This data platform readiness checklist scores one platform on 13 checks in four groups, each with its evidence and its owner. Then three verdicts follow from the total, but the weakest group can lower the verdict by one level.
Last reviewed 2026-11-14 by Manish Sharma.
A data platform is the place where a firm’s data is stored, moved, governed and served to reports and AI work. Also, the checks below look at that platform on AWS or Azure, and at the people and routines around it. Data platform readiness here means the platform can carry more analytics and AI work without surprises. Also, each check asks the team to show something you can open. However, data quality checks sit outside this guide.
Data Platform Readiness Checklist: Verdict Bands and What Each Mark Stands For
First, every check gets one of three marks. Inspected means the reviewer looked at it on the platform or in a record. Reported means the team vouches for it, but nobody has opened it. Lacking means there is no answer.
| Mark | What counts as evidence | Points |
|---|---|---|
| Inspected | A screen, a record or a log the reviewer opened | 2 |
| Reported | A statement from the owner, nothing opened yet | 1 |
| Lacking | No answer, or no owner named | 0 |
Then the data platform readiness checklist adds the points. Group A has 3 checks and a maximum of 6, group B has 4 and 8, group C has 3 and 6, and group D has 3 and 6, so the total tops out at 26.
A total of 21 to 26 gives “Carry the load”. If the total is 14 to 20, it gives “Strengthen, then carry”. When the total is under 14, it gives “Rebuild the base”. Then read each group against its own maximum: a group below half of its maximum lowers the verdict by one level.
Signisys working view: these bands are a starting rule, set to send a mid-range total to a named fix. Basis: our own scoring design, not a provider rule. Adjust them with your own CIO before the first scoring, and write the change down.
Key Takeaway The total sets the verdict, but the weakest group can lower it by one level.
Data Platform Readiness Checklist: Questions, Evidence and Owners by Group
The four groups follow the life of the data. First comes where it lands and moves, then who may see it, then what it costs to run and who owns that, and last how it recovers. Also, each table gives the evidence to ask for and the owner. Then score the groups one at a time.
Where Data Lands and Moves
| Check | Evidence to ask for | Owner |
|---|---|---|
| A1 Storage and compute sized from measured use | A month of usage and wait times | Platform lead |
| A2 Pipelines watched, with an owner and an alert route | Alert history and the owner list | Data engineering lead |
| A3 Reports match after a move | Before and after totals for key reports, signed by report owners | Reporting lead |
A1 asks for measurements, but not a forecast. A2 asks who answers when a load fails overnight. A3 asks for matching totals, as a move that changes a report’s number has failed even when the load ran.
Who May See What
| Check | Evidence to ask for | Owner |
|---|---|---|
| B1 Access by table, column and row, with logs | An access list sample and a log extract | Security lead |
| B2 A named steward on the catalog entry of each key dataset | A catalog export showing steward and last refresh | Data governance lead |
| B3 Sensitivity labels the platform enforces on analytics and AI jobs | The label list and a log line showing a job allowed or refused by label | Security lead with data owner |
| B4 Data location for India and the GCC | Region settings and where backups sit | Security or compliance lead |
Next, for B4, read the Azure regions overview on Microsoft Learn and the AWS Regions and Availability Zones page. As of 14 November 2026, Microsoft’s page says each Azure geography is a data residency boundary, so regions should be chosen inside the geography your organization requires. AWS’s page describes Regions and Availability Zones and gives no residency rule, so the owner reads where each AWS service in use stores its data. However, this guide names no residency rule. Cite a rule you rely on only with its regulator source, and until then B4 stays open, with the owner answering from a record.
Ask the owner to show where each service stores data and where backups sit, then record the page and the date.
What It Costs to Run and Who Owns That
| Check | Evidence to ask for | Owner |
|---|---|---|
| C1 A cost owner with guardrails, no figures in the file | A named owner and a spend alert switched on | Finance partner with platform lead |
| C2 The platform defined as code | A repository with change history | Platform lead |
| C3 A run routine and skills | Runbooks, a named on-call and a skills list | IT head |
C1 asks who answers for spend, but not how much. C2 asks whether a rebuild is a repeat of a recorded change. Also, C3 asks who runs the platform on a quiet Sunday and who covers their leave.
How It Recovers
| Check | Evidence to ask for | Owner |
|---|---|---|
| D1 A backup and restore rehearsal | A dated record of restoring one key table | Platform lead |
| D2 Recovery targets agreed | Signed targets per class of dataset | Business owner with IT head |
| D3 A routine to re-run a failed load | A failed load re-run in the record, with no duplicate rows | Data engineering lead |
However, a backup nobody has restored is a belief. Also, D2 needs a business signature, not only an IT one. Then D3 asks for a record where the re-run left each row once.
Mark D1 Inspected only when you have opened a dated record of one key table restored.
Data Platform Readiness Checklist: A Reporting Platform Scored Group by Group
First, the example is invented for illustration and is no Signisys project. Also, the platform serves finance reports and sales analytics.
| Group | Score | Reason |
|---|---|---|
| A Where data lands and moves | 6 of 6 | Usage record, alert history and matching totals all inspected |
| B Who may see what | 7 of 8 | Three checks inspected; B3 reported, no log line yet |
| C What it costs to run | 2 of 6 | Repository inspected; no cost owner named, no on-call list |
| D How it recovers | 6 of 6 | Restore record, signed targets and a re-run all inspected |
The total is 21, so it sits in “Carry the load”. Group C scores 2 of 6, below half of its maximum of 6, so the verdict drops one level to “Strengthen, then carry”. Then the two items to fix are C1, owned by the finance partner with the platform lead, and C3, owned by the IT head. The other checks need no fix before the next scoring.
Key Takeaway One weak group changes the verdict, even when the total looks strong.
Data Platform Readiness Checklist: Mending the Weakest Group First
Fix the lowest group first, then the next, and score each group again after its fixes. Then a group that reaches half of its maximum or more lifts the verdict back one level.
Signisys working view: work in this order, because a weak run or cost group can undo strong storage and pipelines. Basis: a platform nobody owns for spend or recovery loses the value of its storage and pipelines. Store the proof next to the points, so a later reviewer sees how each mark was reached.
After a verdict of “Rebuild the base”, the next step is a data platform assessment with the owners in the room. That turns each Lacking mark into a named task.
Sources and Neighboring Guides
Also, three guides sit beside this one.
- why an older warehouse starts to limit a firm covers the pressures that push firms to a new platform.
- the build order for each platform item covers the build steps.
- the readiness checks for the AI work itself covers the AI side, which this guide leaves alone.
Again, data quality checks and prices sit outside this guide, and storage model choice is set out in a side by side look at where analytics and AI data is kept.
Also, data quality, storage model choice and prices are not scored here.
Signisys provides this work: see how Signisys builds a cloud data platform for analytics and AI, our data integration and reporting solution and data analytics and AI support. Also, a working session scores your platform on your own records and promises no result. No charge, with a fixed scope.
Data Platform Readiness Checklist FAQ
When should the platform be scored again?
Does this checklist cover data quality?
Which datasets get a sensitivity label first?
What should a team do when overnight loads finish late?
Security Insights, Delivered.
Practical, vendor-neutral analysis of threats, tools, and architecture decisions.