Find the pages Google shows and searchers skip

Picture a page Google showed 4,000 times last month, in third place. Sixty people clicked. At third place, RankHive’s reference curve expects about 400.

That page has done the hard part. It ranks, people see it, and they choose another result. The reason is usually in the few words Google shows, the title and the snippet, or in a page that answers a different question from the one searched. It’s rarely a missing backlink. That makes these the cheapest clicks you can win: a new title takes minutes, while moving a page from ninth to third can take months.

A low CTR on its own is not the signal: a page at position 14 is expected to get few clicks. The report compares each page with what its own average position normally earns, then puts a number on the gap so you can see which pages deserve your time first.

The report reads Search Console’s search analytics for web search over the 28 complete days that end 3 days before today. Search Console’s newest days are often incomplete, so they are left out rather than letting a half-counted day drag a page’s numbers down. Dates are in Pacific Time, so the window can look one day off from where you are.

For every page, the report takes the clicks, impressions and average position for the window. It then looks up an expected CTR for that position on the same reference curve the CTR calculator uses by default: 30% at position 1, 16% at 2, 10% at 3, 7% at 4, 5% at 5, 4% at 6, 3% at 7, 2.5% at 8, 2% at 9, 1.5% at 10, and 1% from 11 to 20.

These are round numbers RankHive chose as a starting point, not a measurement of your site or the figures of any one published study. Real rates vary with the query, the device and what else is on the results page. Positions between whole numbers are interpolated in a straight line, so 6.4 expects 3.6%.

The gap for a page is its impressions multiplied by the difference between expected and actual CTR: missing clicks = impressions × (expected CTR − actual CTR). Only a positive gap counts. A page doing better than the curve scores zero rather than a negative number, so one strong page cannot hide a weak one in the total.

Pages are ranked by missing clicks, with impressions breaking ties. The report shows the top 20, each with up to 5 of its queries ordered by impressions, and the headline counts every problem page, including any past the top 20. The three title fixes are for the top three pages in that order.

Treat missing clicks as the size of an opportunity, not a forecast: what the problem pages would have earned in the same 28 days had each matched the curve. Some will close most of their gap after a better title; some will barely move, because the cause was never the title.

Average position is the biggest source of error. Search Console records the topmost position your site held for each impression and averages it, so a page that ranks first for a small, specific search and fourteenth for a big general one can show an average around 5 while almost all of its impressions come from position 14.

That is why every problem page lists its real top queries: read them before deciding what to change. The rarest queries are also left out of query-level data for privacy, which is why a page’s query rows usually add up to less than its own totals.

Page

The URL as Search Console reports it. A domain property covers every host (www, non-www, subdomains), so a path can appear twice if both versions are indexed, which is worth fixing in itself.

Impressions and clicks

How often the page appeared in web search results in the window, and how many of those led to a visit. Impressions decide what a gap is worth: the same CTR shortfall costs ten times more on ten times the impressions.

CTR

Clicks divided by impressions for the page as a whole, across all of its queries.

Average position

The average of the topmost position the page held each time it was shown. Lower is better. Because it is an average over every query, read it together with the queries underneath.

Expected CTR

What the reference curve expects at that average position. It is the yardstick, not a target you are guaranteed to reach.

Missing clicks

Impressions × (expected CTR − actual CTR), rounded. The list is sorted by it, and it adds up to the headline.

Top queries

Up to five searches the page was shown for, by impressions, each with its own numbers. This is where you tell a title problem from an intent problem or a brand mix.

The numbers below are illustrative, made up for a fictional gardening blog to show the arithmetic. They are not data from a real site and not a benchmark.

The pruning guide ranks third on average and earns 1.5% of its 4,000 impressions. Position 3 expects 10%, so it is missing 4,000 × (10% − 1.5%) = 340 clicks. The colours page sits at 6.4, which expects 3.6% (between 4% at 6 and 3% at 7); at 1.6% it is missing 2,500 × 2% = 50.

