So Outside Technology Security CMS
AnQiCMS
Online Demonstration

More than half of the references in AI answers come from natural ranking, is it still worth making a separate plan for keyword positions?

📅 2026-11-01 👁️ 0

AI responses引用 over half come from natural rankings, the keyword position plan is still worth doing separately. BrightEdge has been tracking the references of AI Overview for 16 months, and the overlap ratio between the reference pages and natural rankings has increased from 32.3% to 54.5%, with a relative increase of 69%. These numbers do not mean that the position plan can stop, but that it has changed roles: the position continues to bear the entrance construction, and the quality of the quoted paragraphs is responsible by the GEO side.

Overlap in references means that positions are still the foundation.

More than half of the references are themselves in the top natural rankings, in other words, the candidate pool of AI abstract references is mostly from the header pages you are already familiar with. There are two practical meanings to the increase in overlap.

The first is not to waste any investment. The content, structure, and external link work you do in the keyword position is also a qualification for AI reference candidates; giving up the position to bet on the generative entry alone is like tearing down the foundation to build the floors. The second is more concentrated competition. The head position has always been the object of position planning competition, and the reference adds another layer of exposure value, amplifying the marginal benefits of the same batch of pages, and for middle-level pages to enter the reference pool, they first need to solve the ranking problem.

What part of the budget should the position planning retain?

Based on overlapping data, it is recommended to retain three types of budgets: the position investment in core transactions and query words, which determines whether the page can enter the candidate pool; content update and maintenance, the sources in the reference pool will rotate, and stagnant pages are easy to fall out; basic technical work, including URL structure, site map and push, to ensure that the page is indexed and understood.

What can be compressed is the technical expenditure on position without relation to content quality, such as the production of pages filled with keywords. On the other side of the increasing overlap is the sensitivity of references to paragraph quality, even low-quality pages are not easy to become the one referenced at the top position.

How is the GEO face and SEO face分工

Both sides use the same set of content assets, with different division of labor. The SEO face solves the issue of whether it can rank and whether it can be indexed: static URL maintains stable structure, 301 redirect inherits the weight after the revision, Sitemap automatically generates and actively pushes to Baidu and Bing to solve the timeliness of inclusion, anchor text management arranges the keywords within the site clearly.

Can the GEO section be quoted in full? Each paragraph is self-contained, and the reader can take away the answer without reading the context; the first sentence directly answers the question. As for the automatically generated llms.txt on the site side, its purpose is to explain the site structure clearly, making it convenient for large language models to understand and index the content of the website. It does not change the reference mechanism itself and does not replace the position work. Using it as a traffic entry plan is a mismatch of expectations.

It is recommended to separate the statistics on the measurement: on the search side, look at clicks and rankings, and on the AI side, look at the list of pages cited and the number of mentions. Both sides point to the same batch of content, but reflect different issues.

Common questions

Ask: If the overlap rate is over half, can we just do one side?Answer: If we can only choose one, choose the position, because more than half of the reference candidate pool comes from the top of the natural ranking. But if the paragraph quality is not good enough, even if the position is high, it can only be exposed without any effect. Two sides solve two questions.

Ask: Should llms.txt be produced separately?Answer: The generation cost is very low, the site supports automatic generation, just turn it on. Don't treat it as an independent traffic plan. Its help to visibility lies in the fact that the model reads the site structure more efficiently, not changing the ranking.

Ask: When the budget is tight, which side should be cut first?Answer: Cut off parts that are not related to content quality first, such as pure word stacking production and mechanical external links. Both basic position investment and paragraph rewriting are retained, the former determines the entry, and the latter determines the probability of being quoted.

Related articles

The third-party vulnerability library can list entries by plugin name, should the plugins installed on the site be checked individually?

Public vulnerability library organized by component name and version entries, the premise of site verification is to have a component ledger first. This article provides the minimum information of the ledger, the verification order based on external accessibility priority, and the differences between the two types of disposal actions with official patches and those without official patches, and explains why the system's built-in injection filtering and content audit layer cannot replace this investigation.

