Exam program: Data and business analysis

IAE-EXAM-DA · version 1.0

Sections, format, sample tasks and the scoring rubric of the IAE written exam in Data and business analysis.

Version 1.0Effective 2 November 2026Exam in English or Russian

01. Who is the exam for?

For data analysts, BI analysts, product, business and systems analysts with a bachelor's degree or higher and at least 3 years of experience (5 years without a degree). The exam tests how a candidate turns a business question into an analytical task, checks the data, calculates, writes requirements and draws a conclusion for management.

Knowledge of a particular tool is not tested: queries and calculations may be written in any widely used SQL dialect, in Python or R, as formulas or in pseudocode. Data engineers and machine learning specialists who build data pipelines and models should take the Software and data program.

02. Format

  • 5 questions, one per section, 10 minutes each.
  • 2 cases of 25 minutes. Case topics: an A/B test result from group data, cohort retention and the payback of an advertising channel, metric definitions and report requirements when stakeholders disagree, a data quality incident after a release, the effect of a price change with a control group, the contribution of an advertising channel from a switch-off test.
  • 1 question about the project from the candidate's questionnaire, 10 minutes: one analytical decision or a situation with stakeholders, and the result in figures.
  • Up to 2 hours in total, in one sitting. Pasting is disabled, and each candidate gets their own figures and circumstances in the tasks. Reference materials may be used.

03. Sections and weights

  • S1. Metrics and problem framing - 20 %: turning a business question into an analytical task, metric trees, decomposing a revenue change, mix effects and averages of ratios, unit economics and payback, KPIs and their side effects, seasonality, cohort retention, the effect of a price change.
  • S2. Data, SQL and data quality - 20 %: joins and row multiplication, aggregates and window functions, cohorts and conversion in a query, deduplication, checks of a daily load, time zones and reconciliation with accounting, missing data, analysis of a data quality incident.
  • S3. Statistics and experiments - 20 %: sample size and duration of an A/B test, significance and confidence intervals, peeking and multiple comparisons, a broken split between groups, heavy-tailed metrics, regression to the mean, correlation and causation, non-significant results, surveys and bias.
  • S4. Requirements and processes - 20 %: the need behind a client's wording, user stories and splitting an epic, acceptance criteria, non-functional requirements, metric definitions and their agreement, prioritising requests, process models, cycle time and bottlenecks.
  • S5. Professional practice - 20 %: conclusions and recommendations for management, uncertainty in a forecast and a decision, honest visualisation, alert thresholds, confidentiality in reports, pressure on the analyst, handling a flow of requests, automating reports, the candidate's own project.

04. Sample question

Question (section S1, 10 minutes). A subscription service with a constant monthly customer churn of 4% is preparing a presentation for an investor. A manager computes annual churn as monthly churn multiplied by 12. The subscription costs $50 a month, and the gross margin is 75%.

  • 1) Calculate the share of customers left after 12 months and the annual churn. How wrong is the manager's calculation?
  • 2) Calculate the average customer lifetime and the gross margin per customer over the lifetime.
  • 3) Why is constant churn an assumption, and how would you check it?

Model answer. 1) After 12 months 0.96^12 = 61.3% of customers remain, annual churn is 38.7%, not 48%: multiplying by 12 ignores that each month a share of a smaller base leaves. 2) Average lifetime 1 / 0.04 = 25 months, margin over the lifetime 50 x 0.75 / 0.04 = $938. 3) Churn is usually higher in the first months and lower later, so the 1 / m formula can be wrong in either direction. I check with cohort retention curves and show the investor actual churn by cohort.

How this question is scored:

  • Correctness (40): 34-40 - annual churn, lifetime and margin are right for the candidate's figures; 20-33 - the annual churn formula is right, but there is an error in lifetime or margin; 0-19 - the solution is wrong or missing.
  • Reasoning (25): explains why multiplying by 12 overstates churn and why the investor needs actual cohorts.
  • Risks and constraints (20): covers non-constant churn, different cohorts and channels, and the risk of overstating customer value.
  • Clarity (15): the answer is structured in points, conclusions and figures are stated explicitly, no padding. Grammar and style do not reduce the score if the meaning is clear.

The example comes from the practice exam. In the exam each candidate gets their own figures: here churn, price and margin vary, and the model answer shows how the result depends on them.

05. Sample case

Case (section S3; 25 minutes in the exam, 15 in the practice exam). A store sent an email with two subject lines, 30,000 recipients in each group, split at random. Variant A: 22% of recipients opened the email and 1.0% placed an order. Variant B: 26% opened and 1.1% placed an order. A marketer proposes using subject B in all mailings because "the open rate grew by 18%".

  • 1) Test the significance of the difference in order rate (5% level, two-sided) and give the 95% interval of the difference.
  • 2) Which metric do you consider primary for choosing a subject line, and why?
  • 3) What will you recommend to the marketer?
  • 4) How many recipients per group are needed to detect a 0.1 pp difference in orders with 80% power? What does this mean for future tests?

