A portfolio piece should make it possible for another person to understand the decision, inspect the evidence and see how you arrived at the recommendation. Visual polish helps, but it is not the first criterion.
1. Decision clarity — 0 to 4
A score of four means the work starts with a specific decision owner and a specific choice. “Analyze ticket data” is a weak brief. “Decide whether to open two additional upper-bowl sections for the next three Saturday fixtures” is a decision.
2. Evidence quality — 0 to 4
Document source, grain, time period, exclusions and known limitations. A strong piece explains why the chosen metric represents the question and shows at least one quality check.
3. Reproducibility — 0 to 4
Another analyst should be able to repeat the core process from your notes, formulas or code. Manual steps are acceptable if they are documented. Hidden spreadsheet edits are not.
4. Communication — 0 to 4
The output should match the stakeholder. A coach may need annotated clips and a short note; an executive may need one chart, one caveat and one recommendation. Do not force every project into a dashboard.
5. Reflection — 0 to 4
State what you would change with better data, more time or a second iteration. Reflection is evidence of judgment, not an apology for imperfect work.
The 15-point publish rule
A piece that scores 15 or more out of 20 and has no zero in decision clarity or evidence quality is usually strong enough for a public portfolio after confidential information is removed. Below that threshold, revise the weakest criterion before adding visual decoration.
Use the rubric during the project
Score the work at the brief, midpoint and final stage. The biggest improvement often comes from rewriting the decision question early, before you invest hours in analysis that nobody needs.
Score in language a hiring manager can challenge
After you assign the numeric score, write one sentence of evidence for each criterion. “Decision clarity: 3/4 because the recommendation names an action but not the owner.” “Reproducibility: 2/4 because the cleaning steps are described but not saved as code or a repeatable checklist.” This prevents the rubric from becoming a decorative total.
Ask a second reviewer to score one project without seeing your numbers. The disagreement is useful. If one person sees a clear decision and another does not, the artifact probably relies on context that exists only in the author’s head.
Use the rubric on a real assignment brief before the project is finished. Early scoring shows where another chart will not help and where a definition, source note or clearer recommendation will.
Review the artifact as if the author were absent
Score the project once without the author in the room. If a reviewer cannot find the question, source, definition, method, limitation, and recommendation from the artifact itself, lower the score even if the analysis is technically sophisticated. Portfolio quality depends on what another person can inspect, not on what the creator remembers.
Apply the rubric to Case: Build a Match Performance Dashboard and compare the result with Case: Write a Post-Match Executive Memo. Use the disagreement between two reviewers as evidence: it shows which definition or explanation is still too dependent on context. Record one revision that follows from the review rather than simply raising the score after cosmetic changes.
Finally, compare the improved artifact with Case: Build Actionable Fan Segments. A rubric is useful when it changes the work before publication. If every project receives roughly the same score and no revision follows, the criteria are probably too vague.
Use the rubric before the project is finished
A rubric is most useful when it changes the work while there is still time to revise it. Score the project after the first complete draft, not after the final polish. For every criterion, write one sentence of evidence beside the number. If you cannot justify the score with something visible in the artifact, the score is too generous.
Apply the rubric to Case: Build a Match Performance Dashboard and ask a second reviewer to score the same piece independently. Compare not only the totals but the reasons. A disagreement about decision clarity, source quality, or reproducibility usually points to missing context in the artifact. Use Case: Write a Post-Match Executive Memo as another test and see whether the same criteria still work across a different output type.
Make one substantive revision: rewrite the decision, add a definition, preserve a transformation step, tighten a limitation, or replace a decorative visual with evidence that supports the recommendation. Score again only after that revision. The point is to create a visible relationship between feedback and improvement.
Finally, use Case: Build Actionable Fan Segments to check transfer. A good rubric should distinguish strong and weak work across roles without rewarding a single visual style. If every project receives the same score, refine the criteria until they reveal meaningful differences in judgment, evidence, and handoff quality.