2026-11-01

The announcement writes the authentication bypass and marks it as 9.8 points, should we stop this kind of temporary login function first?

Seeing the bypass of authentication with a high-risk score, it is not necessary to immediately shut down all at once, but it is necessary to first judge whether it is in use, whether it is accessible from the outside, and whether there is an alternative path. The attack surface of temporary login functions is in three points: who can apply, how long the validity period is, and whether it is protected by secondary verification. The disposal sequence is to collect the exposed range, concentrate on credential rotation, supplement session invalidation and log verification, and then decide on patches or shutdown, while confirming that the back-end authentication entry has anti-automation protection.

2026-11-01

Open-source website building tools have written order callback and login modification into the new version, can the corporate website use it?

The open-source website building tool incorporates order callback, registration login modification, and AI expansion into the new version, which may not be necessary for corporate websites. The criteria for judgment are data ownership and whether additional services need to be resident: orders and finance belong to business system integration, login and group permissions directly affect the back-end security boundary, and editor and template updates are the parts that the content team feels strongly about in daily life. Do not hang externally if it can be covered by built-in modules, and confirm the interface scope first if external system cooperation is needed.

2026-11-01

The release week of Drupal12 has been written into the official schedule, whether the site schedule should be moved up to reserve a window

The release schedule of open-source projects is predictable, the Drupal official schedule has clearly stated that the next major version is scheduled for the week of December 2026, and also listed the end of security support for some branches. The site should reserve an upgrade window in advance, rather than dealing with it on the day of the announcement: first, divide the scheduling into major versions, minor versions, and security fixes, and then prepare reversible backups, independent pre-release environments, and content freeze periods before the window, and write the version tracking into the annual maintenance schedule.

2026-11-01

After the AI summary covers most commercial queries, should the content selection change the approach?

There is tracking data showing that the coverage of AI Overview in business intent queries has increased from 45.75% in September 2025 to 88% in August 2026. After the display position is largely occupied by the summary, the selection should shift from wordy pages to question-based content: the title should be written as a question the user would ask, each paragraph should be self-contained as a small answer, and be able to be quoted directly without context; mass production can be facilitated by AI writing and automatic generation of title keyword descriptions.

2026-11-01

How to schedule the upgrade window for the first beta version of Joomla 6.3 at the end of November, while still in the 5.4 site upgrade window

The Joomla official roadmap shows that the first Alpha of version 6.3 is scheduled for November 24, 2026, with the final release planned for April 2027; regular bug fixes for version 5.4 have ceased as of October 13, 2026, with security support ending on October 12, 2027. Sites still running on version 5.4 should prioritize the last year of security support window and complete an upgrade drill with backup and rollback plans before entering the tail end.

2026-11-01

PHP two high-risk vulnerabilities scored 9.2 points, what steps should the site take after the patch is released?

On October 21, 2026, PHP released the fixed versions 8.5.12, 8.4.27, 8.3.36, and 8.2.35, covering two high-risk vulnerabilities with a score of 9.2. Applying the patch is just the beginning: after the upgrade, you still need to confirm that the actual running process is using the new version, check the high-risk entry points related to the conditions for exploitation, and verify the in-depth defense of the site layer by layer, including authentication, injection, and cross-site scripting protection.

2026-11-01

A lightweight blog program can store data within a few tables, how does Typecho control its size?

The official introduction of the Typecho website states that it uses 7 tables and less than 400KB of code to implement a complete plugin and template system, and takes the ultra-low CPU and memory usage as a selling point. The lightness comes from the convergence of data models and the selection of functions. Another route is to change the implementation language: AnQiCMS is developed based on GoLang, and the official data shows that the memory usage is reduced by about 80% compared to PHP CMS, and can carry about 5 million PV on a single machine.

2026-11-01