A case study earns trust by showing what the evidence can and cannot support. A polished “before and after” without a clear measurement window, denominator, or confounding changes is a story—not proof of causation.
Build a claim-to-evidence ledger
Before writing, list every outcome claim and its source:
| Claim | Evidence needed | Safer wording if evidence is partial |
|---|---|---|
| “Qualified inquiries rose” | Same qualification rule, comparable periods, counted inquiries | “Recorded qualified inquiries were higher in the later period; other changes may have contributed.” |
| “The new landing page caused more sales” | A credible comparison design and outcome data | “Sales increased after launch; this comparison alone does not isolate the page’s effect.” |
| “The campaign generated revenue” | Deduplicated orders, cancellations/returns, attribution limits | Report observed revenue and how it was counted |
Record baseline, intervention, dates, market, sample size, what else changed, and who verified the numbers. Separate platform-attributed conversions from orders or inquiries confirmed by the business.
Protect the people and the truth
Get permission for names, logos, screenshots, and quotes. Remove personal data and details that could identify a customer indirectly. If client data is unavailable or cannot be shared, publish an explicitly labeled illustrative scenario or describe the method without claiming an outcome. Never invent a client, quote, metric, or approval.
Write the useful part
Explain the decision, the change, the measurement method, the observed result, the limits, and what another operator should inspect before copying it. A negative or inconclusive result can still help if it clarifies a decision boundary. Include enough context that readers can judge whether the case applies to their market; do not imply that one Egypt or KSA example proves the same effect across MENA.
AI disclosure: AI tools assisted with drafting. No client outcome is asserted in this guide.