Work should be iteratively reviewed by everyone involved throughout the initiative. There are multiple layers of review:
- Self Review - Check your own work first
- Team/Critical Friend Review - Peer review of code and outputs
- Data Science Lead Review - Leadership oversight
- Customer Review - Validate with stakeholders
Reviews may require returning to earlier stages (data preparation, analysis, or drafting outputs).
Standard Review Checks
Before customer review, all reviewers should:
- Check for errors by sense checking results and comparing to existing resources
- Proofread text to ensure factual accuracy and appropriate audience targeting
Self review
Review your own work continuously, and especially before requesting reviews from others.
Before Requesting Review:
- Ensure all points in the QA section are addressed
- Sense check against the agreed Problem Definition
- Confirm customer agreement for any deviations from the PD
- Sense check reporting outputs to avoid naïve conclusions
- Run code with a ‘clean slate’ to ensure no local dependencies
- Consider writing repeatable tests rather than ad-hoc testing
- Document your code and methodology
Creating Pull Requests:
- Create a PR on GitHub once happy with the code
- If building on existing work, check the code diff and add comments explaining non-obvious decisions
- Review the code diff to catch anything you missed
- Keep PRs small and focused when appropriate
- Request a code review once satisfied with your PR
ImportantGive colleagues sufficient time
Remember that your colleagues have other work; give them enough time to review your work properly.
Team Review
One or more team members should review all code and outputs before merging to the main branch.
Reviewer Responsibilities:
- Sense check against the Problem Definition
- Pull the code locally and run it from a ‘blank slate’ state
- Add comments and suggestions to the PR
- Ask the code author for clarification when needed
- Arrange a call to work through the PR together if several questions arise
Critical Friend Review
The Critical Friend provides independent peer review throughout the project.
Review Process:
- Request feedback frequently - Always before customer reviews
- Validate against the Problem Definition - Ensure alignment with agreed scope
- Review from multiple perspectives:
- As the customer viewing outputs
- As a developer using the codebase
- Review holistically - Not just the code, but the entire work
- Discuss complex feedback - Arrange calls when needed
Customer Review
Involve customers frequently throughout the initiative.
During Development:
- Share draft outputs for review in their own time (not just screen sharing)
- Be available for calls to discuss their feedback
- Address questions and suggestions openly
Final Presentation:
When the initiative is completed, present findings to the customer and:
- Ensure they understand how to use interactive elements (e.g., dashboards)
- Discuss potential for future related work
- Send a follow-up survey (if appropriate) to confirm all needs were met