1. Treat the DWH as the system of record for the week
Our existing rhythm is actually very strong:
- Monday – DWH moves to PRD
This becomes the contract for the week.
Views, semantics, behaviour are now fixed and trusted. - Tuesday – Backlog review
This is where new requirements are discussed against what is now live, not against what people think is coming. - Friday – Documentation only
This locks in what was delivered and avoids late-week destabilisation.
That already gives you something most organisations lack: a weekly data contract.
Reports must key off that.
2. Reports should always trail the DWH by one controlled step
With DEV → UAT → LIVE workspaces, you get a clean handshake:
Monday
DWH PRD is updated.
Nothing in LIVE reports should change yet.
Monday–Thursday
Report teams work in DEV and UAT against the new DWH PRD views:
- Adjust measures
- Validate joins
- Check edge cases
- Fix visuals
No guessing. No “it’ll be there soon”. The data is already real.
Friday
Documentation day:
- DWH publishes what changed
- Reports document what they’re ready to release
- No code churn
Following week (e.g. Tuesday or Wednesday)
Promote reports from UAT → LIVE via deployment pipeline.
That gives us:
- Stable data first
- Reports validated against it
- Clean, low-risk promotion
This also gives stakeholders something they understand:
“Data lands Monday, reports update the following week.”
3. This avoids the most dangerous failure mode
The worst thing we can do is:
Change DWH and reports in the same week in LIVE.
That creates:
- Finger-pointing when numbers don’t match
- No baseline for Finance or PMO
- Impossible debugging (“was it the SQL or the measure?”)
Our cadence prevents that — we just need the workspaces and pipelines to enforce it.
4. How this fits your error-intolerant stance
We’ve said before we’re error-intolerant in production.
This model supports that:
- Monday = only data engineers touch production
- Tuesday–Thursday = analysts can break things safely in DEV/UAT
- Friday = nothing moves, everything is recorded
- Release = controlled promotion, not ad-hoc publish
It also fits beautifully with your CDM freeze + RPV Sunday model — the DWH becomes the stable spine, and reports are consumers of a known, versioned truth.
Bottom line
Our current DWH release rhythm is already right.
The report release strategy should be built to orbit it, not compete with it.
If you align them this way, you get:
- Predictability for Finance
- Safety for Operations
- Freedom for Report Developers
- And no more “what changed?” firefights 🔥