Debunking Five Myths of GEO Optimization

Starting yesterday, I have received a series of questions from students and friends, who have been unsettled by various claims in the market and have developed a string of "illusions":

How is the search volume displayed by GEO monitoring tools calculated? It feels fake?
Is monitoring the mention rate of GEO optimization completely meaningless?
Can user comments really affect GEO rankings?
Is spending hundreds of thousands of yuan on outsourced GEO services completely useless?
Do traditional SEO practitioners not understand GEO?

Today, I want to objectively analyze these illusory issues one by one from a relatively impartial perspective and restore the true face of GEO optimization.

I. Are the "search volumes" in some GEO monitoring tools real?

To accurately understand this issue, we first need to distinguish between two types of tools: official tools and third-party tools.

In traditional SEO, Baidu officially provides the Baidu Keyword Planner, which shows the monthly average search volume for each keyword. This is official statistics based on real user behavior data and is relatively accurate. If you want to view the real click volume and impression volume of your own website, you can also check them in the Baidu Search Resource Platform or Baidu Analytics — provided you are the website administrator with verification permissions.

But third-party tools are different. The "estimated traffic", "keyword search volume" and other data they display are essentially modeled estimates, not real data.
I have seen websites with an estimated 2000+ daily IPs that actually only get 30-50 daily IPs; I have also seen weight-1 websites with an estimated traffic below 100 that have real traffic exceeding 5000+. The huge gap between estimated and real data leads some people to claim "fraud" — in fact, it is not fraud, but the estimation model itself has errors.

In the GEO era, the situation is more complex: all major domestic AI model official platforms (ERNIE Bot, Douban AI, DeepSeek, etc.) have not yet opened APIs for keyword search volume queries, nor will they open them to third parties. You cannot say the data displayed by third-party tools is deceptive, because they never claim it is pulled from official backends — the only issue is that the estimation errors are relatively large.

So is the GEO data from third-party tools still useful?
Yes, but it needs to be correctly understood.
They are market popularity indices estimated through modeling based on multi-source data (social media popularity, news mentions, search trends, etc.), not official precise data. They can be used as a weather vane: which topics may be heating up in the AI circle? What issues are users generally concerned about? This is an important reference for content strategy.

In my understanding: the people who use large AI models for search are essentially the same group of internet users. What they care about in search engines is mostly what they care about in large models. Although this data is not precise, it is a "compass" that can give you a reference direction.

II. Is monitoring the mention rate of GEO optimization really meaningful?

I wrote about this issue before. The concept of mention rate itself is good, but at the current stage, its significance lies not in monitoring the absolute value of a specific entry, but in trends and relativity.

The main reasons are as follows:

1. Huge regional differences
During the Spring Festival, everyone returns to their hometowns. If you ask questions using large models, you will find that the search results are quite different from when you work in other cities. For example, if you ask "recommended decoration companies" in Tencent Yuanbao, it will prioritize recommending companies in your city. So monitoring the mention rate in Beijing may yield completely different results from monitoring it in Shanghai.
API monitoring often ignores regional randomness, while the geographical distribution of real users is random. This leads to deviations — possibly significant ones — between API-monitored mention rates and real-world data.

2. Unknown keyword search volumes
In traditional SEO, many long-tail keywords (e.g., "Which XX company is good?") may have monthly search volumes of only a few dozen times or even less than 5. The same applies to large models. For such low-traffic keywords, even a 100% mention rate has little commercial value. In GEO, if we only monitor a few dozen fixed entries, the sample size is too small, and the resulting mention rate may just be a "lucky number". We can appropriately expand the sample size or observe long-term fluctuations in mention rates to reduce the likelihood of such "lucky numbers".

3. View mention rate monitoring from a different perspective
Every question can be expressed in countless synonymous or near-synonymous ways. For example:

• Website construction company

• Website development company

• Recommended website construction company

• Which website development team offers better services

Users' needs are similar, but their ways of asking questions are thousands.
If the synonyms and near-synonyms of the dozens of core entries we monitor have a certain probability of appearing in users' questions, and the mention rate of the monitored core entries is relatively high, it can indirectly indicate that the overall effect is likely positive.

