CloudRevenue - Schema
=====================

Synthetic cloud subscription revenue dataset. Three tables joined on tenant_id and product_id. 3 years of monthly grain ending 2026-05.

Tables
------

products.csv (8 rows)
---------------------

Column                         Type     Notes
-----------------------------  -------  --------------------------------------------------------------------------------
product_id                     string   Primary key. Format P-XXXX.
product_name                   string   Display name of the product family.
segment                        string   One of Data, Infrastructure, Security, Operations, Productivity.
list_price_per_user_month_usd  decimal  Reference list price; actual revenue applies size-band discount.
growth_band                    string   One of high, stable, declining. Drives the per-month trend in the revenue table.

tenants.csv (247 rows)
----------------------

Column                Type        Notes
--------------------  ----------  ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
tenant_id             string      Primary key. Format T00000.
tenant_name           string      Synthetic name, format Tenant N.
size_band             string      One of SMB, Mid-Market, Enterprise, Strategic.
region                string      Known imperfection: NA tenants are sometimes stored as NA, sometimes as North America (~12% of NA tenants). Chapter exercises can use this for naming-inconsistency cleanup.
vertical              string      One of 8 industry verticals.
seats                 integer     Number of seats on the tenant's master agreement. Drives MAU baseline in revenue.
first_contract_month  date (ISO)  Earliest month the tenant has revenue.
contract_size_tier    string      One of monthly, annual, multi-year.
churn_flag            boolean     true for 12 mid-window churners.
churn_month           date (ISO)  Last month with revenue, if churned. Empty otherwise.

revenue.csv (~21,400 rows)
--------------------------

Column           Type             Notes
---------------  ---------------  -----------------------------------------------------------------------------------------------------------------------------
month            date (ISO)       First day of the month (e.g., 2025-03-01).
tenant_id        string           Foreign key -> tenants.tenant_id.
product_id       string           Foreign key -> products.product_id.
mau              integer or NULL  Monthly active users. Known imperfection: ~2% null (telemetry outage).
revenue_usd      integer or NULL  Computed: mau x list_price x discount_factor. Known imperfection: last 2 months have ~30% nulls (billing close not complete).
discount_factor  decimal          The size-band discount applied (SMB=1.00, Mid-Market=0.92, Enterprise=0.82, Strategic=0.70).

Suggested chapter exercises
---------------------------

- Chapter 3 (story): Build the headline narrative - which product families are driving Q4 portfolio decisions? Pair with the CloudRevenue starter BRD.
- Chapter 4 (chart selection): Tenant-level concentration: which 10 tenants drive what % of revenue per product? Compare bar / treemap / Pareto encodings.
- Chapter 5 (focus and decluttering): Strip the busy default dashboard down to a single 5-visual answer to "where should Q4 investment cuts land?"
- Chapter 9 (Copilot report generation): Direct Copilot to build the portfolio review report; observe how the CSAR loop handles the region-naming inconsistency.
- Chapter 10 (AI visuals): Key Influencers - what predicts revenue growth across product x tenant?
- Chapter 13 (when AI misleads): Watch what Copilot does with the ~30% null revenue in the last 2 months. Does it surface the in-flight state, or compute averages as if the data were complete?

Reproducibility
---------------

Generated by scripts/datasets/generate-cloudrevenue.js using a seeded PRNG. Rerunning the script produces byte-identical output. Source seed: 0xC10D5EED.