- Fictional brief
- Setup and QA
- Worked calculation
- Incident write-up
Use this sequence to organize the investigation. Confirm the details in the affected platform and campaign.
A work sample is most useful when it exposes decisions rather than just a finished dashboard. Someone reading it should understand what you were trying to achieve, what you observed, and why your next action made sense.
This is a suggested portfolio structure, not evidence that every employer requests one. Keep the sample short enough to review and tailor its emphasis to the role you are applying for.
Start with a fictional brief and assumptions
Describe an advertiser, format, flight, goal, targeting, and measurement basis. Use fictional identifiers and mark the brief as simulated. Include unanswered questions rather than silently filling every gap.
A good assumption record might say that the plan uses complete calendar days in UTC and an even delivery curve. That makes later pacing calculations interpretable.
Include a campaign setup and QA sheet
Map each brief requirement to a setting or check. Show an example mismatch you identified and how you would test the correction. Do not fabricate a screenshot from a platform you have not used.
- Requirement from the brief.
- Proposed configuration or QA step.
- Evidence available.
- Open question and owner.
- Verification after the proposed correction.
Show a reproducible delivery calculation
Include goal, elapsed time, actual delivery, planned delivery, and required remaining rate. Show the formula, not only a percentage.
For a 150,000-impression goal over 15 equal days, 40,000 delivered after five days is 80% of the 50,000 planned to date. The remaining 110,000 over ten days requires 11,000 daily. Explain that this is a requirement, not an available-inventory forecast.
Write an incident report with a rejected hypothesis
Present a short evidence packet and an initial hypothesis. Then identify the observation that supports or rejects it. This shows how you update your reasoning.
For example, a bid arrives before the ad server request, but its targeting is attached afterward. The timeline contradicts a simple late-bid hypothesis and supports a handoff investigation.
State the limits of the sample
Label simulator outputs, illustrative logs, and authored calculations. Attribute the documentation and exercise that informed your work. Do not present fictional results as client improvements, measured revenue gains, or certifications.
For any real work sample, obtain permission and remove confidential campaign identifiers and personal information. If permission is unavailable, construct a clearly fictional case instead.
Finish with a concise reviewer summary
Use four sentences: the task, the evidence, the decision, and the verification still required. Add one thing you would investigate next. Keep the detailed notes available below.
Before sharing, ask whether every number can be reproduced and every claim has an appropriate label. A small, clear sample is more credible than many screenshots with unexplained metrics.