This article shares the author’s practical experience with Meituan, Baidu, and Tencent, and discusses how to build product moats at different business stages. The article unfolds from the essence of user value, data product design, interactive transformation in the AIGC era, etc., and provides practical ideas and cases for product users.
Recently, I read an article by Zhang Xiaodong about how to make good products, and I was deeply touched by the “ecological thinking” and “user value anchor” mentioned in it. As a practitioner in the fields of risk control, data products, content strategy, and AIGC in Meituan, Baidu, and Tencent, I would like to talk about how to build product moats at different business stages based on my practical experience.
1. The essence of user value is “net value”, not superficial function
When he joined Meituan in 2017, it was in the period of explosive food delivery business, with an average daily order volume of more than 10 million. But behind the boom is the crazy attack of the black industry: risks such as false transactions, malicious swiping, and account theft are pouring in.
The risk control product team I am responsible for is faced with a life-and-death proposition – how to accurately intercept risks without affecting the user experience.
I tasted the cruel balance of “value and cost” for the first time. In a large-scale marketing campaign, in order to accurately intercept swiping, we designed an exceptionally strict chain of rules: users need to complete five steps of real-name authentication, binding bank cards, and verification of the address of the first order.
The logic was perfect, and the brushing rate plummeted after the launch, and the team cheered.
How can product managers do a good job in B-end digitalization?
All walks of life have taken advantage of the ride-hail of digital transformation and achieved the rapid development of the industry. Since B-end products are products that provide services for enterprises, how should enterprises ride the digital ride?
View details >
But the real data gave us a blow: the conversion rate of ordinary users plummeted by 35%! The merchant’s customer service was overwhelmed by complaints from normal users: “I just want to buy a takeaway, how can it be like a loan?” ”
That lesson was unforgettable:The “security benefits” of risk control must be greater than the “operating costs + trust losses” borne by users.
We eventually moved to a more discreet model: a real-time dynamic scoring system based on device, behavior, and historical orders, which inductively intercepts 99% of normal users and only strengthens verification for high-risk behaviors.
The essence of risk control is not to erect a high wall, but to keep the bad guys out while the good guys are unimpeded.
2. The biggest enemy of the product is “pseudo-insight”: don’t let users do arithmetic problems
When I moved to Baidu as a data platform, I was once superstitious about “data freedom”.
We’ve built an incredibly powerful self-service analysis tool that supports SQL writing, multi-table association, and custom visualization. It is believed that this is a “sharp weapon” to empower users.
The reality is frustrating: core user activity is less than 10%. The survey found that ordinary business personnel were confused when they opened the interface: “All I want is the top 10 sales last week, why let me choose the database and write the fields?” ”
Those cool features have become the threshold that prevents users from obtaining value.
The real turning point is to change “customization” to “scenario”.
Instead of showing cold tool panels, we preset high-frequency scenarios: product managers look at the “feature penetration and retention chart”, operations look at the “campaign ROI dashboard”, and sales look at the “regional performance heatmap”.
Users can click to get key conclusions, and in-depth analysis needs are folded through the “Explore” button.
The core value of data products is to encapsulate complex calculations into insights that can be understood at a glance.
3. The ultimate product power in the AI era: eliminate the interface and return to humanity
After joining Tencent in charge of AIGC, I experienced an interactive revolution.
In the early days, we followed the traditional path: carefully designed a library of prompt templates, parameter adjustment sliders, style selectors…… Complete functionality, but lukewarm user feedback: “It’s still too complicated and I don’t know what to do.” ”
Until the advent of ChatGPT, the magic of its natural dialogue awakened us: the most instinctive human need is to express intentions in their own language, rather than learning the logic of machines.
We cut out all the complex controls, leaving only one input box: “Describe the picture you want to generate”. At the same time, a huge investment is invested in optimizing “intent understanding” instead of adding buttons – when users enter “Spring Festival posters, there should be dragons and fireworks, be festive”, the system automatically understands the style, elements, and themes and generates multi-version solutions.
The data confirms the direction:Conversational products have more than twice the weekly retention rate of traditional interfaces.
A deeper understanding comes from a user test. A designer tried the new feature and said: “In the past, I used professional tools to adjust the light and shadow for half an hour, but now I directly tell the AI ‘the sun shines down from the left and drags out a long shadow on the wooden floor’, and it actually understands.” ”
Now the birth of various agent products also confirms the correctness of this direction.
When technology can understand human emotions and imagery, interfaces become superfluous shackles.
Fourth, the sea of stars of productism
Looking back on this nearly 10-year product journey, I am more and more convinced that the ultimate password for product success isContinuously optimize the “net value” of user acquisition value – that is, the core income perceived by users, minus all operational costs, understanding thresholds and trust losses。
In 2025, when AI tools are blowing out, the role of product managers is undergoing a qualitative change: from functional designers to ecological architects, they need to pre-study technology trends and build a “product-as-platform” ecosystem. From requirements movers to value judges, find a balance between data utilization and privacy protection, business goals and user experience. From business executives to strategic prophets, just as Steve Jobs paved the way for AR with a gyroscope, product managers need to lay out their technical reserves 3-5 years in advance. For example, we started developing multimodal interaction technology in 2023 to lay the foundation for immersive content experiences in 2025.
A good product manager is the user’s subconscious translator, transforming vague desires into concrete value, and then hiding complex logic between simple breaths.
In the era of technology running, what is more important than chasing the outlet is to return to the essence of user value: so that every ordinary individual can enjoy the beauty given by technology painlessly.
This may be the sea of stars that product people are most worthy of.
