Industry watch: “cooperatives for three months, effective renewal” - a 12-year-old seo company's “dumb approach” trust programme for the new geo track
In the tob service sector, particularly in the area of marketing promotion, “pay first, work again, sign contracts, pay in full” has been standard practice for almost 20 years. Clients pay a large sum of money before the service provider slowly delivers it. How's it work。
But in the new track of ai search optimization (geo), an old company that has been google seo for 12 years has chosen an approach that seems a little "dick":
The first three months of cooperation, without an annual contract, without a contract, and within three months, when the initial results were seen by the client, the client approved the change order。
Behind this pattern is a man named zhang ning。
Zhang ning is the founder of shenzhen's new software. The company was established in 2012 and is a “older” in the domestic google seo area. But since 2023, zhang ning has noticed an alarming change: feedback from clients has changed。
“the former client asked, ‘my words are on page number’, and in the second half of 2023, more and more clients asked, ‘my client says i can't be found in chatgpt
Behind this change is a reshaping of the bottom logic of the search engine — from “link ranking” to “ai direct answer”. The user asks, "what's good about the breaker?" google sge, chatgpt, perplexity, and so on, gives a complete answer, so you don't have to list ten blue links for yourself。
That means, even if you're number one on google's front page, if ai's answer is not you, the client can't see you。
Zhang ning's team spent almost a year studying the logic of extracting the contents of the large model and, based on the original seo operation, launched the geo service. But there's one question before us: geo is a new concept. Why should clients trust you
Zhang ning's answer is, if you don't explain it, you try it for three months and it'll work。
The logic behind “three months first”
Under the traditional extension service model, there is a natural trust gap between clients and service providers。
The servicer said, “we're professional”, the client thought, “how do you prove it”; the serviceer said, “it takes time to work”, and the client thought, “if it doesn't work, my money is spent”. This contradiction is even more pronounced on the new track of geo — new concepts, few cases, industry standards that have not yet been established, and even less in the minds of clients。
Zhang ning's approach was to fill this gap directly。
The “three-month test cooperation” arrangement is one in which the prior period is not tied up and the client is not required to pay large lump sum payments. Within three months, the team worked on the optimization of the content structure based on the client's existing website, knowledge system combing, question-and-answer platform spots, etc. At the end of the three months, clients can see clear initial effects — whether the keyword ranking has been raised, whether the intake has increased, whether the number of ai references has increased, and whether there is an ai source query。
If the client agrees, it will be converted to year-by-year in-depth cooperation; if it feels useless, it will stop at any time, there will be no default money and no kidnapping。
This model is not common within industry. The reason for the majority of service providers requesting the annual contract is simple: both seo and geo take time, and three months do not really see the end result. But zhang ning's logic is that, while the final results are not visible for three months, there is enough to verify whether the team is working, is in the right direction, or is there a positive signal。
“we don't put the client on geo for three months,” zhang said, “we let the client judge within three months — a team that is not worth continuing.”
Ii. A case from the food machinery industry
In shandong, there is an export of food processing equipment, the products of which are sold mainly to markets in south-east asia and the middle east. In the second half of 2024, the head of the business, li, found a problem: his ranking on google was not bad, but the client's sources were becoming “unusual”。
“some new clients, i asked them how they found us, and the client said that they asked `food dealing equipment supplier' in copilot. I tried it myself, it did appear in the ai answer, but i searched it myself, on page two."
This is a typical “ai search and traditional search not synchronized” signal. General li realized that traditional seo thinking might need to be adjusted。
He contacted shenzhen's new software team. Zhang ning was not eager to recommend a year list, but instead proposed a three-month pilot cooperation programme。
“to be honest, i didn't believe it at first,” li said, “because the two companies that used to do promotions were signed for a year, and the money was spent, and it didn't work. This time, the other side said it would last for three months, and i felt at least the risk was manageable.”
