Which of your pages get fewer clicks than they should?

The checker assumes roughly 28% at position 1, 15% at 2, 7% at 5, and 2.5% at 10, falling under 1% past position 16. Published CTR studies differ a lot, and your own rates vary with the query: brand searches earn more, and results under ads, maps, or an AI answer earn less. That's why the checker treats these rates as a starting assumption and flags only gaps too big to be chance.

Usually the title and description don't promise what the searcher wants. Or the result already answers the question, so there's nothing left to click for. A featured snippet, an AI answer, or a stronger brand above it can take the click too. And the position is an average across many queries: a page at 7 overall may rank 3 for some searches and 15 for most.

No. Your browser reads the file and never uploads it. Nothing reaches our servers and nothing is stored. Close the tab and it's gone.

For every page with at least 50 impressions, the checker takes the typical click-through rate for its average position, works out the clicks that rate predicts, and asks how likely your actual clicks would be if the typical rate held (a Poisson test). It lists the pages where that chance is under 5% and at least 2 clicks are missing, worst first. It's the same test Site Guard, our check for articles losing clicks, runs on the articles we write. Clicks on links to a section of a page (#jump links) count for that page. Their impressions don't, because Google counts one result once for the page and again for each section.

The last 28 days or 3 months. A longer range averages a page's position over months when it may have moved, and an old title you've since fixed still counts against it.

Not from the export. It lists queries and pages in separate tables, so it never shows which page answered which query. That needs the Search Console API. Once you connect Search Console, Site Guard checks every article we write for two articles competing for one query.