - For each system, I would define trigger, sender and receiver, required fields, business validations, expected response, frequency or volume, and the authoritative source. - I would document duplicate, late, missing, and rejected messages with visible business statuses and ownership for resolution. - I would validate one successful booking and 6 interface exceptions as sequence diagrams or scenarios, then obtain agreement from product, operations, and each system owner.

A checkout redesign must support 3,000 concurrent users before a campaign launch in 6 weeks, but stakeholders wrote only 'fast and reliable.' How would you define the NFRs?

I would replace adjectives with thresholds tied to user journeys and business risk. - I would define p95 page or API response times for basket, payment initiation, and confirmation against a workload model covering 3,000 concurrent users, arrival rate, journey mix, and think time, with a separate maximum for the payment provider. - Availability would become a percentage and measurement window, such as 99.95% monthly excluding agreed maintenance, plus recovery time and recovery point targets. - I would confirm how each threshold and the workload model will be measured in a campaign-like load test and identify who accepts a missed threshold before launch.

A healthcare portal for 25,000 patients must define privacy and access requirements within 3 weeks. How would you make security requirements measurable from a BA perspective?

I would translate privacy obligations into observable access, audit, and retention behavior. - I would define which of the 5 roles can view, create, amend, export, or delete each data category, including emergency access and approval rules. - I would define an audited-event catalog covering view, create, amend, export, delete, emergency access, and permission changes; 100% of those events would record actor, patient, action, timestamp, outcome, and reason where required, with an explicit policy-approved retention period and search SLA. - I would specify session timeout, failed-login lockout, consent withdrawal, and data deletion timelines as acceptance scenarios reviewed by privacy, security, and operations before week 3.

A release contains 75 requirements tied to 6 objectives and 110 planned tests, and scope review is in 5 days. How would you build traceability from objective to test?

I would create bidirectional links that expose missing value, missing coverage, and unnecessary work. - Each objective would link to a measurable outcome, then to requirement IDs, design or story references, and one or more acceptance tests. - I would flag requirements with no objective, objectives with no requirements, requirements with no test, and tests that do not prove any requirement. - The review would focus first on high-risk many-to-many links and changed requirements, with owners assigned to close every blocker before day 5.

A subscription product has 5 epics, 60 stories, 14 business rules, and a refinement deadline in 8 days, but reviewers keep finding contradictions. How would you structure the requirements models?

I would separate concerns into linked models and use scenario walkthroughs to test consistency. - A context and process model would set boundaries and handoffs, while state, decision, data, and permission models would each own one type of rule. - Stories would reference those shared rules instead of copying eligibility or status logic into 60 separate descriptions. - I would walk 8 representative scenarios through every model, recording contradictions in a single issue list with an owner and resolution date before refinement closes.

A claims process crosses 5 teams, has a 48-hour document SLA, and includes 6 exception types; the BPMN model is due in 7 working days. How would you model boundary events?

I would attach boundary events to the activity whose normal execution they interrupt or supplement. - An interrupting timer on 'Wait for documents' would route overdue claims to an escalation path after 48 hours and make the original wait stop. - A non-interrupting message boundary event could capture an additional document while assessment continues, if policy allows concurrent handling. - I would model the 6 exceptions only where their triggers and outcomes differ, then validate token behavior with operations so the diagram does not create duplicate active work.

Business Analyst interview questions

100 real questions with model answers and explanations for Business Analyst candidates.

See a Business Analyst resume example →

Practice with flashcards

Spaced repetition · Hunter Pass

Questions

stakeholder-managementrequirements-elicitationonboarding

I would sequence discovery by risk and dependency rather than schedule 12 disconnected conversations.

In days 1 and 2, I would map each stakeholder to the onboarding decisions, evidence, and systems they own, then mark compliance and operations as mandatory inputs.
I would review the current forms, policy documents, funnel data, and 20 recent applications before interviews so questions test known gaps instead of asking people to restate the process.
I would run role-specific interviews in week 1, a cross-functional playback in week 2, and reserve week 3 for unresolved rules, acceptance criteria, and sponsor approval.