In addition, we need to note the difference between API monitoring and real user conversations. API calls usually yield relatively stable results, while real user conversations are random and context-dependent, leading to potentially different answers to the same question in different sessions.

Mention rate monitoring should be used as a probabilistic reference, not an exact count.

III. Can user comments really affect GEO rankings?

This is a very interesting question.

A student told me: Some technical bloggers write articles claiming that comments on Zhihu, Baidu Tieba, Xiaohongshu, and WeChat Official Accounts can significantly affect GEO results.

I want to say: These people have definitely not practiced it themselves.

At least in domestic large models (ERNIE Bot, Douban AI, Kimi, etc.), comments currently have a negligible impact on GEO optimization results and can be directly ignored. Those who claim comments have a huge impact must not have run their own GEO projects — they only read foreign theories, which differ greatly from the domestic context.

Of course, there is one type of "comment" that is useful: embedding testimonials and reviews from clients after cooperation in article content. This structured evaluation is easy for large models to crawl and understand. However, user comments scattered in comment sections that require a second click to view are currently difficult for large models to incorporate into the ranking criteria for real-time answers.

Instead of spending energy on fake comments, it is better to focus on producing high-quality content — especially content with real cases and user testimonials. These are the hard currency of GEO optimization.

IV. Is spending hundreds of thousands of yuan on outsourced GEO services completely useless?

With fierce competition now, many jokes have emerged:

• Monthly fee of 30,000 yuan, zero cost

• Spent hundreds of thousands in six months, zero results

• Spent 100,000 yuan, only got 2 inquiries

In fact, GEO is a promotion method with a generally high ROI. In some industries, even a monthly cost of only a few thousand yuan can generate hundreds of user inquiries. It is impossible to spend 100,000 yuan and only get 2 inquiries — this is just an exaggeration, essentially a sales pitch to encourage users to cooperate with them. Nor is it possible to have a monthly fee of 30,000 yuan with zero cost — what doesn't require cost?

The market is stratified, and price differences should stem from service depth, not moral differences. Truly effective outsourced GEO services are based on in-depth understanding of content strategy, data monitoring, and model logic — not simply "brushing mention rates".

V. Do traditional SEO practitioners really not understand GEO?

Excellent SEO practitioners are inherently experts in the search marketing ecosystem. They understand not only official website SEO but also social media SEO and video SEO. GEO, in essence, is the optimization of large model search results, and its underlying logic is similar to search engine optimization:

• Keyword research: SEO requires mining user search terms, while GEO requires mining user question patterns

• Content quality: SEO emphasizes matching content with search intent, while GEO also emphasizes content that answers user questions

• Authority building: SEO requires external links and trust, while GEO requires brand mentions and positive evaluations across the entire network

• Technical optimization: SEO requires website structure optimization, while GEO requires data structuring

Of course, some SEO practitioners may indeed lose to operations personnel in GEO projects. But this is not because SEO practitioners "don't understand" GEO, but because their time is spread thin. Given the same time and energy, people who understand SEO will find doing GEO natural and effortless. So the jokes that "SEO practitioners don't understand GEO" should just be taken lightly, not seriously.

Currently, many of my GEO peers are from SEO companies that have transformed. People who understand SEO find GEO naturally easy to handle and grasp quickly. I also wrote a separate article before: A good SEO practitioner is a good GEO practitioner. However, a good GEO practitioner does not necessarily need to be a good SEO practitioner.

Conclusion

As an emerging field, GEO optimization is inevitably accompanied by various misunderstandings and exaggerated marketing rhetoric. Faced with these "illusions", we need to return to common sense: technology has boundaries, data has errors, but business logic and content value will never be outdated.

Instead of being led by the numbers from various monitoring tools, it is better to focus on two things:

1. Produce high-quality content that truly solves users' problems;

2. Under compliance premises, enable the brand to gain continuous, positive exposure across the entire network.

If you are also paying attention to GEO, SEO, or have questions about search marketing, press release distribution, etc., welcome to communicate more.