How case studies on this site are built
The same evidence standard applied to statistics and research on this site applies here — including what happens when the honest answer is a range, an estimate, or nothing disclosable at all.
Case studies are where the temptation to round a good month up, cherry-pick a favourable period, or quietly blur "we did this work" into "this happened because of us" is strongest — precisely because nobody but the business and the person writing it up can easily check. This page is the standard we hold ourselves to, so that check is possible in principle even when it isn't made in practice.
Every figure is labelled by its actual source
- Client-reported — a figure the client told us directly (revenue, bookings, jobs won), which we haven't independently verified against their systems.
- Platform-measured — a figure pulled directly from Google Search Console, Google Business Profile Insights, analytics, or ad platform reporting, which we can see and verify ourselves.
- Modelled or estimated — a figure calculated from other known numbers (for instance, estimating revenue impact from a measured traffic and conversion rate change) rather than observed directly.
A number appears with its label attached, not buried in a footnote. If a client is comfortable disclosing platform-measured rankings and traffic but not internal revenue figures, the case study says exactly that rather than inventing a plausible-sounding number to fill the gap.
Before, work, and after — kept genuinely separate
Every case study follows the same three-part structure, and the sections aren't allowed to blur into each other:
- Before — the starting position: rankings, visibility, leads, conversion rate, whatever's relevant and measurable, at the point work began.
- Work performed — specifically what was done, in enough detail that the connection to the result is checkable rather than asserted.
- After — the resulting position, over a stated time period, with the same figures measured the same way as the "before" section.
Where other factors plausibly contributed to a change — a seasonal shift, a competitor closing, a broader market trend — that's stated rather than left for the reader to assume the case study's work was the sole cause.
What we never do
- Invent a result. If a client's outcome isn't strong enough to be worth publishing honestly, it doesn't get published with the numbers improved.
- Present a single client's results as an industry benchmark. One business's outcome is one business's outcome — see our research methodology for how that differs from a genuine benchmark, which needs a real sample across many businesses.
- Claim credit for something we didn't do. Where a client made changes to their own site, staff, or offering alongside our work, that's disclosed rather than folded silently into "our" result.
- Publish a result without the client's informed agreement on what's being disclosed and how it's described.
If you're a client wondering how your results will be presented
You decide what's disclosed and what isn't. A case study can run with platform-measured figures only, with ranges instead of exact numbers, fully anonymised, or not published at all — any of these is a legitimate outcome, and none of them requires inventing a number to make the story feel more complete.
This is the standard every published case study is held to.
Where a specific figure isn't disclosed, that's stated plainly rather than filled in — the same rule this site applies to its statistics and research.
Replies usually go out within one working day.