DATIMORE · WEB-047
Warehouse-readiness checklist and worked history
Use this worksheet to keep a useful reporting history with a setup your team can maintain. Complete it with the business owner and the people responsible for the source systems.
Synthetic planning example, not a client case or a working integration. The records and results below are invented to explain the method. This static worksheet does not load data, preserve history or enforce permissions.
Recurring comparison / people who need it:
Source systems / history required:
Business owner / backup / review date:
1. Decide what needs to stay available
History
Which past states or changes must the report explain? How long must they remain available? Check whether each source already retains them. A new warehouse cannot recreate missing past records.
Evidence / named owner:
Ready / Open / Unknown — next action and date:
Sources and shared definitions
List each source, the matching identifier and the agreed calculation. Confirm that exports cover the same cutoff and include every required record. Name who approves changes to definitions.
Evidence / named owner:
Ready / Open / Unknown — next action and date:
Completeness and update timing
Agree a reporting deadline and the last complete period. State what happens when a source is late or an identifier is missing. A completed transfer alone does not prove that the figures are complete.
Evidence / named owner:
Ready / Open / Unknown — next action and date:
Access
Name who may view detail, change matching rules and approve a period. Limit stored data to what is needed. Confirm access and retention requirements with the people responsible.
Evidence / named owner:
Ready / Open / Unknown — next action and date:
Ownership and recovery
Name the owner and backup for each source and the report. Decide who investigates failed updates, checks corrections and restores a saved copy. Plan a recovery check before relying on the setup.
Evidence / named owner:
Ready / Open / Unknown — next action and date:
Cost and skills
Compare storage, data transfers, reporting access and maintenance effort for both options. Name who has time and skills to maintain each one. Record assumptions and obtain current costs for your own scope.
Evidence / named owner:
Ready / Open / Unknown — next action and date:
2. Choose the smallest suitable setup
A report inside one system can serve a current operational question. A small shared dataset with agreed calculations can combine a few sources when the required history is retained and updates are manageable. Consider a warehouse when saving and reusing history across systems and reports becomes recurring work. Source count or company size alone is not a threshold.
Proposed option / why the smaller option is or is not sufficient:
Do not treat this as a points-based score. An unknown source, unclear owner or missing permission needs an answer before the affected data is used. Pilot one reporting question, then review the evidence before expanding.
3. Follow the saved records
Synthetic distributor, one product, units rather than money. At each month-end, both systems provide one complete record per order. Dispatch records show the total sent up to that date. There are no cancellations, returns, multiple identifiers for the same order or duplicate rows. The dispatch source explicitly confirms quantities, including zero; missing information is not zero.
| Cutoff | Order number | Order source: units ordered | Dispatch source: cumulative units sent |
|---|---|---|---|
| 30 September 2026 | A | 20 | 12 |
| 30 September 2026 | B | 15 | 15 |
| 31 October 2026 | A | 20 | 20 |
| 31 October 2026 | B | 15 | 15 |
| 31 October 2026 | C | 10 | 4 |
For each cutoff, match on order number, subtract cumulative dispatched units from ordered units, then add the outstanding quantities. Keep the dated input copies and the approved result. Do not add the two months together: these are states at two dates, not monthly shipments.
September: (20 − 12) + (15 − 15) = 8 units outstanding.
October: (20 − 20) + (15 − 15) + (10 − 4) = 6 units outstanding. September stays at 8.
4. Try an incomplete input
In the October dispatch export, replace the order number C with a blank while keeping its quantity of 4. The rule can no longer connect that shipment to an order. The October total stays provisional; it must not be approved as 10 by ignoring the unmatched shipment. The dispatch owner checks the source, confirms C and corrects its identifier. Keep the received copy and the corrected version with the reason and approval time. Rechecking yields October 6; the separate September record remains 8.
Expected checks before approving the example
- Complete September inputs → 8 outstanding.
- Complete October inputs → 6; retained September → 8.
- Blank order number on the four-unit dispatch → October provisional, no approved total.
- Confirmed source correction to C → October 6; September still 8.
- Missing or late source export → period provisional, not zero.
- A later correction to an approved period → retain the original, record reason and approval, label the revised version.
These are small paper rules. Real systems need checks for cancellations, returns, duplicate records, source changes, simultaneous updates, access and recovery. The warehouse does not perform these checks merely because it stores the data.
5. Minimal source-to-report architecture
Download the editable draw.io SVG
Microsoft’s cited small/medium business design is for modernizing an existing warehouse. It explicitly excludes building a new warehouse from scratch; do not copy it as a default first-warehouse architecture. Microsoft architecture and its exclusions.