A page that holds a top ten spot and gets almost no clicks used to mean something was wrong with it. Bad title, weak description, mismatched intent. You rewrote the snippet, waited a few weeks, and the clicks came.
I ran exactly that repair on one of our own pages in May. New title, sharper and more specific, written against the search intent rather than the topic. Then I left it alone for two months and measured. The click-through rate stayed where it was, around a tenth of a percent, while the page held its position and kept collecting impressions.
That result was more useful than a success would have been. It ruled out the explanation everyone reaches for first, and it forced me to look at what had actually changed.
The Position Is Fine. The Result Page Is Not.#
For a growing share of informational searches, Google answers the question above the results. The user reads the summary and never scrolls. Your listing is still there, still ranked, still counted as an impression. It simply sits below the point where the search ends.
I should be precise about what I can and cannot prove here. Search Console gives you position and clicks; it does not tell you which result page carried an answer box. What I have is a strong correlation on our own data plus the failed title test, and the explanation that fits both. If you want certainty for your own pages, search the affected queries yourself and look at what sits above the first organic result. That takes ten minutes and beats any inference.
For the page I tested I did run that check, and what sits above the organic results explains the flat click line better than anything on the page does. So for this one case I am comfortable saying it was neither a penalty nor a content problem. For your pages that remains a hypothesis until you look.
What makes it hard to see is that every dashboard still reports the old metrics as if they meant the old things. Position two looks like a win. Impressions climbing month over month looks like growth. Only the click column tells you the truth, and it is the one column people explain away.
How to Recognise It in Your Own Data#
There is a signature, and you can check it in Search Console in about ten minutes without any tools.
Look for pages with a high impression count, a stable position in the top ten, and a click-through rate near zero. Not low, near zero. A weak title still produces a real rate, just a disappointing one. What I am describing produces something closer to nothing at all, across hundreds of impressions, month after month, without moving.
Then look at what those queries actually are. On our own pages the affected ones were questions. Definitional searches, "what is" and "how does", the exact shape of query that an answer box handles completely. The commercial queries on the same domain still converted impressions into clicks at an ordinary rate, because a summary does not finish that job.
That split is the useful part. It tells you which of your pages are affected and which are not, and it stops you from applying the wrong fix to the whole site. I would not assume our split generalises to your site, but the check takes ten minutes and tells you your own.
Why the Failed Repair Was Worth More Than a Fix#
I want to stay on the title rewrite for a moment, because the way it failed is the most useful thing in this article.
The standard advice would have had me rewrite, see no improvement, and conclude the new title was not good enough either. Then rewrite again. That loop can absorb months, and it never terminates, because you can always imagine a better headline.
What stopped it was writing down beforehand what a success would look like. A meaningful lift within eight weeks on a page whose position was already stable. When the eight weeks produced nothing, the hypothesis was dead and I had to look elsewhere. Without that number written down in advance, I would have squinted at a flat line and found something encouraging in it.
This applies well beyond titles. Most SEO work is untestable in practice because nobody says in advance what would count as the measure failing. If your agency proposes a change, ask what result would prove the change did not work, and by when. A proposal that cannot fail cannot be evaluated, and you will pay for it either way.
Three Things That Actually Move#
Look at other search engines, because they did not all change at once. This was the finding that surprised me most. I took a page whose Google click-through rate had collapsed and pulled the same query set from Bing Webmaster Tools. Same page, same search terms, roughly four percent click-through rate against a Google figure that rounds to zero.
Bing is smaller, and for most businesses it will not replace Google traffic. But it is not nothing, and the effort required is close to zero because your pages already rank there. In the client accounts I have looked at, a claimed Bing Webmaster property is the exception rather than the rule, which means the channel is usually not even being measured. Whether that traffic converts as well as Google's is something I have not measured and would not claim.
Push the position instead of polishing the snippet. If the answer box is taking the click, cosmetic changes below it are wasted. The lever that remains is moving genuinely higher, and the cheapest candidate is internal linking, because most sites have valuable pages carrying a single incoming link from their own navigation and nothing else. I would call this the best available bet rather than a proven fix: it is cheap, it is under your control, and unlike a title rewrite it addresses the thing that actually changed.
Write to be the source, not to win the click. If the summary is going to be read either way, the question becomes whether it is built from your material. That means content an answer engine can lift cleanly: a direct answer near the top, specific numbers, clear structure, and something genuinely yours in it, a measurement or a case that exists nowhere else. The reasoning is simple enough that it does not need a study behind it. A summariser has to choose what to quote, and material that exists in fifty other places gives it no reason to name you.
What to Stop Measuring the Old Way#
Rankings as a standalone success metric no longer survive contact with reality. A report showing a climb from position eight to position three is not a result if the click column did not move, and any agency handing you that report without the click column next to it is showing you the wrong number.
The same applies to impressions. Impressions rise when Google shows you more often, including on queries that get answered above you. Rising impressions with flat clicks is not momentum, it is the pattern described in this article.
What I would put in a report instead: clicks by intent category, so commercial and informational pages are never averaged together. Click-through rate compared against your own history rather than an industry benchmark. And traffic from every search source you have, not just the biggest one.
Answer Searches and Buying Searches Need Different Pages#
The practical consequence of all this is that one content strategy no longer covers both halves of your site.
Answer searches, the "what is" and "how does" queries, are the ones being absorbed. Pages built for them still have a job, but the job changed. They are now there to establish that you know the subject, to be quoted, and to catch the small share of readers who want more than a summary. Measuring them by sessions will make them look like failures even when they are working.
Buying searches held up in our data and I expect them to hold up longer, for a structural reason rather than a measured one. Somebody looking for a supplier in their region, comparing options, or ready to make contact still needs to arrive somewhere and speak to someone, and a summary cannot complete that. These pages deserve the disproportionate share of your effort now, and for most small companies they are the underbuilt half, because informational content is easier to produce and feels more productive.
If you only change one thing after reading this, separate those two page types in whatever report you look at. The moment they stop being averaged together, most of the confusion about whether SEO still works resolves itself.
The Uncomfortable Part#
If most of your traffic came from informational content, you should plan on the assumption that a part of it is not coming back. I cannot prove that as a permanent state, and anyone who tells you they can is guessing with more confidence than I have. But planning around it costs you nothing if you are wrong, and the alternative is spending another year improving titles on pages whose problem sits above them on the result page.
The useful response is to move the effort to where clicks still exist. Commercial and local searches, where a summary cannot finish the transaction. Channels you already rank in but never measured. And content strong enough to be quoted, which builds recognition even when it does not build sessions.
None of that is as satisfying as a ranking chart pointing up. It has the advantage of being true.
One Test You Can Run This Week#
Open Search Console, filter to the last three months, and sort your pages by impressions. Take the ten with the most impressions and look only at the click column. Several near zero while holding a top-ten position gives you your candidate list, not your diagnosis. Take three of those queries, search them, and look at what occupies the space above the first organic result. That is the step that turns a suspicion into an answer, and it is the one everybody skips.
Then open Bing Webmaster Tools, add the same site, and give it long enough to collect a meaningful sample. Compare the click-through rate on the same queries. If the gap looks anything like the one I measured, you have a channel that is already sending you traffic and that nobody on your side is measuring or working on. Claiming it costs you an afternoon.
