On this page
A work sample becomes more useful when you can explain what it proves. A polished chart can show presentation skill, but an interviewer may ask why its inputs belong in the calculation, what would invalidate it and which conclusion follows from the example.
Choose an example with a clear question
Use original synthetic data or appropriately sourced public information. Avoid employer forecasts and client records in a public portfolio. Define one bounded question, such as how an indivisible budget changes a ratio-ranked project list or how a denominator changes an emissions trend.
| Question | The quantity or comparison being answered |
| Inputs | Definitions, dates, units and source status |
| Method | The operation connecting inputs to result |
| Independent check | A representation that exposes a possible error |
| Limits | Conditions omitted or still uncertain |
Distinguish your contribution from the source
Name which records were supplied, which assumptions you chose and which calculations you performed. A source’s reputation does not validate every adaptation. If your demonstration factor is fictional, say so. If a boundary decision is unresolved, do not turn it into an assured category.
The ESG data work-sample guide provides a source-and-unit record. The capital-allocation work-sample guide supplies a dated project bridge and constraint-aware comparison.
Explain one correction in full
A small diagnosed error can reveal more analytical skill than an unexplained large model. For example, show that a ratio-ranked choice is feasible but not optimal in a whole-project budget. Or show how a kilogram number was mislabeled as tonnes. Keep the original statement and the corrected reasoning side by side.
| Highest-ratio-first allocation | Enumerate the small feasible subsets | The rule fails for this indivisible case |
| Lower intensity described as lower total | Recalculate both numerator and ratio | The two changes move in different directions |
| Imported rate called an IRR | Substitute it into the original equation | It is not a root of that schedule |
Answer the question about what you have not proved
An exact calculation on synthetic data does not establish that a real project should be financed. A recorded emissions product does not certify inventory compliance. A study exercise does not grant a professional designation. These limits can be stated precisely without dismissing the value of the work.
Separate data-quality uncertainty from model scope. You may know the arithmetic is reproducible while leaving the factor applicability unresolved. Or you may have trustworthy supplied amounts but a simplified feasible set. An interviewer can assess judgment more clearly when those distinctions are visible.
Link the artifacts so the explanation is reproducible
Use the model review log or factor-source sheet beside the displayed table. The companion version-control article explains how to retain the source and correction history.
Present the work as an original analytical demonstration, with authorship and source roles accurately described. Let the evidence support the claim rather than adding an unsupported title, assurance badge or commercial-success story.
Frequently asked questions
Must a work sample use a real client dataset?
No. An original synthetic case or appropriately sourced public data can show the method while protecting private information.
Does admitting a model limitation make the sample useless?
No. A precise statement of what is and is not established helps a reviewer assess judgment and reproduce the useful part of the analysis.
Sources
- CFA Institute: capital investments and allocation · cfainstitute.org · Retrieved
- GHG Protocol: published Scope 2 Guidance · ghgprotocol.org · Retrieved