Why interviewers ask this: The interviewer is assessing whether the candidate can turn a crowded stakeholder landscape and a fixed deadline into a risk-based elicitation plan.

stakeholder-managementstakeholder-mappingcommunication

I would use the map to set the depth and cadence of engagement, not to exclude low-power users.

The procurement director and finance controller have high power, so I would validate objectives, budget rules, and final decisions with them twice during the 10 days.
Buyers and approvers have high interest but less formal power, so I would observe 6 real purchase requests and test detailed workflows with them.
I would record an owner and contact frequency for every quadrant, then revisit the map when a stakeholder gains decision authority or a hidden dependency appears.

Why interviewers ask this: A strong answer shows that stakeholder mapping changes concrete elicitation activities while still protecting the voice of operational users.

racidecision-rightsdesign

I would facilitate a RACI around 8 to 12 decisions and deliverables, not assign authority myself or map every project task.

For pricing rules, refunds, invoice content, data retention, and release acceptance, I would propose one Accountable candidate and ask the sponsor or process-governance owner to confirm that authority; the people doing the analysis or implementation would be Responsible.
I would review overloaded rows, especially any with two Accountable people or all 14 marked Consulted, because those patterns predict delayed decisions.
I would test the matrix in a 45-minute session using two real decisions; if the gap is decision rights rather than delivery responsibility, I would use DACI or RAPID instead and record its decision process.

Why interviewers ask this: The interviewer wants evidence that the candidate can use RACI to remove decision ambiguity rather than produce a decorative stakeholder table.

facilitationstakeholder-interviewsconcurrency

I would combine the techniques because each exposes a different part of the process.

I would observe 8 agents handling at least 24 calls to capture workarounds, screen switching, and steps that policy owners may not know about.
I would interview the 3 policy owners separately to clarify exceptions and controls that agents cannot decide in a group workshop.
I would use a workshop with agents and supervisors to validate the as-is map and prioritize gaps, then confirm disputed details against recordings and system timestamps before day 15.

Why interviewers ask this: The interviewer is checking whether the candidate selects elicitation techniques from the information needed rather than defaulting to meetings.

stakeholder-managementonboardingcommunication

I would establish decision rights before asking the group to approve detailed requirements.

I would create a decision register for identity checks, contract eligibility, service levels, support ownership, and exceptions, naming one decision owner and a due date for each.
I would ask the sponsor to resolve unowned or multiply owned decisions within 48 hours, presenting the delivery impact of leaving each one open.
Every workshop conclusion would be recorded as a decision, assumption, or open question, so silence is never mistaken for approval at the 3-week baseline.

Why interviewers ask this: This tests whether the candidate can make decision ownership explicit early enough to keep analysis moving.

stakeholder-managementcommunicationworkflow-analysis

I would design an asynchronous-first cycle with short regional sessions and one controlled source of truth.

I would send a 10-minute process walkthrough and focused questions 48 hours before each session, then collect comments directly on the same BPMN model in Miro or Confluence.
I would run two mirrored workshops, publish one decision log within 24 hours, and give absent stakeholders a 2-business-day review window.
For the 5 cross-region decisions, I would propose candidate decision owners for sponsor confirmation, then schedule a final 60-minute overlap session rather than reopen every local detail.

Why interviewers ask this: A strong answer demonstrates a concrete remote elicitation rhythm that prevents attendance limits from creating conflicting requirement sets.

stakeholder-managementrequirements-elicitationrequirements

I would treat regulations and control evidence as primary requirement sources alongside stakeholder interviews.

In week 1, I would map each policy clause to the application step, data captured, responsible role, retention period, and evidence an auditor must be able to retrieve.
I would validate interpretations with compliance and legal, while operations demonstrates 10 real anonymized cases including approvals, declines, and manual reviews.
I would make every regulatory rule traceable to a requirement and test, and require named compliance approval before the 4-week baseline.

