Guide

Read Search Console in ten minutes: four numbers, one trap

For someone who never opened Search Console: the four numbers that matter, the one that misleads, and two filters that turn a report into a decision.

SK Semih Kavaklıoğlu·8 Sept 2026·3 min read Share

Search Console is free, first-party and mostly unread. People who build products open it once, see a chart with four coloured lines, and close it. This is the ten-minute version: which numbers to read, in what order, and the two filters that make them mean something.

Which four numbers should I read first?

Answer. Clicks, impressions and average position over 28 days against the previous 28, and the share of clicks that are not your brand name.

Clicks
145
▲ +31%, was 111
Impressions
2,632
▲ +338%, was 601
Position
11.1
▲ from 13.2
Non-brand clicks
11
▲ +10, was 1

Clicks are visits from Google. Impressions are how often you were shown at all; they move first when SEO work starts to land, weeks before clicks do. Average position is where you were shown, weighted by impressions; under 10 is page one. The fourth number is not on the default screen: it is clicks from queries that do not contain your name, and it is the only one that says whether new people are finding you.

Set the date range to "Last 28 days" and tick "Compare" to the previous period. Search Console data lags two to three days, so the range ends a few days ago; that is normal.

Which number is the trap?

Answer. Site-wide click-through rate. It falls when you grow, because impressions grow faster than clicks.

The worked numbers above produce a CTR that dropped from 18% to 6% while everything else improved. The site started being shown for four times as many queries, mostly at positions 8 to 20 where few people click. Read per query, at a stable position, CTR is useful. Read as an average across a growing site, it is noise that looks like a warning.

What are the two filters that turn the report into a decision?

Answer. Split brand from non-brand with a query filter, and sort non-brand queries by impressions to find the ones you rank for but do not win.

First filter: Query, "Custom (regex)", "Doesn't match", and a pattern with your brand name and its misspellings. Everything left is non-brand: the demand you are competing for. On the example site non-brand impressions went from 191 to 895 in a month while brand clicks slipped from 91 to 77; the two facts point at different work.

Second filter, on the non-brand view: sort by impressions and scan the top rows for two shapes.

Within reach
position 8 to 20
shown, rarely clicked
impressions high
demand exists
what to do
make the page answer the query in its first sentence and title
Absorbed
position 1 to 7
should be clicked
clicks near zero
something above you takes the click
what to do
search it yourself; if an AI Overview or carousel sits on top, become the cited source or stop investing

That is the whole method. Everything else in Search Console (device, country, search appearance) refines it.

What about the Pages report?

Answer. Open Indexing, Pages, once a month, and read one row: "Crawled, currently not indexed".

That status is Google saying it read a page and chose not to keep it. A few login and tag pages there is fine. Guide or product pages there is lost traffic, and the reasons are usually thin or duplicated content, not technical errors. Errors and "Excluded by noindex tag" rows are worth a glance too; a sitemap URL under the noindex row means your sitemap and your page disagree.

How does Moonleap use the same data?

Answer. With a read-only connection it does this pass for you every Monday, keeps the brand split, and flags the absorbed queries as findings your agent can act on.

The weekly report carries the four numbers with the previous period, the brand and non-brand split, the top within-reach queries and the count of absorbed ones. The content-ideas engine uses the non-brand query list to propose pages, and holds a proposal back when the page it targets changed in the last 30 days, so you do not measure two changes at once.

FAQ

Do I need Search Console if I have Google Analytics?

Yes. Analytics sees people who arrived; Search Console sees the searches where you were shown and not clicked, which is where the work is.

How long until a change shows up?

Two to three days of data lag, then however long Google takes to recrawl the page. Read the result at 28 days, not before.

Can I connect it read-only?

Yes. Verification gives you ownership; a tool like Moonleap asks for the read-only scope and never writes anything.

Sources

  1. Google, "Performance report (Search results)" https://support.google.com/webmasters/answer/7576553
  2. Google, "Page indexing report" https://support.google.com/webmasters/answer/7440203
  3. Worked numbers from omniroom.ai, 28 days to 4 September 2026 (our own Search Console export)

Share the number

4

numbers worth reading in Search Console before anything else, and one to ignore until you split it

moonleap.io · 8 Sept 2026Post

Related