So Outside Technology Security CMS
AnQiCMS
Online Demonstration

The average click-through rate of questions with AI overview has dropped by about 20%, what other indicators can be attributed to

📅 2026-11-01 👁️ 0

The average click-through rate of questions with AI overview has dropped by about 20%, what other indicators can be attributed to it? First, let's determine the nature of this number: it comes from a study covering ten million-level keywords, comparing the mean difference before and after the overview of the same batch of questions relative to the baseline. The mean means that the results are calculated by combining different intent and different industry questions together, and using it to predict the decline of a single site will be significantly biased - a site with a high proportion of brand words is facing a completely different traffic structure compared to a site with a high proportion of long-tail information words.

What does this 20% number measure?

The sample is counted by question unit and not divided by site; the baseline is the performance before the overview appears. Therefore, it describes that 'questions with an overview are more likely not to be skipped out', rather than 'your site has dropped by 20%'. Taking it as a ratio that can be applied to your own backend is the most common way this type of research is misinterpreted.

Why can't it be taken directly as the decline of a single site

Three reasons: Firstly, the intention structure is different; the click behavior of navigation and transactional questions is not on the same distribution as information questions; secondly, the position structure is different; the loss of pages originally ranked first and those pushed to the second screen after is very different; thirdly, the overview trigger is unstable; the results of the same question at different times and on different devices are not consistent, and the fluctuation within a short period of time is easily interpreted as a trend.

Three indicators that still hold steady.

A category is the amount of in-coming traffic brought by brand words and questions containing brand names, which is less affected by the overview and can reflect whether the actual demand changes; another category is the manifestation of non-informational intent, such as the behavior brought by pages such as inquiry, comparison, and download; another category is the conversion process after the entrance - form submission,留言, consulting and ordering, which occur outside the overview and are closer to business results than click-through rate. Split the keyword library into these groups according to intent and observe the trend, which is closer to the explainable attribution than focusing on the total amount.

Make the attribution reproducible.

Leave the control period first: Do not change a large number of templates and content at the same time, otherwise it will be impossible to distinguish between the impact of the overview and the impact of the revision. Next, look at the push and inclusion: If only part of the questions trigger the overview, use link push to actively hand over the new content and revision address to the search engine, so that the variable of 'whether included' can be fixed first. Finally, check the structure: Whether the anchor text and internal links hide the conversion page too deeply, when the deep link is skipped, the decline in click-through rate will appear simultaneously on the result side and the site side, and it is easy to be mixed into one thing.

Common questions

Can questions without an overview be used as a control group?Can it, and better. Comparing the performance of the same batch of keywords by triggering a preview group is more reliable than directly citing the external average.

If the click-through rate decreases, does that mean the content has worsened?Not necessarily. After the preview puts the answer in the foreground, some needs are met on the result page, and the clicks naturally decrease, which is a separate line from content quality.

How often should one look at the conclusion?It is recommended to look at trends on a monthly basis and structural changes on a quarterly basis. Short-term fluctuations after a single algorithm or layout adjustment should not be conclusive.

Where should one start?Prioritize the observability on the conversion side, and then discuss content rewriting; when one does not know what the user does after entering the station, any click-through rate conclusion cannot be verified.

Related articles

The program itself is free, but the plugins are charged by seat. How do you calculate this cost when selecting?

The cost-free is the software authorization, not the entire cost. The selection should be broken down into four segments for pricing: software license, hosting resources, plugins charged per person or per site, and upgrade backup and other maintenance labor. The larger the team, the more sites, the more obvious the difference in seat fees and hosting fees, and the total account should be calculated based on the same number of people and the number of sites when comparing.

2026-11-01

The screening conditions generate a large number of similar pages, whether to block crawling first or block indexing first.

The screening page should first determine whether the waste is the crawling quota or the inclusion quota: the address combination is still being repeatedly requested, which belongs to the crawling side issue, and can be收敛 using robots rules; the page has been crawled and entered into the inclusion before talking about indexing, the handling of whether to index should be canceled indexing or merge the standard address into the main page, blocking crawling itself will not cancel the existing inclusion.

2026-11-01

How to control the crawling depth of list pages when pagination and infinite scrolling coexist

The depth of crawling on the list page is determined by the internal link structure: Infinite scrolling must retain direct access to pagination addresses, each page returns complete HTML and real links, and page turn marks are no longer considered as inclusion signals. The control means three things - fixed pagination parameters, hanging the entry text into a crawlable link, and supplementing the deep content with site maps and anchor text, rather than expanding the page numbers infinitely.

2026-11-01

Only two of the top ten sources cited by the three AI engines overlap, do we need to prepare the content separately according to the engine?

There are only two sources of median overlap among the top ten references of three engines, which means that the domains selected by each search pool are different, and it is not the same problem that has been split into three sets of requirements. Copying pages by engine will make the internal signals thin, and it is more cost-effective to write the same content into paragraphs that can be independently cited, accompanied by verifiable data and a clear site structure, and then monitor the citation results on a periodic basis.

2026-11-01

Does a small site need to worry about crawling budget, what two numbers do the backend look at first?

The crawl budget usually only becomes a bottleneck on large websites or websites with high frequency updates, small stations should first clearly see the total number of daily crawl requests and the average response time in the search console, then deal with these three types of waste: broken links, redirect chains, and bulky site maps, and the real problem of most small stations is not the quota, but the inclusion and content.

2026-11-01

After the redesign changed the domain name, do we need to keep the redirection of the old address active?

The redirect after changing the domain name is not an obligation forever, but addresses that have received external backlinks or have stable traffic should be retained for a long time, usually at least one year before evaluation; after removing, the old address will directly return an error page, and the backlinks and weight will be cut off accordingly. The cleaning sequence is to do address mapping first, then cut the traffic, and finally batch replace the internal links within the site to avoid diluting the signal of multi-level redirect chains.

2026-11-01

Where is the difference in construction costs between static pages and dynamic sites, and who has higher post-maintenance costs?

The price range of static sites during the construction period is lower, while dynamic sites are expensive in terms of backend, database, and security patches; during the operation period, the situation is reversed, where the content update of static sites depends on technical intervention, and dynamic sites require continuous upgrades and backup work. Sites with frequent updates are more suitable for content systems with backend and scheduled publishing, separating the update actions from the development time.

2026-11-01

Should the new article be submitted to the search engine actively, and which companies does this protocol cover?

The proactive submission resolves the section of

2026-11-01