Why interviewers ask this: The interviewer is evaluating whether the candidate can preserve regulatory traceability without reducing elicitation to policy transcription.

warehouse

I would sample frontline evidence across shifts and compare it with management assumptions.

I would observe 3 pickers per shift through complete picks and interview them at the workstation, including one new hire and one experienced picker on each shift.
I would compare observed travel time, scan failures, and exception handling with the standard operating procedure and 4 weeks of system data.
I would play back discrepancies to managers and pickers together, labeling whether each requirement comes from policy, measured behavior, or a proposed improvement.

Why interviewers ask this: The interviewer wants to see a practical sampling strategy that exposes real work without expanding discovery beyond the deadline.

stakeholder-managementdesigncommunication

I would assess completeness against the product scope and models, not by counting documents or approvals.

I would walk every actor, trigger, main flow, alternate flow, business rule, data item, interface, permission, and measurable NFR through a coverage checklist.
I would trace the 120 requirements to 6 business objectives and 18 process steps, flagging objectives or steps with no supporting requirement or acceptance test.
I would run scenario reviews for a normal claim and the 5 highest-risk exceptions, then classify remaining gaps as design blockers, explicit assumptions, or later-scope items before day 5.

Why interviewers ask this: A strong answer separates evidence of requirement coverage from the false assurance of a large requirement count.

requirementsrequirements-documentsfunctional-specification

I would use a concise BRD for the agreed business scope and a linked functional specification for build detail.

The BRD would fix the 5 objectives, in-scope departments, target outcomes, policy constraints, and exclusions so the vendor cannot interpret features as the business goal.
The FRD would describe workflows, rules, data, roles, interface behavior, and measurable NFRs at the level needed for estimation and acceptance.
I would avoid duplicating text by giving every functional requirement an ID linked back to its BRD objective and forward to acceptance scenarios.

Why interviewers ask this: The interviewer is testing whether the candidate chooses documentation from contract and delivery needs instead of treating BRD and FRD as mandatory templates.

user-storiesuse-casesdesign

I would use a use case to control the end-to-end interactions and stories to slice delivery.

The use case would capture actors, preconditions, the normal return, 7 alternate paths, and postconditions across store, web, warehouse, and payment systems.
I would derive stories by valuable increments such as standard refund, exchange, damaged item, and manager override, each with acceptance examples.
I would keep the use case as the shared behavior model and link stories to its flows, avoiding two independently maintained versions of the same rule.

Why interviewers ask this: A strong answer matches each artifact to a delivery purpose and explains how to prevent duplicate specifications.

I would use a decision table to expose every meaningful condition combination and its outcome.

Rows would cover segment, basket threshold, promotion eligibility, stacking permission, and date validity, with actions for discount amount and rejection reason.
Based on the agreed stacking rule, I would choose an explicit DMN hit policy: Unique when exactly one rule may match, Priority when one outcome outranks another, or Collect when eligible promotions may stack; I would then test gaps and overlaps against that policy.
Product, finance, and QA would review examples at the boundaries, such as exactly $100 and one cent below, before the table becomes the implementation reference.

Why interviewers ask this: The interviewer is checking whether the candidate can turn dense business rules into a complete and testable model.

workflow-analysis

I would define states by stable business meaning and make every transition explicit.

For each of the 9 states, I would document allowed entry events, permitted next states, responsible actor, data changes, and whether the customer can still edit the application.
I would challenge ambiguous statuses such as pending by separating waiting for customer, waiting for bureau, and manual review when their rules or timers differ.
I would test the model with a standard approval, a decline, a withdrawal, and two resubmission paths, then link transitions to acceptance scenarios.

Why interviewers ask this: A strong answer shows that a state model controls behavior, permissions, and exceptions rather than merely listing labels.

data-dictionarysystem-design

I would make the dictionary the agreed business definition and mapping for each field.

Each of the 35 entries would include business name, precise definition, source, type and format, allowed values, nullability, owner, sensitivity, and transformation rule.
I would resolve terms such as customer, active, and primary address with examples because naming differences often hide different populations or time rules.
I would validate 20 sampled records through the mappings and record unresolved fields as assumptions with owners and dates rather than silently choosing one source.

