The Page-2 Problem: Why Your Impressions Get No Clicks, and the Fix

Seen 1,745 times in Google, clicked 5 times

We published eleven careful articles about putting AI to work. Ninety days later, Search Console said Google had shown them 1,745 times.

People clicked them 5 times. Five. One click for every 349 appearances.

The articles were fine. Their page 2 Google ranking was the problem. They sat on pages two to four, where almost nobody looks.

So lots of impressions and almost no clicks means your page is listed too low to be seen.

More articles will not fix it. You fix the pages Google already shows, in five steps, and you judge them after 90 days.

We call those five steps the Page-2 Playbook. Each step below comes with the numbers we used on our own pages.

Why do I have impressions but no clicks?

You have impressions but no clicks because your link is listed where searchers do not look. An impression is not a person reading your link.

Search Console counts an impression each time your site appears in the results. A link on a page the searcher never opened still counts.

Average position is the spot of your highest link, averaged over many searches. Roughly ten regular results fit on one page.

So position 18 is page two. Position 29 is page three. Position 35 is page four.

Our three most-seen articles sat at 35, 18 and 29. Together they drew 1,457 impressions and 5 clicks.

We call this the Page-2 Problem. Your report says you are visible. The searcher never knew you existed.

Why does it feel like progress? Impressions keep climbing. A climbing number looks like growth, even when nobody walks in.

The Lean Analytics authors, Alistair Croll and Benjamin Yoskovitz, argue that the most useful numbers are rates, not counts. A rate tells you whether to speed up or slow down.

Impressions are a count. Clicks divided by impressions, your click rate, is the number that tells you to act.

A big count on a page nobody visits is a billboard in a locked basement.

How much traffic does page two of Google get?

Page two of Google gets almost no traffic. In one study of 4 million search results, fewer than 1 in 100 searchers clicked anything on page two.

The exact figure was 0.63 percent. The top result alone took 27.6 percent of the clicks.

Our own data agrees. Queries where we ranked between 21 and 50 drew 2,012 impressions. Zero clicks.

Page one is shrinking too. When an AI summary sits at the top, people click fewer regular links, as Pew Research Center found by tracking the real browsing of 900 adults.

The size of that drop keeps moving, so the latest figures sit in the dated box below. The direction is what matters. Fewer clicks reach the regular links.

Page two was near zero to begin with. More searches now end before anyone scrolls, so where buyers ask the question matters as much as where you rank.

Close enough? Not even close. Page two is the storage room of the internet.

Should you write more content or fix old content?

Fix old content first when Google already shows it. Write new content when Search Console shows almost no impressions.

The check takes two minutes. Open the last 90 days and look at impressions per page.

Hundreds of impressions and almost no clicks means a Page-2 Problem. Google has already matched the page to real searches. It just ranks it too low.

Our reasoning: a page Google already matches starts ahead of a page that does not exist yet. Moving it from 18 to 8 is a shorter trip than going from nowhere to 8.

Almost no impressions is a different problem. There is nothing to move up, so you need pages, starting with searches you can win and that pay.

So why does most advice say publish more? Publishing is easy to sell and easy to count. Our take: a content calendar is the industry's favorite way to look busy.

How to stay on page two forever in four easy steps:

1) Publish a new post every week.

2) Never open Search Console.

3) Give every post a clever title nobody searches for.

4) Link to nothing. Wait.

That is not a strategy. It is a hobby with a hosting bill.

Being outranked is not permanent. In Hacking Growth, Sean Ellis and Morgan Brown tell how Zillow lost traffic to a rival, Trulia, that ranked its listings higher in Google.

Zillow made search the whole company's focus for its 2008 growth push. Every team found ways to improve. Zillow caught and passed Trulia, and seven years later bought it for $3 billion.

Our reading: ranking got fixed when it became everyone's job for a set stretch. Not a side task. The main one.

Steps 1 and 2: find the seen-not-clicked pages and retitle them

The Page-2 Playbook is five steps run on pages Google already shows: find, retitle, link, deepen, then date and wait. Steps 1 and 2 take one afternoon.

Step 1: find the pages. In Search Console, open the Performance report, set the last 90 days, and sort pages by impressions.

List every page with 100 or more impressions and fewer than 5 clicks. Then click each one and note its top three queries and their positions.

Start with pages at positions 11 to 30. They are closest to page one. They move first.