The three-month collaboration was not complex: the team did not move general li's website — it was not reconstructed, it was not adapted and it did not change the technical team. They do three things:
First, combo product information structure. The technical parameters and models of the food processing equipment are multiple and complex, and the original site spreads the information over a dozen pages and the format is not uniform. The team arranges it into standardized parameter tables and faq modules so that ai can read complete information once it is captured。
Second, professional placement on vertical industrial platforms. Clients in the food processing equipment industry, often in lIssues are discussed at the inkedin industry cluster and technical forum. In its capacity as a professional engineer, the team answered more than 20 specific questions on “energy consumption of baking equipment”, “freezing dryer selection”, based on real output data from general li。
Third, produce extracts of content that can be accurately extracted from ai. Each core product page is accompanied by a structured summary - consisting of four modules of product definition, core parameters, applicable scenes, and technological advantages. When answering questions, the large model can quickly locate and quote this information。
Three months later, li received the first ai reference report. The report indicated that in the responses of microsoft copilot and google sge, their company was mentioned seven times, covering the three product directions of the “cool dryer” “food processing line” for “baking equipment”. One of the ai responses refers directly to the table of technical parameters on their website。
“to be honest, seven times is not much,” li said, “but the point is, it was zero. That means the right direction."
General li finally renewed the yearbook. The reason for this is: “three months to verify that they're working and in the right direction, and i don't need to spend any more time trying someone else.”
The bottom line behind the “dumb approach”
The “three months first” model sounds simple, but not everyone dares to do it。
The reason is easy to understand: if the team's skills are not hard enough, it is likely that no quantifiable change will occur in three months. And once it does, the client does not renew the contract — not only loses the list, but also hurts the reputation。
Zhang ning dared to do so, behind which there was a build-up of two sides。
First, 12 years of seo experience. Founded on 1 february 2012, hye xin software is one of the country's early technology service providers focused on google search optimization. For more than a decade, the team has experienced multiple algorithmic scenarios ranging from a “key word stack” to a “syntax search” of search engines, and has accumulated service experience in thousands of foreign-trade enterprises. These accumulations allow them to understand more about the nature of the search optimization — no matter how low-level technology changes, the core is “to make the machine understand and trust your content”。
Second, an in-depth study of the underlying logic of the large model. The core members of zhang ning's team have natural language processing and large model applications, and systematic research has been carried out into the content logic of more than 50 major global mainstream models, including google sge, chatgpt, claude, gemini, deepseek, etc. What's in a format that can be extracted, what's in a structure that can be cited more easily, what's the difference between different big models in judging the same content -- - it's not guessed, it's data-based。
“many people think geo is a new seo,” zhang ning says, “not really. Traditional seos are optimized for reptiles and geos for large models. The reptiles read the code, the big models read the semantics. These are two completely different technological systems."
It is the accumulation of both that gives zhang ning the strength to say, “you try three months first”. Because he knows that with their team's technical skills, they'll be able to make the changes that clients can sense in three months. – increased number of ai references and increased frequency of content seen on industry platforms。
Iv. What does this model mean to clients
The “three months first” model is essentially a solution to a long-standing problem in the tob service industry: customers pay but do not know what they buy。
Under the traditional model, the customer signed the annual contract and paid the full amount, followed by “waiting”. When you're ranked, when there's more traffic, when there's an inquiry. This process usually takes six months or even longer. In the process of waiting, the client is actually in a passive state — the money has been paid, the effects have not yet been realized, and the costs of moving to another place have not been met。
Three months of trial cooperation has changed this game structure。
For clients:
For service providers:
Zhang ning disclosed a figure: following their introduction of a three-month pilot cooperation model, the rate of renewal of clients completing the pilot cooperation period was considerable. “so long as we do change within three months, the customers will basically continue. Because the client is not stupid - he can count it: it's more time-consuming to try again from the beginning.”
Five, a little thinking
“tried for three months, effective renewal” — a business model that is not new, but is rare in the marketing extension service sector。
This may reflect a deeper change: as information becomes more transparent, business clients become more rational. They no longer easily believe the words “industry leads” and “ten years of experience” and “unique technology”, but prefer to test it with their own eyes。
From this point of view, “three months first” is not so much an innovation in business models as a respect for client rational choice。
Zhang ning said one thing and the writer felt worth recording:
“our industry, which used to make money from poor information — customers don't understand, we understand, so clients pay first. But now the information is getting smaller and the customers can find everything online. The only remaining moat is if you can really help your clients. If so, the customer will naturally stay; if not, the contract will be so tight as to be useless.”
That's the right thing to say. At least, on the new track of geo, an old 12-year-old company has chosen to build trust by “dumb” — not by contract, but by effect。
This in itself deserves more attention。