Why interviewers ask this: The interviewer is assessing whether the candidate can make cross-system data meanings precise enough for design and validation.

requirementssystem-designtypes

I would specify the business exchange and failure expectations for every interface before discussing implementation details.

For each system, I would define trigger, sender and receiver, required fields, business validations, expected response, frequency or volume, and the authoritative source.
I would document duplicate, late, missing, and rejected messages with visible business statuses and ownership for resolution.
I would validate one successful booking and 6 interface exceptions as sequence diagrams or scenarios, then obtain agreement from product, operations, and each system owner.

Why interviewers ask this: A strong answer covers the information and business behavior at system boundaries without drifting into low-level engineering design.

stakeholder-managementcommunicationconcurrency

I would replace adjectives with thresholds tied to user journeys and business risk.

I would define p95 page or API response times for basket, payment initiation, and confirmation against a workload model covering 3,000 concurrent users, arrival rate, journey mix, and think time, with a separate maximum for the payment provider.
Availability would become a percentage and measurement window, such as 99.95% monthly excluding agreed maintenance, plus recovery time and recovery point targets.
I would confirm how each threshold and the workload model will be measured in a campaign-like load test and identify who accepts a missed threshold before launch.

Why interviewers ask this: The interviewer is evaluating whether the candidate can make non-functional expectations measurable, testable, and commercially meaningful.

requirements

I would translate privacy obligations into observable access, audit, and retention behavior.

I would define which of the 5 roles can view, create, amend, export, or delete each data category, including emergency access and approval rules.
I would define an audited-event catalog covering view, create, amend, export, delete, emergency access, and permission changes; 100% of those events would record actor, patient, action, timestamp, outcome, and reason where required, with an explicit policy-approved retention period and search SLA.
I would specify session timeout, failed-login lockout, consent withdrawal, and data deletion timelines as acceptance scenarios reviewed by privacy, security, and operations before week 3.

Why interviewers ask this: A strong answer turns broad security language into business-visible controls and verifiable outcomes.

requirementstestingrequirements-traceability

I would create bidirectional links that expose missing value, missing coverage, and unnecessary work.

Each objective would link to a measurable outcome, then to requirement IDs, design or story references, and one or more acceptance tests.
I would flag requirements with no objective, objectives with no requirements, requirements with no test, and tests that do not prove any requirement.
The review would focus first on high-risk many-to-many links and changed requirements, with owners assigned to close every blocker before day 5.

Why interviewers ask this: The interviewer is checking whether the candidate uses traceability to answer practical coverage questions rather than maintain a passive spreadsheet.

requirementsbusiness-rulesestimation

I would separate concerns into linked models and use scenario walkthroughs to test consistency.

A context and process model would set boundaries and handoffs, while state, decision, data, and permission models would each own one type of rule.
Stories would reference those shared rules instead of copying eligibility or status logic into 60 separate descriptions.
I would walk 8 representative scenarios through every model, recording contradictions in a single issue list with an owner and resolution date before refinement closes.

Why interviewers ask this: A strong answer demonstrates model-based consistency control rather than trying to solve contradictions with more prose.

concurrencyerror-handlingprocess-modeling

I would attach boundary events to the activity whose normal execution they interrupt or supplement.

An interrupting timer on 'Wait for documents' would route overdue claims to an escalation path after 48 hours and make the original wait stop.
A non-interrupting message boundary event could capture an additional document while assessment continues, if policy allows concurrent handling.
I would model the 6 exceptions only where their triggers and outcomes differ, then validate token behavior with operations so the diagram does not create duplicate active work.

Why interviewers ask this: The interviewer is assessing whether the candidate understands boundary event behavior well enough to model real exceptions accurately.

