Case Study

A case study is an account of one piece of work for one client: the situation before, what was done, and what measurably changed. It is evidence in narrative form, and its worth depends on whether the figures in it can be checked.

Case study or testimonial

A testimonial is an opinion about the work. A case study is a record of it: a before figure, an after figure, the dates each was taken, and the method that produced them. The client may be anonymised if the method and the figures are real. What a case study may not be is assembled from a template with the numbers filled in, because the first thing a careful reader does is ask where a number came from, and a number with no origin fails that test on sight.

Why the figures in one get quoted

Aggarwal et al., “GEO: Generative Engine Optimization” (KDD 2024), altered source pages by adding statistics, quotations and citations, and measured a rise in how visible those sources were in generated answers. A case study is where those three things occur naturally on a business site: a specific number, attributed to a named piece of work, with a date. A services page that says the work is excellent gives an assistant nothing to lift. A page that says what changed, by how much, over what period, gives it a sentence.

What this site has published

No client case study. This site’s one published study is an audit of what 9,483 small-business websites serve to an AI crawler, a sample-level finding rather than an account of one engagement; it is at what 9,483 small business websites serve to an AI crawler. A case study on a client’s visibility would need a full question panel run before and after the work, with both readings kept, and will be written from that record or not at all.

A common question

Can a case study keep the client anonymous and still count as evidence?

Yes, as long as the method and the figures are genuine. What makes the account evidence is the reading taken before, the reading taken after, the dates of each, and the description of how both were produced, none of which depend on naming the buyer. The failure case is the opposite one: a template with numbers dropped into it, because the first thing a careful reader does is ask where a figure came from, and a figure with no origin does not survive that question.

Sources

Source-side statistics and visibility: Aggarwal et al., “GEO: Generative Engine Optimization” (KDD 2024). The audit named above is this site’s own; no figures from it are quoted on this page.