The visits brought by AI only account for one percent of the website traffic, why is the conversion rate higher?
AI recommended visits only account for slightly more than one percent of the site traffic, why the conversion rate is higher on the contrary, and whether it is worth investing? Whether it is worth it depends on whether you first understand that these two numbers are not the same denominator. The proportion is calculated as the average of all sessions on the entire site, and the conversion rate is calculated as the achievement rate separately for each recommended channel. The former is smaller, and the latter is larger, which is not contradictory, nor does it constitute a direct answer to whether to invest or not.
What is the statistics of the item with a small proportion
A study covering more than 13,000 domain names shows an average proportion of only 1.08%, note that this is the share calculated by putting different types of channels in the same pool, including a large number of industry sites that almost do not rely on content-driven traffic. That is to say, this average value describes 'the current overall scale is still very small', not 'this type of access is not valuable'.
The fact that the scale is small is a fact, and the growth rate and structure are also facts: most of the sites that truly treat this channel as a decision-making basis first look at whether they have been cited, rather than first look at the industry average proportion.
Why is the conversion rate so different
In the same set of statistics, the recommendation conversion rate from the conversational engine is significantly higher than the traditional natural results, with a magnitude difference of ten times or so. The reason is not that the channel itself is more 'advanced', but that the user has already completed the selection at the moment of recommendation: the answer names your product or page, indicating that the user's question and your content have been determined to be related, and the decision-making progress at the time of reaching the site is more advanced.
| Indicator | What is the denominator? | What questions can be answered? | Easy to misread as |
|---|---|---|---|
| Recommendation share | All sessions on the site | How big is this channel now? | The channel has no value |
| Channel conversion rate | Visits brought by this channel | How close are the people coming in to the decision-making process | Content quality is better than that of peers |
| Number of citations | Number of times cited by Q&A system | Whether the content enters the shortlist | Equals bringing clicks |
| Completion rate of landing page | Behavior after citation falls on a specific page | Whether the reception matches | Irrelevant to citation rate |
Cannot be divided directly if the denominators are different. Placing 1.08% and high conversion rate in a sentence to compare will lead to the wrong conclusion of 'small and beautiful, so don't do it.'
First, strive to be cited or to create the landing page first
The sequence is to solve the 'read' first, then the 'cited', and finally the 'landing'.
- Being read: the site structure should be understandable to the model, AnQiCMS will automatically generate the site LLMs.txt, organizing the sections and key content into an entry for the reader;
- Being cited: the paragraph should be able to stand alone as a conclusion, data and sources should be written clearly, and question-and-answer titles are easier to摘;“
- To be able to land: the referenced section should be consistent with the landing page, the reference talks about configuration steps, but the landing page stops at the product introduction, the conversion will drop immediately.
The AI Chat intelligent dialogue in this chain can only perform checks and bulk revisions, such as rewriting scattered conclusion paragraphs into forms that can be independently cited, but it cannot replace the design of the page itself.
Judging the three steps of investment
- Take a group of real questions related to your business, ask several engines, and record whether your brand appears and in what form it appears;
- Check the links and landing pages of the parts that have already appeared, to see if the cited text and the content on the first screen of the page are about the same thing;
- Fill in the content for the parts that have not appeared, retest monthly after one round is completed, and do not determine KPIs based on single-round results.
Common questions
Is the proportion so small that we need to set up a KPI specifically?You can initially set up controllable indicators such as 'number of citations' and 'completion rate of landing pages cited', treating the proportion as an observation item rather than an assessment item.
What should you do if cited but not clicked?This means that the user has stopped at the answer, and at this point, you need to check whether the content provides enough motivation to go to the page, such as tables, lists, or tool-type information.
Should content be created separately for different engines?There is no need to write articles according to the engine, what is needed is to allow the same batch of conclusion paragraphs to be extracted by different engines.