Risks and Dependencies in Analysis Planning

How to identify, document, and mitigate things that could delay or invalidate an analysis before they cause problems.

Dependency	Example	How to resolve
Data access not yet granted	Need access to a vendor's export	Raise access request immediately; don't start the analysis until confirmed
Data doesn't exist	Metric was never tracked	Scope out or use a proxy; document the limitation
Data is unreliable / known quality issues	A known ETL bug affects the column	Document the issue; decide whether to proceed with caveated results or wait for the fix
Data is too fresh (not enough history)	New feature only 2 weeks old	Narrow the question or set a longer timeline

Dependency	Example	How to resolve
SME input required	Need product manager to confirm event definition	Schedule immediately; flag as a blocker if unresolved
Stakeholder needs to confirm scope	Requirements not yet signed off	Do not start until sign-off; treat as day 0
Another analyst is needed	Two people needed for parallel work	Confirm availability before committing the timeline

Dependency	Example	How to resolve
Warehouse quota / cost approval	Large query needs DBA sign-off	Get approval before running
Dashboard tool access	Need a BI tool licence	Request early; use an alternative format as fallback
Third-party data delivery	Vendor sends a file on the 5th of the month	Build the timeline around the delivery date

For each identified risk, record:

Field	What to capture
Risk	Brief description of what could go wrong
Probability	High / Medium / Low
Impact on timeline	Hours / days it would add
Mitigation	Action taken before the risk materialises
Contingency	What to do if the risk materialises anyway

Risk	Probability	Typical impact	Mitigation
Data quality issue discovered mid-analysis	Medium	+0.5–1 day	Run EDA on data before committing to approach
Scope expands after requirements sign-off	High	+0.5–2 days	Document scope boundaries explicitly; get written sign-off
Stakeholder unavailable for feedback	Medium	+1–3 days	Book feedback slot in advance; set a response deadline
Key assumption turns out to be wrong	Medium	Restart sub-questions	Make assumptions explicit early; validate before building on them
Approach doesn't work on the actual data	Low–Medium	+0.5–1 day	Run a proof-of-concept before full implementation
External data delivery delayed	Low–Medium	+1–5 days	Identify alternative or contingency approach

Escalate to the requestor when:

Escalate early. A stakeholder told on day 1 that the delivery will be late has options. A stakeholder told on the eve of the delivery has none.

/analysis-planning