Locked questions

21
An online application waits for either customer documents, a cancellation request, or a 5-day timeout, and the process design is due tomorrow. How would you use an event-based gateway?
designgatewayresilience
22
A fulfillment process has 42 activities repeated across 3 regions, and stakeholders need a readable BPMN model in 6 days. How would you use subprocesses?
stakeholder-managementcommunicationconcurrency
23
A marketplace sends 20,000 orders per day to 3 logistics partners and receives asynchronous accept, reject, and delivery messages; the interface model is due in 2 weeks. How would you model message correlation?
asynctypescorrelation
24
You have 2 weeks to document an as-is service process performed by 40 agents across 2 sites, but the procedure manual is 18 months old. What evidence would you collect?
concurrency
25
Returns take 9 days against a 3-day target, and managers blame slow agents; you have 8 working days to analyze 500 recent cases. How would you separate root cause from symptom?
root-cause-analysis
26
A 22-step supplier onboarding process must reach a 5-day cycle time from its current 12 days, and you have 10 days to complete a gap analysis. How would you proceed?
gap-analysisonboardingconcurrency
27
A mortgage review has 12 handoffs, 4 approval queues, and 18% rework, and a Lean improvement workshop is scheduled in 1 week. How would you identify waste beforehand?
facilitationdata-structures
28
A verification team receives 600 applications per 8-hour day, each requiring 9 minutes of work, and demand will rise 20% in 6 weeks. How would you assess capacity and queue risk?
capacitydata-structures
29
A refund process crosses 7 teams, handles $2 million per month, and needs a controlled to-be design in 3 weeks. How would you analyze controls and handoffs?
formsdesignconcurrency
30
You have 3 days to explain why 12% of 80,000 customers with paid orders have no onboarding call, using customer, order, and call tables. How would you structure the SQL analysis?
onboardingsql
31
A 7-stage sales funnel contains 2 million events, and leadership needs a source-level conversion analysis by Friday. How would you use CTEs while keeping the logic reviewable?
funnelfunnel-analysis
32
Support wants a 10-day analysis of response times for 300,000 tickets, including each team's weekly trend and the slowest 10% of cases. Where would window functions help?
window-functions
33
A subscription product needs a 6-month retention view for 24,000 new customers before next week's planning session. How would you define the cohort analysis?
retentioncohortssessions
34
A checkout funnel has 500,000 sessions per month and 5 tracked stages, and product needs drop-off findings in 5 working days. How would you build a trustworthy funnel?
funnelfunnel-analysissessions
35
CRM, billing, and support contain 1.2 million customer records with no shared ID, and a duplicate estimate is due in 2 weeks. How would you define duplicates at BA level?
estimation
36
The definition of a premium customer changed from $5,000 annual spend to $7,500 on July 1, affecting 18 months of reports that must be reconciled in 7 days. How would you handle the changing definition?
data-reconciliation
37
ERP revenue is 2.4% higher than the CRM dashboard for the same quarter, covering 160,000 transactions, and finance needs a reconciled answer in 4 days. How would you investigate?
data-reconciliationtransactions
38
Operations wants to publish a 'first-contact resolution' KPI for 50,000 monthly cases in 2 weeks. How would you validate the KPI and its data-quality rules?
kpi-definitionvalidation
39
Regional managers need a Power BI or Tableau sales dashboard for 6 regions and 4 product lines in 4 weeks. What requirements would you define before development?
birequirements
40
You have 2 weeks to test a new insurance-claim workflow with 6 user roles before engineering estimates it. What prototype fidelity would you choose?
prototypingprototypesestimation
41
A scheduling prototype covers a 12-step workflow for dispatchers, technicians, and customers, and you have 5 days to validate it. How would you run the validation?
prototypingprototypesjobs
42
A case-management product has 4 roles and 25 planned screens, and permission requirements must be ready in 8 working days. How would you validate permissions in wireframes?
design-collabrequirementsvalidation
43
A payment flow has 12 screens and must be handed to design in 6 days, but current wireframes show only the happy path. Which error and edge states would you add?
wireframesdesigndesign-collab
44
Ten warehouse users must validate a tablet picking prototype across 3 device sizes before build starts in 7 days. How would you turn feedback into requirement decisions?
prototypingprototypesfeedback
45
A partner API must support 3 order actions and 8 business validation failures, and its contract review is in 10 days. What would you specify from the business viewpoint?
