The claim: one client's result proves the system
A case study can demonstrate that something happened under a particular set of conditions. It rarely establishes how often the result happens, whether the provider caused it or whether a different buyer can expect it. Those are additional claims requiring additional evidence.
Begin by identifying which outcome is actually shown: a ranking, sessions, attributed conversions, positive replies, booked meetings or collected revenue. A screenshot of one unit should not be treated as proof of another.
Ask for the missing denominator
If a provider presents ten successful clients, ask how many comparable engagements were attempted. Ask which time window, eligibility criteria and exclusions produced that selection. Without that context, the examples show possibility rather than a reliable success rate.
A result can also be affected by an existing audience, a new sales team, a pricing change, seasonal demand or another campaign. A before-and-after comparison is useful, but it does not isolate every possible explanation. Distinguish observed association from demonstrated causation.
Original arithmetic: a large percentage can conceal a small base
Suppose a hypothetical site increases organic clicks from five to 25 in a comparable period. That is a 400% increase, representing 20 additional clicks. Another site rises from 2,000 to 2,500, a 25% increase representing 500 additional clicks.
Neither is automatically the better commercial result. Relevance, inquiries and costs still matter. The point is that a percentage without starting and ending counts is difficult to interpret. Show both, and state whether the periods and measurement definitions match.
Reconstruct a search or AI example
For SEO, request the exact query, market, destination, dates and result type. Identify whether branded or invented terms dominate the wins. For AI answers, retain the full prompt, platform context, date and saved answer. A brand-seeded question should remain visibly separate from unbranded discovery.
For advertising, request spend and the conversion action definition. An imported event and a platform-attributed conversion may not represent the same timing or person as a CRM record. For revenue, reconcile with the authorized commercial system rather than infer a sale from an inquiry. FTC: advertising guidance for small business.
A case-study evidence packet
- Starting and ending counts, with a consistent period and scope.
- The actual intervention, implementation date and relevant concurrent changes.
- Source exports or inspectable records, with private information protected.
- Definitions for each outcome and the treatment of duplicates or invalid records.
- Costs and operational responsibilities where the claim concerns profitability.
- The selection method and known limitations of the example.
These requests do not require a vendor to disclose another client's confidential data. Redacted evidence, an authorized demonstration or appropriately aggregated comparable results can be useful. If the necessary information cannot be supplied, keep the conclusion narrower.
Apply the same standard to ZeroDark
A SharkBlade demo screenshot illustrates product behavior with synthetic data. It is not a case study proving that a real client achieved those results. Any future ZeroDark performance story should identify its authorized evidence, scope and limitations rather than borrow credibility from a polished interface.
Thought leadership requires a standard that survives being applied to our own claims. A useful case study helps the buyer understand the conditions for success, not just admire the strongest number.
Supporting documentation
FTC: advertising guidance for small businessGoogle: Search Console performance reportGoogle Ads: conversion measurement