Your Search Console data, already read

Sixteen months of your own search data, the queries that rose and fell already named, and five kinds of opportunity computed from numbers Google measured, not from an invented benchmark.

Analyze your website free The whole product

Uses Google Search ConsoleFree plan: one website of up to 900 pages, re-analyzed every month. No card.

The Search Performance screen: 90 clicks, 4,023 impressions, a 2.2% click-through rate and an average position of 10.7 over 28 days, each compared with the 28 days before, above a list of what the numbers mean. — Search Performance: Search Console's figures, compared and explained.

The problem

Search Console has the answer, but will not volunteer it

Search Console is the most honest data a website has about itself, but using it well means exporting sixteen months, pivoting it and comparing periods by hand. Most teams look at the last 28 days, see clicks are down, and stop there.

The useful questions are all comparative. Which queries moved. Which pages earn impressions but no clicks. Where the site sits on page two for something it nearly ranks for. None of those are on a default screen.

How it works

The first read is everything Google has

Connecting takes one Google OAuth grant. The first read covers sixteen months, everything Search Console keeps, fetched a month at a time so no single request grows with the window. It is lag-aware: the page states the range it actually read rather than implying it is current to today.

After that, a daily refresh re-reads the last ten days, the part Google still revises. The rest is stored and final. Your Google connection is read-only and encrypted at rest.

The opportunities are the point. They fall into five kinds, and each is arithmetic over measured numbers. That is why the click-through test stops at the first page of results instead of telling a query at position 65 to rewrite its title.

The figures

Clicks, impressions, click-through rate and average position, against the previous period
A daily trend chart with a range picker
Top queries and top pages, sortable, searchable and paginated
Country and device breakdowns
Position spread: how your queries distribute across the results

What moved

The queries that rose and the queries that fell between two periods
Which pages appear for which searches over 90 days, replaced on every sync
The comparison that Competing Pages and search-based link text both read from

Five kinds of opportunity

Click-through: ranking well, being ignored
Ranking: close enough to the first page to be worth the push
Visibility: impressions without the position to convert them
Emerging: queries that were not there last period
Declining: queries going the other way, before the traffic does

Good to know

It is optional. Unconnected, it is an empty state explaining what connecting unlocks, and everything else in the product keeps working.
It reports the range Google actually returned, including the lag, rather than implying today's numbers exist.
It does not estimate, model or fill gaps in Search Console data.

Questions

About Search Performance

Do I have to connect Search Console?

No. The product works without it: the audit, the content inventory and the link analysis all run on the crawl alone. Search Performance is an empty state until you connect, and it explains what connecting unlocks rather than pretending to have data.

What happens to my Google tokens?

Access is read-only and encrypted at rest. You can disconnect from Settings at any time, which deletes the stored data, and removing access from your Google account works too.

How far back does it read?

Sixteen months on the first sync, which is everything Search Console keeps. It is read a month at a time so the request never grows with the window, and after that a daily refresh re-reads the last ten days, which is the part Google still revises.

Why do the totals not match the sum of the rows?

Because Search Console counts anonymized queries in the totals and omits them from grouped rows. That is Google's behavior, not a bug, and the product does not reconcile the difference by inventing rows to cover it.

Keep exploring

All nine parts