validationapi
46
A nightly supplier file carries 50,000 records and 32 fields into a procurement system, and interface requirements must be approved in 2 weeks. How would you define the contract?
requirementssystem-designtypes
47
You must recommend one of 3 case-management options within 8 working days: configure the current tool, buy a SaaS product, or build a custom module. How would you score them?
configcloud
48
A self-service portal is expected in 6 weeks with a $180,000 budget, 2 engineers, and 5 external dependencies. How would you validate feasibility assumptions before confirming scope?
feasibility-analysisdependenciesvalidation
49
A release has 45 functional requirements across 6 workflows and must agree its acceptance model in 1 week. How would you structure acceptance?
requirementsworkflow-analysis
50
After 4 weeks of analysis, 95 requirements for a payroll feature must be baselined by Friday so 2 delivery teams can estimate consistently. What would the baseline contain?
requirementsestimation
51
Sales wants phone number collection to be mandatory, while Support says it should remain optional; the current form has 18% abandonment across 12,000 monthly visits. How would you resolve the conflicting requirements?
requirementsforms
52
Three department heads propose 4 different approval flows for a purchase request, and development needs one decision within 2 days. How would you facilitate it?
53
The decision owner is unavailable for 5 working days, but a team of 6 developers will be blocked in 48 hours by an unresolved refund-rule question. What do you do?
54
A stakeholder adds 6 new requests after a 24-point sprint has started, and the team has only 3 points of spare capacity. How do you handle the scope creep?
soft-skillscommunicationscope
55
A change request adds 12 customer fields and a new validation service 3 weeks before release; the current forecast is 8 developer-days and 5 tester-days. How would you assess it?
change-controlvalidation
56
Operations ranks a 20% reduction in handling time first, while Customer Service ranks a 15% reduction in repeat contacts first; the next release can fit only 12 of 20 backlog points. How do you settle the priority dispute?
backlog-managementbacklog
57
A customer campaign fixes the launch date at 4 weeks, but the approved scope is estimated at 28 points and the team's demonstrated capacity is 20 points. What would you recommend?
estimationcapacity
58
An external address API is now 10 days late, and your release is due in 3 weeks with 14 stories depending on it. How do you manage the dependency conflict?
dependenciesapi
59
Team A can deliver a new order status in sprint 7, but Team B freezes its reporting release after sprint 6; 9 reports depend on that status. What do you recommend?
agile
60
You have 25 UAT cases, 5 business testers, and 3 working days. How would you define ownership and execution?
ownershipuat
61
A release has 80 UAT cases but capacity for only 45 before the deadline; 6 cases cover payment and refund paths. How would you set the strategy?
estimationcapacityuat
62
Forty UAT cases require customer profiles, but the test environment contains only 12 usable profiles and none has a failed-payment history. What do you do?
test-environmentsuat
63
The UAT environment fails with timeouts in 9 of 30 cases, while the same steps pass in the system test environment. How would you respond?
system-designresiliencetest-environments
64
UAT produces 18 defects with only 2 days before the release meeting: 3 stop checkout, 7 have workable workarounds, and 8 are cosmetic. How do you triage them?
defectsuat
65
During UAT, 7 of 50 cases fail because users expect bulk cancellation, but the signed requirement describes only single-order cancellation. Is this a defect or a requirement gap?
defectsuatresilience
66
Fourteen defects were fixed after UAT round 1, and the regression suite contains 60 cases with 2 tester-days left. What would you rerun?
regressiondefectsuat
67
After 72 of 75 UAT cases pass, Operations refuses sign-off over 3 failed reports, while the product owner says the release can proceed. How do you handle the disagreement?
soft-skillsconflictuat
68
In UAT, 16 of 40 executed cases lack the build ID, tester, and evidence. How would you recover a trustworthy acceptance record?
uat
69
Forty-five UAT cases were executed against the wrong release build and all passed. What do you do before go/no-go?
governanceuat
70
A cross-team epic spans 3 products, 4 teams, and a proposed 6-week delivery window, but refinement keeps producing different assumptions. How would you refine it?
71
A 24-point epic for online returns is too large for a sprint, and the team proposes splitting it into UI, API, and database stories. How would you split it?
agileapidatabase
72
A story depends on an unfamiliar vendor API, and estimates range from 5 to 18 points because nobody knows its rate limits or error behavior. What do you recommend?
