Case studies often fail in opposite directions. Some reveal almost nothing: a logo, a glowing sentence, and a claim that everything improved. Others borrow scientific language they cannot support, treating one project as proof that a tactic causes a universal result. Credible proof is more modest and more useful.
The seven-part evidence record
- Starting condition.What was happening before, for whom, and how was it known?
- Constraint.What budget, timing, systems, approvals, data, capacity, or market conditions shaped the work?
- Decision.What was chosen, what was deliberately not chosen, and why?
- Intervention.What actually changed in the website, process, content, campaign, software, or service?
- Observed outcome.What evidence exists after the change, over what period, using which definition?
- Contribution and uncertainty.What else may have influenced the result, and what cannot be claimed?
- Permission.Which names, images, numbers, quotes, and details did the client approve for public use?
Replace transformation language with comparison
“We transformed the brand” is difficult to inspect. A comparison is concrete: the previous homepage opened with three internal slogans and seven equal actions; the revised version names the service, intended customer, operating region, primary action, and one approved proof signal in the first screen. The reader can judge the change even without a revenue claim.
Useful evidence can be qualitative: fewer repeated customer questions, a clearer intake, an easier approval path, improved staff confidence, a specific accessibility barrier removed, or a process the client can finally operate. Name how the observation was collected and avoid converting an anecdote into a percentage.
When a number is allowed to speak
Define the metric, period, source, baseline, and denominator. “Leads increased 40%” is incomplete without knowing whether the count moved from five to seven, whether tracking changed, whether spend increased, or whether the comparison covers equal seasons. Use exact numbers only when the client approves them and the records support them.
“Conversions doubled.”
No definition, period, volume, source, spend, or qualification.
“Qualified consultation requests increased from 8 to 15.”
The case then states the 90-day periods, unchanged definition, campaign context, and limits.
Design proof for privacy and permission
Some of the best client work cannot be shown in full. Replace sensitive records with approved redactions, recreated diagrams, aggregated ranges, cropped interfaces, or a process explanation. Label synthetic or reconstructed visuals. Do not publish a customer’s dashboard, private communication, identity, location, or business result because it looks persuasive.
Keep an approval trail: artifact, intended context, approver, date, expiry or review condition, and any wording constraint. A testimonial approval does not automatically authorize project files, numbers, staff images, or a public logo relationship.
What to publish when outcome data is immature
Publish a field note rather than a finished victory story. Explain the problem, decision, build, verification performed, current status, and what will be measured next. OneZero’s evidence standard keeps verified client work separate from Studio Lab concepts and unverified claims.
Make the case useful beyond the trophy
A strong case study teaches the reader how to recognize a similar problem, which trade-offs matter, what preparation improves the project, and which signals should be measured. Link to the related service and a deeper article, but do not hide the practical lesson behind a contact form.
