The reports you rent and the reports you own
Delivered reporting runs your operations and does it well. Then an analysis question crosses two objects, or three years, or one restructuring, and the walls appear. They are structural, and renting more analytics does not move them.
◆ The seven wallseach one structural, none of them configuration mistakes.
Practitioners rediscover these in the same order. A report needs data from two places and the report writer allows one primary data source per report, so the workaround chain begins. Composite reports look like the answer and turn out to assemble sub-reports side by side rather than join them. Then the trend request arrives, and the delivered history is a window, not an archive.
| Wall | What it means | The workaround it forces |
|---|---|---|
| One data source | Every report has one primary business object. Related objects reach only as far as delivered relationships go. | Chains of lookup fields, or export and join in spreadsheets. |
| Composite assembly | Composite reports lay sub-reports side by side. They do not join unrelated sources, and grouping breaks on lookup columns. | Fragile layouts rebuilt by hand each cycle. |
| No windowed math | Rolling averages, cohorts, and period-over-period ratios have no native home in the report writer. | Calculated-field chains, then spreadsheets anyway. |
| History is a window | Workforce snapshots are monthly and purge after roughly 36 periods, practitioner-observed. Covered on the trending store page. | Manual archives of old exports. |
| Capacity-priced analytics | The analytics add-on is licensed by record allocations. Growth in data is growth in fees, and overage charges arrive mid-contract, practitioner-reported. | Budget meetings about row counts. |
| Walled output | Add-on results stay inside Workday security and Workday screens. Direct download of a published dataset is not a given. | Reports built on top of the tool that was the workaround. |
| The data still cannot leave | None of the above moves your data into a warehouse you control. The walls travel with the tenant. | This catalog's extraction pattern. |
◆ What ownership changesthe same questions, answered by structure instead of workaround.
An owned model dissolves the walls rather than climbing them. Two objects in one answer is a join. A rolling average is one window function. History is a table that grows, not a window that purges, and the toolchain on top is whatever your team already uses. The enterprise model shows the shape; the module pages show every report built on it.
The delivered stack keeps the job it is good at: transactions, approvals, and the operational lists that run the week. The analytical questions move to a model priced like storage and compute, because that is all it is. Ownership is the difference between paying for answers and owning the thing that answers.
- business object
- What Workday stores instead of tables. A report reads one of them.
- data source
- The population a report is built on. One primary per report.
- composite report
- A delivered report type that assembles sub-reports side by side.
- lookup field
- A field reaching into a related object, one delivered hop at a time.
- window function
- Database math across neighboring rows: rolling averages, ranks, cohorts.
- capacity pricing
- Licensing metered by stored record counts rather than by use.
- star schema
- Facts joined to shared dimensions. The owned model’s shape.
- practitioner-reported
- Figures seen in the field but not published by the vendor. Labeled until verified.