rate-limitingestimationprocurement
73
A feature crosses 4 teams and has 11 identified dependencies, but the plan lists only target dates. How would you make the dependency map actionable?
dependencies
74
A story estimate changes from 8 to 21 points across 3 refinement sessions. How would you diagnose the churn?
churnestimationsessions
75
On day 7, the acceptance criteria contain 4 mutually incompatible cancellation rules from different sources. What do you do?
specsresilienceacceptance-criteria
76
Two weeks after release, 35 users report that a new search takes too many clicks, but analytics show completion improved from 62% to 70%. How would you act on the feedback?
feedback
77
A vendor withdraws the API that underpins the Sprint Goal with 4 days left. How do you respond?
agileprocurementapi
78
A vendor requires a signed specification 4 weeks before build, while your internal team works in 2-week sprints and expects requirements to evolve. How would you manage the hybrid handoff?
agileprocurementrequirements
79
A BPMN model says every refund receives manager approval, but logs for 500 refunds show 22% skipped that step. How would you debug the discrepancy?
process-modeling
80
A support process has a 4-hour SLA, but 31% of 2,400 monthly tickets miss it even though average handling time is 2.1 hours. How would you investigate?
concurrency
81
An event retry creates duplicate shipment notifications for 1,800 of 60,000 records. How would you investigate and correct the behavior?
resilience
82
An order marked Shipped in the warehouse appears as Processing in the customer portal for 14% of 3,000 daily orders. How would you debug the status mapping?
warehouseconcurrency
83
The API accepts discounts up to 30%, but the approved business rule caps agents at 20%; 46 orders exceeded the cap last week. What would you do?
business-rulesapi
84
Six API clients receive HTTP 400 without a business error code for 14 validation rules and retry each request 3 times. How would you improve the contract?
httpresiliencevalidation
85
Customer onboarding time rose from 12 to 19 minutes after a release, but no single step shows an obvious error across 600 sessions. How would you find the root cause?
root-cause-analysisonboardingsessions
86
A manual refund process produced 17 amount-entry errors in 3 months across 4 agents. What control would you design?
designconcurrency
87
A nightly file integration sends 10,000 rows, but the target loads only 9,940 and reports no error for the missing 60. How would you debug and prevent recurrence?
88
A signup funnel drops from 68% completion to 49% at the identity step after a release, affecting about 4,000 weekly visitors. How would you investigate?
funnelfunnel-analysis
89
Marketing defines a retained customer as active within 30 days of signup, while Product uses any activity in calendar month 2; their reports differ by 11%. How would you resolve the cohort conflict?
cohorts
90
A dashboard week starts Sunday while Finance uses Monday, shifting 374 orders across months. How would you resolve the reporting mismatch?
91
Overall conversion rises from 40% to 43%, yet conversion falls in both mobile and desktop segments across 20,000 visits. How would you explain and investigate this?
92
Agents are measured on closing 35 tickets per day, and reopen rates have climbed from 8% to 17% in 6 weeks. What would you recommend?
93
A new scheduling feature targets 60% weekly adoption, but only 34% of 1,200 eligible users use it after 5 weeks. How would you diagnose the gap?
decision-makingjobs
94
An A/B test with 18,000 users raises checkout completion from 71.0% to 72.2%, but average order value falls by 2.5%. How would you interpret it?
ab-testingrevenue
95
Two days before a promised launch, analysis shows the core flow fails for 18% of mobile users and needs about 5 developer-days to fix. How would you communicate the bad news?
communicationpromises
96
In a 90-minute workshop, 2 managers repeatedly interrupt each other and the group has resolved only 1 of 6 requirements after 40 minutes. What do you do?
requirementsfacilitation
97
A junior analyst produces a 14-page requirement document for a 3-story feature, but developers still ask who can cancel an order and when. How would you mentor them?
mentoring
98
A vendor's 38-page proposal says customer records are synchronized regularly, but your process requires updates within 15 minutes and the contract is due in 3 days. What do you do?
procurementconcurrency
99
You have 5 working days to hand over a feature with 42 requirements, 9 open decisions, and 6 active defects to another analyst. How would you structure the handover?
requirementsdefects
100
You are asked to improve your BA delivery over 1 quarter after 28% of stories entered sprints with unresolved questions and UAT found 11 requirement gaps. What metrics and actions would you choose?
agilemonitoringuat