Model answer. 1) The average order rate is 1.05%; z = 0.001 / sqrt(0.0105 x 0.9895 x 2 / 30,000) = 1.20, p = 0.23; the interval of the difference is from -0.06 to +0.26 pp - not significant. Opens differ significantly (z = 11.5), and the growth of 26 / 22 - 1 = 18% is computed correctly. 2) The primary metric is orders or margin per recipient: the goal of the mailing is orders. More opens without more orders may mean the subject attracts less interested readers; opens are also distorted by email programs that load messages automatically. 3) Sales growth is not proven: subject B can stay, but it cannot be claimed to increase orders; for the next mailing, orders per recipient is written down as the primary metric. 4) n = 7.84 x (0.01 x 0.99 + 0.011 x 0.989) / 0.001^2, about 163,000 per group, more than five such mailings. So small subject-line differences cannot be tested on one mailing: combine several mailings under a plan written in advance, or test larger changes (the offer, the segment).

How this case is scored:

  • Correctness (40): 34-40 - z and the interval are right for the candidate's figures, the primary metric is chosen and explained, the recommendation follows from the result, the sample size is right; 20-33 - the metric conclusion is right, but there is an error in z, the interval or the sample size; 0-19 - the solution is wrong or missing.
  • Reasoning (25): the choice of metric and the recommendation are explained by the goal of the mailing and the test result.
  • Risks and constraints (20): covers distorted opens, several metrics in one test, insufficient power of one mailing, and extending one mailing's result to all.
  • Clarity (15): the answer is structured in points, conclusions and figures are stated explicitly, no padding. Grammar and style do not reduce the score if the meaning is clear.
  • An answer in general terms that does not work through the figures in the task scores at most 19 for correctness and at most 12 for reasoning.

06. Scoring rubric

Each task is scored on four criteria; the total is converted to 100 points using the section weights. The examiner checks the answer against the rubric and the model answer for the task with the candidate's figures.

  • Correctness of the solution - 40 %: a correct calculation, query or requirement for the candidate's figures. Without a calculation on the candidate's own figures - at most 19 of 40.
  • Reasoning behind decisions - 25 %: why this option was chosen over another. General recommendations not tied to the figures and circumstances of the task - at most 12 of 25.
  • Risks and constraints - 20 %: data quality, precision of the estimate, assumptions, the conditions under which the conclusion changes.
  • Clarity - 15 %: structure and precise wording. Grammar and style do not lower the score if the meaning is clear.
  • Naming a method or test without applying it to the task's figures adds no points. Minor SQL syntax errors do not lower the score if the logic of the query is right.

07. Result and certificate

  • Pass mark - 70 of 100.
  • Result in 5 business days. Those who pass receive an IAE Certified Professional certificate in "Data and Business Analysis", valid for 3 years; scores by section are in a separate report.
  • The certificate confirms that the candidate passed the IAE written exam under this program. It is not a license or a government qualification and does not replace certificates from vendors of data analysis software. Employers and authorities make hiring and other decisions about the candidate under their own rules.
  • Retake 30 days after the first failed attempt and 60 days after the second, with no more than 3 attempts in 12 months.
  • Full terms: the Exam terms page.
  • The Credentialing Council reviews the program at least once a year.

08. Out of scope

  • Work in particular BI tools (Power BI, Tableau and others), their settings and features.
  • Building data pipelines, warehouses and machine learning models: this is the Software and data program.
  • Database administration and query tuning at the database engine level.
  • The syntax of a particular SQL dialect or of Python or R libraries.
  • Formulas and statistical tables from memory: the exam tests how the candidate chooses a method and applies it to the task's figures.
  • Financial statements, accounting and legal assessment of personal data processing: requirements needed for an answer are given in the task.

09. Preparation

  • Practice exam: 2 questions, 1 case and a project question in the same format. Afterwards the model answers and the rubric are shown; no score is given. Terms are on the Exam terms page.
  • Calculations found in the tasks: decomposing revenue into traffic, conversion and average order value; weighted averages and mix effects; acquisition cost, payback, revenue and margin over the customer lifetime; sample size for two proportions; the z-test and confidence interval for a difference of proportions; a chi-square check of the group split; difference in differences; utilisation of a process step.
  • Queries found in the tasks: joins without row multiplication, conditional aggregates, ranking and moving windows, cohorts and retention, finding and removing duplicates, reconciling totals with the source.
  • Requirements: user stories, acceptance criteria in the "Given, When, Then" format with testable figures, a metric definition with a formula, source and owner.
  • The project question is built on the project from the questionnaire: recall in advance the business question, the data and its defects, your calculation, the result in figures with its precision and who made the decision based on your conclusion.

Exam

Book the exam

The questionnaire takes 15 minutes. Admission decision in 1 business day.

Candidate questionnaire

Received

Admission decision within one business day

FieldData and business analysis
DurationUp to 2 hours
Result5 business days