Why query level and page level detail is left off a public proof page

The query and page breakdown is deliberately left off our proof page, and the page says so in its own words rather than showing a redacted screenshot. That costs us the most convincing slide available: the list of exact phrases a site now appears for, and the individual pages carrying the clicks. It is also the one part of a Search Console export that belongs to the business whose searches produced it, which is why a stated omission is a better object than a blurred table.

What a query and page breakdown actually discloses

Totals describe a programme. Query rows describe a strategy, and page rows describe which assets are load bearing. Put them together and you have published, for anybody who wants it:

• The commercial phrases a business has decided to compete on, in priority order by volume.
• Which of those it is winning and which it is not, which is a map of where to attack it.
• The individual pages doing the work, so a competitor can read them rather than guess.
• Enough of the demand shape in a niche to rebuild the keyword research without doing any.

None of that is sensitive in the legal sense and all of it is valuable. The question is not whether it may be published. It is whether the party who benefits from publishing it is the same party who created it.

Why a redacted version is worse than an omission

A blurred or part redacted table invites a reader to do two things at once: trust the figures they cannot read, and assume the redaction was applied evenly. Neither is checkable. A stated omission asks for less and gives more, because the reader knows exactly which part of the record is absent and can weigh the rest accordingly.

It also avoids the specific failure where a redaction is incomplete. A table with the queries removed and the pages left in is not redacted, since a page slug usually names the query it was written for.

There is a quieter version of the same problem in the opposite direction. A table that has been thinned rather than blurred, with a handful of rows shown and the rest dropped, looks like a sample and behaves like a selection. Nobody reading it can tell whether the visible rows are the top of the list, a representative spread or the only ones that flatter the claim, and the author rarely says which. Shown rows from a withheld table are the least checkable object in this whole category, and they are more common than either a full table or an honest omission.

What is published instead, and why it is enough

Published	Withheld	Why the published half still supports the claim
Totals for the window	Per query totals	A total cannot be assembled from a flattering subset of queries
Daily and monthly series	Per page series	The shape of the result is visible without naming the assets
Average position	Position per query	Enough to show whether clicks could plausibly follow
Window and export dates	Nothing	Every figure is scoped and datable as published

The last row is the one that makes the first three usable. A withheld breakdown beside unscoped totals would be a weaker document than either half alone, because then nothing on the page can be checked and the omission starts to look like the reason.

The rule we work to, and where it is written down

Publish the aggregate, name the window, name the export date, and withhold anything that is a competitive asset of the business the data came from, with the withholding stated. Those decisions belong in writing before the first case study is drafted rather than being settled page by page, which is what a claims and evidence clause in an internal policy is for. A marketing policy you can copy section by section has example wording for both the claims clause and the customer data clause, and the two together cover most of this.

How an engine treats a self published figure

It is worth being clear about what a disclosed omission costs in the place these pages get read. Two of the things ChatGPT and Gemini asked for most often, across the 102 build spec conversations of 15 October 2026, were a disclosed methodology at 99 and the underlying data published so the work could be run again at 78. A withheld breakdown sits awkwardly against the second of those, and the resolution is not to publish it anyway. It is to name what is absent, so a reader is weighing a stated limit instead of discovering one. Measurement and the writing that follows it is what we sell, which makes our own pages the first place that trade off has to be made where anyone can see it.

Questions about publishing client data in a case study

Can I publish Search Console figures for a client at all?

With their written agreement, and it is worth asking for that agreement at the level of detail rather than in general. A client who is happy with totals is often not happy with their query list, and those are two different permissions.

Is anonymising the client enough?

Often not. A query list in a narrow niche identifies the business to anybody in that niche, so removing the name removes the attribution and not the disclosure.

What if the figures are from my own domain?

Then the decision is yours alone, and the same reasoning still applies, because the breakdown is a competitive asset whoever owns it. Our own published totals come from properties we own outright, and that ownership is what made printing raw figures possible in the first place.

Does withholding the detail weaken the case study?

It narrows what the case study claims, which is different from weakening it. A total with a window and an export date is checkable in the way that matters: the claim cannot be assembled from a selected part of the record.

What should the page actually say?

One sentence, in the source note, stating that query level and page level detail is left off on purpose. Readers who care will notice the sentence. Readers who do not will not have been misled by a blur.

The slide we chose not to make

Somewhere in the same export is a table that would be more persuasive than anything on the published page: the phrases, in order, with the clicks beside them. Knowing it exists and not using it is a more informative fact about how we work than the table itself would have been, and it is a fact a reader can verify by noticing what is absent. The strongest evidence available is not always the evidence it is yours to publish, and saying which part you are holding back is the closest thing to a substitute.