MBA 506 · Business Analytics
Prove what actually moves your numbers, then act on it
In about six weeks at nine hours a week, learn to design metrics, run fair tests and forecast honestly, so decisions rest on evidence.
For managers and analysts with plenty of dashboards but no clear answer on what drove results.
One-time paymentInstant access14-day money-back guarantee
What you will produce
DuolingoMeasuring What Moves the Needle
You analyze Duolingo from its public information and write an analysis of which metrics really drive its growth. It is marked against this rubric, with written feedback on every criterion, usually within the hour. You finish with work you can show.
- 01Metric Framework and Growth Model20%
- 02Data Quality Assessment20%
- 03Sampling and Statistical Inference15%
- 04Experimental and Causal Evidence20%
- 05Recommendation and Professional Ethics10%
- 06Format, APA 7, and Integrity15%
Each criterion is judged Exemplary, Proficient, Developing or Beginning.
Who it is for
Sound familiar?
“We track everything, and I still can't say what actually drove last quarter.”
What changesYou'll define the few metrics that matter and link them to the decision in front of you.
“The test showed a lift, we shipped it, and nothing changed.”
What changesYou'll size tests properly and tell a real effect from noise.
“Every forecast I give turns into a promise I can't keep.”
What changesYou'll forecast with honest prediction ranges and explain the uncertainty.
Not for you ifyou want live lectures or a study group. Everything here is self-paced.
What you will be able to do
By the end, you will be able to:
- 01
Run A/B tests you can trust
Size tests from the effect you care about, then read significance and practical impact correctly.
- 02
Find causes without an experiment
Use difference-in-differences and matched comparisons when you can't randomize.
- 03
Forecast with honest ranges
Build seasonal forecasts with Holt's method and back-test them, so plans rest on realistic ranges.
- 04
Segment customers and target well
Use RFM, clustering and logistic regression, then set break-even thresholds for who to target.
- 05
Know what keeps customers coming back
Model customer states with Markov chains, calculate lifetime value and rank the retention levers.
- 06
Put scarce resources where they count
Formulate a linear program and read shadow prices to find the bottleneck worth fixing.
Look inside
See a class, a study card and the marking before you pay.
Watch the opening of class one, flip a real study card, and see exactly how your work is judged.
One of 48. Each returns just before you would forget it, so the course stays with you.
Case 2: Analytics Audit. You see the rubric before you start, and get written feedback on every criterion.
The full textbook and the study guide written to it, so you can look anything up after you finish.
The ten classes
Ten classes, in the order you will need them.
01Analytics, Decisions, and Metrics That MatterStart from the decision, not the data. Define metrics precisely and link them with a steady-state model.
- Five Courses Built a Plan on Numbers
- Decision, Question, Metric, Data, Analysis, Action, Review
- Four Parts: Metrics, Models, a Case, and Ethics
- Four Kinds of Analytic Question
- Start From the Decision, Not the Data
02Data Quality and Describing DataReconcile data across systems. Flag outliers with z-scores and the 1.5×IQR rule. Report medians and spreads to describe distributions.
- Can We Trust the Numbers in Front of Us?
- A Metric Is Only as Good as Its Data
- Reconcile Before You Report
- Four Parts: Sources, Reconciliation, Description, and Honesty
- Transactions, Logs, Surveys, and Third Parties
03Sampling, Surveys, and InferenceDesign random surveys. Calculate confidence intervals. Apply post-stratification to correct sampling biases.
- From Bad Numbers to Bad Samples
- A Sample Stands In for a Population
- Four Parts: Margins, Bias, a Real Case, and Ethics
- Populations, Samples, and Parameters
- Random Sampling Gives Every Customer a Fair Chance
04Experiments and A-B TestingRandomize test groups. Size tests from effect size. Interpret p-values and practical business impact.
- Class Three's Warning About Samples
- Randomization Makes Two Groups Look Alike
- Four Parts: Randomization, Testing, Judgment, Ethics
- Randomization Removes Confounding
- An A-B Test Splits Real Visitors by Chance
05Causal Inference Without an ExperimentUse DiD and matching to estimate causation. Avoid confounding, selection bias, and regression to the mean.
- Causal Inference Without an Experiment
- Why Many Decisions Cannot Be Randomized
- Four Parts: Traps, Comparisons, Natural Experiments, and Assumptions
- Correlation Is Not Causation
- Confounding: A Hidden Third Factor
06Regression and PredictionFit regressions and logistic models, predict in business units with prediction intervals, and spot omitted-variable bias.
- A Prediction Is Not the Same as a Cause
- Four Parts: Fitting a Line, Predicting, Extending, a Real Case
- Eight Cafés: Seats and Weekly Orders
- Least Squares: Finding the Best-Fitting Line
- R Squared and Correlation: How Well the Line Fits
07Forecasting Under UncertaintyForecast with seasonal indices, exponential smoothing and Holt's method. Back-test accuracy and size inventory with prediction ranges.
- How Do We Plan for a Future We Cannot Know?
- A Forecast Adds Time's Own Uncertainty
- Four Parts: Error, Methods, a Real Case, and Judgment
- Inventory, Staffing, Cash, and Capacity Depend on a Forecast
- Level, Trend, and Seasonality Make Up a Time Series
08Segmentation and RecommendationSegment customers with RFM and clustering, find market-basket lift and test offers with holdout groups.
