After the AI overview appeared, there were only thirty-six clicks out of a thousand exposures, does SEO still have meaning?
A thousand exposures leave only thirty-six clicks, these numbers are true, but the conclusion is not 'SEO is meaningless', but 'the goal of SEO should change'.A measurement in late November 2026 shows: When the AI overview appeared, the natural click rate dropped by about 58.6%, the clicks per thousand impressions dropped from 87 to 36;The page that was most affected was the one originally in the first position, with a drop in click-through rate of 56.9%, from 27.4% of clean result pages to 11.8%.Clicks are fewer, but this link is still supported by the original infrastructure that discovered, extracted, and cited the page.
How are these click-through losses measured?
Need to see three layers of aperture at the same time. The first layer is the sample: This is a measurement on a specific query set, not a general conclusion for all sites, and there are great differences between sites with different traffic structures. The second layer is the larger page referenced on the same page: Zero clicks are already the majority in the overall search, and Similarweb's quarterly report gives a range of more than 65% for desktop and more than 71% for mobile, while AI overview appears in about 78% of information queries. The third layer is distribution: losses are concentrated in information queries, while transactional and navigational queries will still click.
So the 'One thousand exposures only leave thirty-six' describes the average of the type of page that appears when the overview is shown, not the fate of the site as a whole. Use it as a premise to diagnose your own site, and the first step should be to look at how much of the traffic structure is information queries.
Clicks are down, what metrics should SEO change?
| Old metrics | The limitations in zero-click environments | Recommended replacement specifications |
|---|---|---|
| Click-through rate | The overview directly answers, clicks are intercepted | Times extracted, number of sources cited |
| Ranking positions | The first position has dropped more than half of the clicks | Coverage of languages and page types |
| Session duration | No duration without visits | Whether the answer page is self-contained and can stand alone |
| Exposure volume | Exposure is not equal to visibility | Whether the search party can find and correctly categorize the page |
The key action of the replacement is to measure 'being read' and 'being chosen' separately. Being read depends on collection and push, and being chosen depends on content structure. The solutions to the two types of problems are completely different, and mixing them together will lead to misjudgment.
Site map and link push ensure that the page is discovered
The easiest mistake to make in a zero-click environment is to think that since there is no click, there is no need to worry about inclusion. In fact, the search party must first be able to access the page before they can extract information. Automatically generated sitemaps ensure that each address has a stable discovery path; for content with a time limit, link submission and active submission can directly deliver the new address to the search engine, without waiting for the next crawling cycle.
There is an additional layer of verification for the inclusion of information pages: a large number of similar pages under the same theme will be merged or ignored, and the appearance in the sitemap does not equal the crawling of each one. Verification should be done by sampling visits rather than just looking at the submission results. Baidu push and Bing push are also different in terms of use, the former is biased towards speeding up the inclusion of Chinese search, and the latter is used to cover the crawling of another search engine, both of which need to be verified according to their respective interfaces for the amount of submission and failure.
Anchor text and keyword library help the search party locate the theme
The premise of extraction is to judge which question this content is answering. Anchor text management makes the keywords within the site point consistently, so that the search party can judge the theme of the page from the internal links; Keyword library management concentrates these words in one place, avoiding the use of different names for the same concept in different columns.
These two functions are underestimated in the zero-click environment: they do not bring clicks, but decide which theme the page is categorized into and whether it is selected among similar pages. In comparison, the old practice of improving ranking by stacking keywords is almost ineffective here, because the overview refers to statements that can stand independently, not word frequency.
Common questions
If not clicked, should SEO still be done?
It should be done, but the goal changes from getting clicks to being discovered and extracted. The infrastructure of inclusion, promotion, and anchor text determines whether the search party can read the page, and this step has no alternative.
Which type of content is affected the most?
Pure definition and pure explanation information content. Transactional and navigational queries will still bring clicks, while comparative queries are between the two.
How to judge the loss degree of your own site?
Look at the query type separately rather than the total clicks. Use existing data to check the proportion of information queries, and then compare the overview appearance ratio to give a rough but usable loss range.
How do you handle a brand being mentioned without a link
Treat it as a brand exposure and count it separately. The mention rate and citation rate should be considered as two separate indicators, as only looking at references with links will underestimate the actual visibility.