Questions

1You have 3 weeks to elicit requirements for a customer onboarding redesign involving 12 stakeholders across sales, compliance, operations, and engineering. How would you plan the elicitation?

2A procurement portal must be scoped in 10 working days with 10 stakeholders whose influence and daily involvement vary widely. How would you use a power-interest map to focus your analysis?

3Fourteen people from product, legal, finance, support, and engineering are contributing to a billing feature due for design approval in 2 weeks. How would you create a useful RACI?

4You have 15 business days to analyze a call-center returns process used by 80 agents, 6 supervisors, and 3 policy owners. How would you choose between workshops, interviews, and observation?

5A partner onboarding product has 11 stakeholders and must freeze its core requirements in 3 weeks, but nobody can say who decides identity checks, commercial terms, or support policy. What would you do?

6You must elicit a shared order-management workflow from 15 stakeholders in 5 time zones within 4 weeks, and no meeting slot reaches more than 9 people. How would you work remotely?

7A regulated lending workflow serving 30,000 applications per month needs requirements ready in 4 weeks, with input from 9 stakeholders and mandatory audit evidence. How would you adapt elicitation?

8A warehouse picking redesign has 8 managers in discovery but affects 120 pickers on 3 shifts, and findings are due in 12 days. How would you avoid capturing only management's view?

9After 3 weeks of discovery, a claims product has 120 draft requirements from 13 stakeholders and must enter solution design in 5 days. How would you assess completeness?

10A vendor will build a 6-month case-management solution, and you have 3 weeks to provide requirements covering 4 departments and 2 external interfaces. Would you use a BRD, an FRD, or both?

11An omnichannel returns feature has 4 user roles, 7 exception paths, and a design deadline in 10 working days. Would you model it with user stories, use cases, or both?

12A discount engine has 6 customer segments, 4 basket thresholds, and 3 promotion types, and engineering needs unambiguous rules in 7 days. How would you model them?

13A loan application can move through 9 statuses with automated and manual transitions, and the workflow must be specified in 2 weeks. How would you build a state model?

14Three source systems use different names and formats for 35 customer fields, and a unified profile specification is due in 8 working days. How would you create the data dictionary?

15A booking product must exchange customer, inventory, payment, and confirmation data with 4 systems, and interface requirements are due in 10 days. What would you specify?

16A checkout redesign must support 3,000 concurrent users before a campaign launch in 6 weeks, but stakeholders wrote only 'fast and reliable.' How would you define the NFRs?

17A healthcare portal for 25,000 patients must define privacy and access requirements within 3 weeks. How would you make security requirements measurable from a BA perspective?

18A release contains 75 requirements tied to 6 objectives and 110 planned tests, and scope review is in 5 days. How would you build traceability from objective to test?

19A subscription product has 5 epics, 60 stories, 14 business rules, and a refinement deadline in 8 days, but reviewers keep finding contradictions. How would you structure the requirements models?

20A claims process crosses 5 teams, has a 48-hour document SLA, and includes 6 exception types; the BPMN model is due in 7 working days. How would you model boundary events?

Locked questions