- How Do We Treat Customers Differently, and Fairly?
- Four Roadmap Parts: Groups, Recommending, a Case, and Ethics
- The Roaster's Four Judgment Segments
- Recency, Frequency, and Monetary Value
- A Worked Example: Two Customers by the Numbers
09Customer Retention and State ModelsModel customer states with Markov chains, calculate lifetime value and rank which retention lever matters most.
- Why Customers Stay or Leave
- A Question Class One Left Open
- Part One: Churn, Retention, and Survival Curves
- Cohorts: Customers Who Started Together
- Three States: Active, Paused, and Cancelled
10Optimization, Governance, and the Integrated CaseAllocate resources with linear programming and shadow prices, govern models responsibly and solve the integrated case.
- The Final Question of the Course
- What Class Nine Left Open
- Four Parts: Optimize, Then Govern
- From Prediction to a Recommended Action
- The Gift Season's Capacity Problem
When you will see results
You will use it at work before you finish.
- After class 1
You can define a metric so precisely that two teams would compute the same number, and tie it to a decision.
- After class 4
You can size and run an A/B test and tell statistical significance from practical impact.
- After class 7
You can forecast with an honest prediction interval that reflects seasonality and past error.
- After class 10
You can allocate resources with a linear program and read the shadow prices on its constraints.
Your time
Built to fit around a full-time job.
About 54 hours in total, at whatever pace your week allows. Nothing is live, the portal remembers exactly where you stopped, and study cards take about five minutes a day.
Before you decide
You don't need to code, a statistics degree or a data team. A spreadsheet is enough.
- Each class ends with a short quiz, so you know it stuck before you move on.
- Study cards bring back what you would forget, five minutes a day.
- The portal always shows your next step, so a busy week never becomes a lost month.
What $149 gets you
The complete MBA 506 module, sold on its own.
- Ten filmed classes you can fit around workPause anywhere and pick up on any device. Captions on every class, so you can study with the sound off.
- The full Business Analytics textbook, yours to keepPDF and EPUB, so you can look things up long after you finish.
- A study guide that covers every classThe key ideas, models and worked examples in one place, so you never write your own summary notes.
- 12 quizzes and exams, marked instantlyThe reasoning behind every answer, so you know what stuck and what to revisit.
- Two real-company cases with written feedbackStarting with Duolingo. You finish with analysis you can show, marked against a published rubric.
- 48 study cards that keep it freshFive minutes a day. Each card returns just before you would forget it.
- Your progress, saved everywhereThe portal remembers exactly where you stopped, so a busy week never sets you back.
- Lifetime access and updatesIt is there when the real decision arrives, including every future improvement to the course.
One payment. No subscription, no hidden fees.
What one better forecast is worth
If a forecast with honest prediction intervals cuts excess inventory by $5,000 in one season, that is more than 30 times the price. The course costs $149, once.
One-time paymentInstant access14-day money-back guarantee
Questions
What people ask before they buy.
How much time will it take?
About 54 hours in total: ten classes, the readings, 12 quizzes and exams, and two case assignments. At nine hours a week that is about six weeks. Nothing is scheduled, so you set the pace.
Is this right for beginners?
Yes. No prior statistics required. You'll learn probability, hypothesis testing, and regression from fundamentals. Prior SQL or Excel helps but isn't needed. Case studies ground concepts in operations, product, and marketing.
What if I'm too busy?
Nothing is live and nothing expires. Each class stands on its own, the portal remembers exactly where you stopped, and your study cards take about five minutes a day. Pause for a month and pick up where you left off.
I've started online courses before and never finished. Why is this different?
It is built so you finish. Every class ends with a short quiz, study cards bring back what you would otherwise forget, and the portal always shows your next step. Two graded cases give you a reason to apply it, not just watch it.
When will I see results?
Class 1 changes how you define metrics at work. After class 4 you can run a valid A/B test, after class 6 you can predict outcomes with regression, and after class 7 you can forecast with honest error ranges.
Why is it worth $149?
You get the complete MBA 506 module: ten classes, the full textbook and study guide, twelve marked quizzes and exams, and two cases with written feedback. One avoided bad decision can be worth far more than $149.
What if it isn't right for me?
If the course isn't right for you, ask within 14 days of buying for a full refund. It applies as long as you've watched fewer than three classes and haven't submitted an assignment.
Will I learn Python, SQL, or specific tools?
No. This course teaches statistical and analytical concepts, not software. You'll apply them with spreadsheets, case studies, and worked examples so you can translate them to any tool later.
Can I use this at a small company?
Yes. Metric design, A/B tests, simple forecasts and segmentation work with a spreadsheet and a few hundred rows of data. None of it requires a data team.
How do I get access?
Straight after checkout you are signed in to the CBI student portal and can start class one. We also email you a sign-in link so you can come back from any device.
Is this the same as MBA 506 in the CBI MBA?
Yes. It is the same ten classes, textbook, study guide, exams and case assignments that MBA students take for MBA 506, sold on its own.
MBA 506 · Business Analytics
Prove what actually moves your numbers, then act on it
Start tonight. If it isn't right for you, ask within 14 days for a full refund.
One-time paymentInstant access14-day money-back guarantee