Reading time:

8 min read

|

Last updated:

Why Position 1 Now Gets You Zero Clicks

Pages ranking top five with no clicks are often answering conversational AI queries. How to spot them in Search Console and what to change on the page.

We design websites and products that make AI companies more money.

Siddarth Ponangi

Founder, Studio Maydit

We design websites and products that make AI companies more money.

Web and product design for AI companies

We help AI companies build fast, clean, and conversion-focused websites and products.

If Search Console shows pages ranking in the top five with almost no clicks, look at the queries before you change anything. A growing share of them are full sentences, written the way people talk to an assistant rather than the way people type into a search box. Someone asks a question, an AI surface answers it, your page is one of the sources it used, and the impression is recorded without anyone visiting you. A better meta description will not fix that, because the snippet is not what the reader saw. What helps is writing pages that answer the question directly, in sentences that can be quoted on their own, and measuring something other than clicks.

Here is what those queries look like on our own site, how to find yours, and what actually changes about the way you write.

What these queries look like

These are real queries from our own Search Console over the last 28 days, with the position we held and the clicks we received.

"our website no longer reflects who we are. which digital agency should we hire for rebranding?" sat at position 1.0 with 36 impressions and no clicks. "which design retainer fits multi-product saas rebranding needs?" sat at position 2.5 with 40 impressions and no clicks. "we need continuous design support post-series c funding" sat at position 4.1 with 40 impressions and no clicks. "what is the best framer agency for startups?" sat at position 6.2 with 78 impressions and no clicks.

You can see the pattern. They are long. They are written in the first person. Several are two sentences, with the situation first and the question second. Nobody types like this into a search box. People talk like this to an assistant.

One of our pages collected more than 400 impressions in a month at an average position of 6.3, and almost every query on it had this shape. It received zero clicks. By the old reading, that page was one small improvement away from real traffic. By the actual data, there was no traffic to win.

How to find yours

Open Search Console, go to Performance, and switch to the Queries view for the last 28 days. Then look for three things together.

First, sort by impressions and read the queries rather than scanning the numbers. Second, look for anything longer than about eight words, especially if it starts with a word like our, we, which, who, or what. Third, check the position and the clicks side by side. A query at position 1 to 6 with a healthy impression count and zero clicks is the signal.

Do the same at page level. Filter to one page, look at its queries, and ask whether a person would type any of them. If most of them read like something said out loud, you have found a page whose numbers you have been misreading.

One thing to filter out first. Some long queries are not people at all. If you see many queries that begin with a phrase like "i am a 35-44 year old founder" repeated with different ages or job titles, those are automated visibility checks run by tracking tools, not buyers. Ignore them when you judge demand.

Why a top position produces no click

The reason is simple. The question was answered where it was asked.

If someone asks what the average conversion rate is for a software website, the answer is a number. Once the number appears, the visit has no purpose. If someone asks which agency they should hire for a rebrand, the answer is a short list of names with a sentence each. The person now has what they wanted, and if they act on it they will probably search the name of one company directly rather than click a list.

This is not a penalty and it is not something you did wrong. It is what happens when the surface that receives the question is also able to answer it.

What it changes about how you write

If the page is being read by something that summarises it, then the page has to be easy to summarise correctly. In practice that means four habits.

Answer the question in the first paragraph, in full, without needing the rest of the page. If your opening sets up context and delivers the answer in paragraph six, the summary will be built from someone else's paragraph one.

Make the facts specific enough to be worth quoting. A number, a date, a named example, a range with its conditions attached. Vague sentences survive nothing. "Most teams see improvement" is unquotable. "A ten to fifteen page rebuild usually takes six to eight weeks" can be lifted and attributed.

Use headings that are the questions people actually ask, and answer each one underneath in a paragraph that stands alone. This is the same discipline as a good FAQ, applied to the whole page.

Keep every claim next to its conditions in the same sentence. A claim that gets separated from its limits is a claim that will be repeated without them, which is how a reasonable statement turns into something you would not have said.

What this does not mean

It does not mean search traffic is finished, and it does not mean you should stop writing pages.

Ordinary typed queries still exist and they still convert. On our own site the pages that earn the best click rates are the ones matching things people genuinely type, like a specific service plus a specific kind of company. Those behave exactly as they always did.

It also does not mean an impression without a click is worthless. Being one of the sources an assistant uses puts you in the consideration set at the moment somebody is choosing. That is worth more than a browsing visit, even though it will never appear in your session count. The cost is only that you cannot see it happening.

What to measure instead

Start by separating the two kinds of demand. Once a month, take your top queries and split them into typed and conversational, then track the two impression totals separately. Judging one page against a blended click rate hides both stories.

Then watch the signals that show up after a citation rather than during one. Searches for your company name are the clearest. Direct visits are next. Then the thing that actually matters, which is booked calls and demo requests, including the ones where someone says they were recommended a shortlist and you were on it. Ask that question on your enquiry form, because it is often the only place this shows up at all.

Finally, keep a list of the questions your pages rank for, and check each quarter whether the page answers that question in a single paragraph near the top. If it does not, that is the edit worth making. It is also, not coincidentally, the edit that makes the page better for a human who does click. If you would rather have someone look at your pages with you, book a 30-minute call and bring your Search Console.

If you are trying to work out what your pages are worth once the clicks are this scarce, our guides on software website conversion rates and testing a landing page with low traffic cover how to read small numbers without fooling yourself.

Frequently Asked Questions

Table of Contents
Scroll to view headings
0%