Produce the same file our proof page is built from, on your own domain

The file behind the clawlaw.in case on our proof page is a Google Search Console Performance on Search export. It is one screen, four settings and one download. The window on that export is 2026-07-13 to 2026-10-12, 92 days with data, and the file was pulled on 2026-10-14. It contains 18,537 clicks and 929,405 impressions. Nothing on that case was typed by hand, which is why it can be quoted, and the same screen on your own verified property produces the equivalent file for your own site in about five minutes.

Before the screen: the one prerequisite

You need a verified property in Search Console for the site you are measuring. Verification is ownership, not access, so there is no version of this that works on a domain you do not control. If a supplier has been reporting your search figures and you have never seen this screen, the first request is to be added to the property rather than to be sent another slide.

Choose the property type deliberately. A domain property covers every subdomain and both protocols. A url prefix property covers exactly the prefix you entered. Two properties on the same site return different totals for the same window, and that difference is not an error in either of them.

The four settings on the Performance report

• Search type: Search. The report also carries Discover and News, and those are different surfaces with different behaviour. Performance on Search is the one a published search result figure comes from, and naming it is part of the claim.
• Date range: custom, with two dates. Pick the start and the end explicitly rather than using a preset. A preset is relative to today, so a chart made from one cannot be reproduced tomorrow.
• Metrics: all four on. Total clicks, total impressions, average click through rate and average position. Turning on all four is what stops the chart becoming an argument for whichever one looked best.
• The Dates tab. Switch the table under the chart to Dates. The row count is your days with data, and it is the divisor behind every daily figure you are about to publish.

The export, and the field to write down immediately

The export control sits at the top right of the report and offers a spreadsheet, a comma separated file or Google Sheets. Any of them is fine. What matters is what you do in the next thirty seconds: write down the date you pulled it.

That field does not appear in the file. A performance export does not include the most recent day or two, and figures for recent days continue to settle for a short period afterwards, so the same window pulled a week later returns slightly different rows. A published figure without an export date is a figure nobody can reconcile against a later pull, and the discrepancy will be read as an error rather than as settling.

Turning the export into a page somebody can check

Four things go above the chart, in this order, and they are all in your hands by now.

• The two dates of the window, written out.
• The day count from the Dates tab, labelled as days with data.
• The export date.
• The name of the export: Google Search Console, Performance on Search.

Then publish the daily table underneath the chart rather than only the chart. A chart shows a shape and a table shows the rows, and a reader who distrusts the shape can check it without writing to you. This is the cheapest credibility on the page and almost nobody does it.

Generate the figures in the text from the file rather than retyping them. A proof page whose sentences are typed by hand drifts from its own data the first time somebody updates the chart, and then the page is wrong in a way no gate will catch.

What to do with the file once you have it

The same export answers questions beyond a proof page. The clicks and impressions columns per position band give you your own click share curve, which is more use than any published one, and building that curve from your own Search Console data takes about an hour from the file you have just downloaded.

If nobody on the team has worked this screen before, the sequence above is learnable in an afternoon and is worth learning in order rather than by searching for each step as it arrives. Where that needs to be somebody's actual skill rather than a one off, picking a class that teaches the work rather than the vocabulary is the difference between a person who can pull this file and a person who can read a slide about it.

The reading this export cannot give you

Everything above exists because Google records the data and hands you a file. The same buyer question asked of ChatGPT, Gemini, Perplexity or Claude produces one written answer naming two or three companies, with no results page, no position and no export behind it. There is no screen to visit, so a figure about that surface exists only if somebody asks the question and records what came back, with the date and the apps named.

That record is the product we sell, and the reason this page gives away the Search Console half for nothing is that the two measurements answer different questions and a business needs both.

Questions about pulling the Performance on Search export

What is the earliest start date the report will accept?

Sixteen months before today, and that limit moves forward as time passes, so the oldest months fall out of reach. A year on year comparison therefore depends on having exported and kept the file at the time rather than on the console still holding it.

Which figure should the published total come from?

The chart, not the query table. Rare queries are withheld to protect the people who typed them, so the per query rows are a subset by design and will not reconcile to the headline. Take the total from the chart and use the query rows as explanation underneath it.

Does the export include AI Overview clicks?

A click on a link inside an AI Overview is counted as a Google click and filed with the others, so it is in your total and is not separable in the report. That is a known limit of the file rather than something a different setting will fix.

Should I publish the file itself or the figures from it?

Publish the figures and the daily table, and keep the file ready to send. 78 of 102 build spec conversations run on ChatGPT and Gemini on 15 October 2026 asked for the underlying work to be published so it could be re-run, and being able to send the export on request is what satisfies that without putting a spreadsheet on a marketing page.

What if the figures are small?

Publish them with the day count attached. A small figure over a stated window is a measurement, and the credibility it buys on the rest of the page is worth more than the figure costs. A page that only shows its good columns is read as a page that has other columns.