Step 2: rewrite the title for the search. Google builds the blue link mostly from your page title. So the searcher's words go first.

Our top query for one article was "ai maturity assessment." Our title led with "AI readiness assessment." Close, but not what anyone typed.

Our rule: the query first, under 60 characters, and the same promise as the page. The clever headline can stay on the page itself.

No Search Console yet? Our free SEO audit tool gives you a first look while you set it up.

A clever title nobody searched for is a sign written for nobody.

Steps 3 and 4: link to each page and give it what the results above it have

Pages climb when other pages point to them and when they answer more than the results above them. Those are steps 3 and 4.

Step 3: link to it. Google finds and understands pages through links, and it reads the words inside each link.

Add at least two links to each listed page from pages that already get clicks. Use words a searcher would type. Never "click here."

Ours had almost no links. Now each article sits in a topic hub and links to its neighbors, the method in our hub-and-spoke SEO guide.

Step 4: deepen it. Search the top query yourself and read the three results above you. List what they answer that you do not.

Then add the missing pieces: a worked example, named sources, the doubt a buyer would raise. Google's guide to helpful content asks if a page gives substantial value next to other results.

Eight of our eleven had four sections and few sources. Each now has seven sections and twelve or more linked sources.

Proof of who wrote the page matters too. Our E-E-A-T checklist covers that part.

Thin pages do not get promoted. They get skipped.

Step 5: date it honestly, wait 90 days, and judge with a range

Change the updated date only when the page truly changed, resubmit your sitemap, and wait 90 days before you judge. That is step 5.

Google looks at several signals to estimate a page's date, including the date you show. Our rule: a new date means new content. Then resubmit the sitemap.

Before the 90 days start, write down what counts as a win. Croll and Yoskovitz call this drawing a line in the sand. You decide the action before you see the data.

Our line has two parts. The page reaches positions 1 to 10 for its top query. And its click rate clearly beats the old one.

What does clearly mean with 5 clicks? Small counts swing a lot. So we judge them with a range, not one number.

Douglas Hubbard and Richard Seiersen, in How to Measure Anything in Cybersecurity Risk, show a simple way to turn hits and misses into a range. You give the count one free hit and one free miss.

Your best single guess is then (1 + clicks) divided by (2 + impressions). For the range, a spreadsheet does the work: =BETA.INV(0.05, 1+clicks, 1+impressions-clicks) gives the low end, and 0.95 gives the high end.

Counting each impression as one try is our own use of the method, not the book's. The book counts hits and misses in security samples. The math is the same.

Our 5 clicks on 1,745 impressions give a range of about 0.15 to 0.60 percent. Say the next 90 days bring 12 clicks on the same impressions: 0.44 to 1.11 percent.

Those ranges overlap. Not proof yet. Twenty clicks would give 0.81 to 1.66 percent, clear of the old range.

Our rule is strict on purpose. The new low end must sit above the old high end. Overlap means keep working or keep waiting.

Why not wait for statistical significance? The same authors point out that significance does not answer the question you care about: how sure you can be now.

Two weeks of data is a coin toss with a damn chart attached.

Where people get stuck on page two

Three snags are common. Each one has a plain cause and a plain fix.

Google shows a different title than the one you wrote. Google uses several sources for the title link, including the big headline on the page. Make the title and that headline say the same thing.

The page moved up, but clicks did not. The result may sit under an AI summary, or its description may not answer the search. Google mostly builds that description from your page, so open the page with a two-sentence answer to the query.

Nothing changed after two weeks. Search is slow. The Hacking Growth authors note that organic channels take longer to pay off than paid ones.

Check the page is indexed with the URL Inspection tool. Then leave it alone until day 90.

Editing a page every week restarts your own test. Leave it alone. Patience is part of the method.

Run the Page-2 Playbook with AI this week

The prompt below runs the whole playbook in your preferred AI. Paste in your Search Console export, and it builds your shortlist, new titles, link plan and 90-day scorecard.

It never invents numbers. It works only from the export you give it. When something is missing, it asks.

This week: export 90 days of page and query data, then run the prompt on it.

Ship new titles and two links for your top three pages. Write your win line, set a reminder for day 90, and move on.

Next problem: getting quoted inside AI answers. Read what AI answer engines actually cite, or ask us about our SEO work.

Our eleven articles were shown 1,745 times and read almost never. Being shown is not being seen. Stop painting new signs, and move the ones you have to where people walk.