The about page looks healthy at 13.3%, but at position 1.8 the curve expects 18.8%, a gap of about 49 clicks that the report shows as 50. A compost bin review with 80 impressions and a tools page averaging position 24 are not listed at all: one has too little data, the other has a ranking problem.

The headline would read: 3 pages are missing clicks · about 440 extra clicks a month available · 3 fixes ready. The total comes from the unrounded gaps (340 + 50 + 49.2), then rounded once.

Now the part numbers cannot do. The pruning guide’s top query is “when to prune hydrangeas” and its title is “Hydrangea Care, Part 3”: a clear title problem. On the about page, the blog’s name ranks first with a high CTR, and “gardening blog uk” ranks fourth with almost no clicks. That searcher wants a list of blogs, not one blogger’s biography, and a new title would not change that.

You can reproduce the report’s first step in Search Console itself. It takes longer, and the menus below may be labelled slightly differently as Google updates the interface.

1. 1.Open Search Console, choose the property, and go to Performance, then Search results.
2. 2.Set the date range to the last 28 days, or a custom range ending a few days ago, with Search type set to Web.
3. 3.Turn on the Average CTR and Average position cards so those columns appear in the table.
4. 4.Open the Pages tab. Use the table’s filter to keep rows with more than 100 impressions and a position under 20, then sort by impressions.
5. 5.Compare each page’s position and CTR, and note the ones far below what that position usually earns.
6. 6.Click a page to filter the report to it, then open the Queries tab to see which searches it is shown for.

Search Console will not compare each page with an expected CTR or total the gap. For that, export the Pages table and upload it to the CTR calculator, which runs the same comparison in your browser without an account. For what counts as a good CTR in the first place, see What Is a Good Click-Through Rate for SEO?.

Each cause starts with what it looks like in the report, then how to confirm it and what to do. Confirm by searching the top query yourself in a private window, on the device most of your searchers use: the results page is the evidence the report cannot see.

Your report lists each page’s top searches, so you can usually tell which of these it is in a minute.

Google builds the title link mostly from your title element, but can also use the main heading, other prominent text, or the anchor text of links to the page. A title that clearly describes the page is more likely to be shown as written, and clicked.

• Start from the page’s real top query, not the keyword you hoped it would rank for. Put its main words near the beginning.
• Say what the searcher gets: the answer, the list, the price, the calculator, the steps. Specific beats clever.
• Keep the important part early. Google shortens long title links to fit the results page and sets no fixed character limit, so 50 to 60 characters is a rule of thumb, not a rule.
• Make the title, the main heading and the opening of the page agree. A title that promises more than the page delivers may get clicks once and lose the ranking later.
• Put your brand at the end, short, and avoid repeating the same phrase in every title on the site.

Three before-and-after examples, again illustrative, drawn the way the title and meta description checker previews a Google result on a phone:

If you want more candidates to choose from, the free SEO title generator drafts several, and the title and meta description checker shows what the live page currently serves. On WordPress, the title usually lives in your SEO plugin’s field rather than the post title; see How to Change Your WordPress SEO Title.

A gap in the report is a reason to look, not an instruction to edit. Besides the causes above that a title cannot fix (a sitelink, a different intent, a crowded results page, a brand mix), leave the title alone when:

• The page changed in the last few weeks. Its window mixes the old and the new; wait for a clean 28 days.
• The page has a seasonal peak and the window caught the edge of it. Compare the same weeks of last year first.
• The page is about to be merged, redirected or rewritten. Fix the title as part of that work, not before it.
• The current title already names the top query and Google shows it as written: look at the snippet and the intent instead.

Google has to recrawl the page before searchers see the new title. For an important page, ask for that with Request Indexing in Search Console’s URL Inspection.

Once the new title is showing, wait for a full 28-day window and compare it with the 28 days before the change, using Search Console’s compare dates option filtered to that page. Look at the same top queries, and check that their average position stayed roughly where it was: a CTR that rises because the page also moved up is a ranking change, not a title win. A fresh report from this tool after the window shows whether the page has left the list.