Platform Usage Paradox: Why Don't Developers Use Platforms?

Platform Usage Paradox: Why Don't Developers Use Platforms?

From platform strategy to developer behavior, explore the psychology and mechanisms behind it

In the seventh year since its launch, Weixin Mini Programs have exceeded 600 million daily active users. Ten independent developers can be consulted, and nine and a half of their main businesses are still making native apps. This is not that they do not understand the general trend, but that they settle the bill. The platform gives you traffic entrances, payment channels, and trust endorsements, which seem to be packaged and delivered to your hands; from another perspective, the platform locks "user relationships" in its own pool and holds "distribution rules" in the algorithm black box., stuck the "monetization cap" in the percentage. Developers do not "use the platform" because they are not lazy, but afraid-they are afraid that after handing over their lifeblood, they will not even be qualified to bargain.

Let's first look at the developer's psychological accounts. At the end of 2023, I chatted with a founder who does vertical SaaS. He has a team of 20 people and an annual turnover of 30 million yuan. I asked why I didn't go to the application market of Flybook/Dingpin/Enterprise Micro? He threw out a chart: the platform earns 15%-30%, but the cost of acquiring customers is 40% higher than that of self-built official websites. The most heartbreaking thing is that customer data cannot be exported, and the migration cost is close to infinite. He said: "It's not that I don't want to borrow money. I don't dare to store the deed in someone else's safe." This is not an exception. Stack Overflow's 2023 survey shows that 68% of professional developers prefer "self-built/controllable" rather than "platform hosting". The top three reasons are: data sovereignty, pricing autonomy, and technology stack lock-in risks. Quantifying "control" as migration costs instantly increases the threshold in developers 'hearts.

Let's look at the way China's Internet platform plays. Ten years ago, Zhang Xiaolong said that "WeChat is a way of life." Ten years later, Mini programs became the "App Store of Super Apps." But carefully disassembled, the platform never does "ecology", only does "traffic distribution." Meituan has pushed takeout merchants into the prisoner's dilemma of "must-buy promotion". Douyin has stuck e-commerce anchors in the double harvest of "drawing commission + streaming". Lianyun manufacturers '"cloud subsidies" are essentially based on low prices. Lock up the three-year contract. The business logic of the platform is extremely clear: Turn high-frequency demand into infrastructure, and turn low-frequency long tail into material for traffic monetization . Every "access" a developer makes to the platform is a transaction with "zero customer acquisition costs" on the platform ledger. This is not cooperation, it is harvesting.

At this time, someone always jumped out and said: Is the platform worthless? Without Weixin Pay, cloud vendors 'GPU clusters, and application store distribution capabilities, how many wheels must entrepreneurs build themselves? Steelman's version of the opposite view : The platform lowers the infrastructure threshold and compresses the time from 0 to 1, allowing small teams to reach hundreds of millions of users. This is a real dividend. Especially for early projects, the trust endorsements and payment channels provided by the platform are irreplaceable by self-construction.

But this cliché misses a key variable-* diminishing marginal return in the time dimension . The platform value is positive in the "0 to 1" stage and gradually turns negative in the "1 to 10" stage. When your DAU rises from 10,000 to 1 million, the proportion of incremental traffic given to you by the platform is declining, but the proportion of withdrawals is rising; when you want to do differentiated functions, platform API restrictions make it difficult for you to move; when you want to migrate, it is found that the user relationship chain cannot be derived at all. What's even more ironic is that platforms like to use "we helped you solve infrastructure" as an excuse, but didn't AWS also solve infrastructure? Why doesn't AWS take 30% of your business flow? Because AWS sells standardized resources that can be migrated , while China platforms sell proprietary niche that cannot be migrated . The former is a tool dealer, and the latter is a landlord.

Putting this matter into urban planning will be more intuitive. The developer is the "merchant who opens the store" and the platform is the "mall operator." The mall will give you decoration standards, crowd guidance, cash register system, and even help you bring customers. The premise is that the rent is drawn according to the flow rate, the decoration must follow the unified style of the mall, the customer list belongs to the mall, and if you want to move, you have to smash the sign and re-establish it. Silicon Valley's model is like an "industrial park": buy land and build your own factories, connect the water and electricity network according to standards, add it at any time if you want to expand, and relocate the entire factory and pack it away. No one will choke you. The Internet in China is accustomed to building "supermarkets" where merchants are locked in to collect traffic, convert, and withdraw profits; Silicon Valley is accustomed to selling "standardized land", allowing merchants to compare product power, service, and brand strength in their own territory. The former is the distribution logic of the stock game , and the latter is the infrastructure logic of incremental competition . This is the most essential footnote to the theme "The paradigm difference between China Internet vs. Silicon Valley".

One detail is often overlooked: the platform's dual-standard treatment for "big customers" and "long tail". Leading manufacturers can access the platform and get customized APIs, exclusive customer service, and even reverse share; long-tail developers can only eat standard terms, go through the work order system, and cannot even find an appeal entry. In 2022, a cloud vendor's "database is read-only but not exported" incident was finally compromised by major customers, and long-tail users have not yet waited for an explanation. This power asymmetry is written in the contract terms and also engraved in the platform architecture: OAuth only authorizes but not opens, Webhooks only push but not callbacks, and data export only has CSV and no API. The platform uses "security compliance" as a shield, but in fact it turns "migration costs" into a moat. Developers are not stupid, they can see through it at a glance.

The deeper mechanism is incentive mismatch . Platform KPIs are DAU, GMV, and advertising revenue-all need to keep users in a closed loop within the platform. Developer KPIs are LTV, repurchase, brand assets-all of which need to deposit user relationships in their own hands. The more successful the platform is, the more marginal the developer is; the stronger the developer, the more anxious the platform is. This is not a misunderstanding, but a structural contradiction. In the early days, WeChat Mini games gave developers 100% of their revenue, and when they were bigger, they took 30% of their profits + forced the use of Weixin Pay; Douyin E-commerce attracted 0 commissions in the early days, and the scale grew to double harvest "commission + streaming". Every platform is repeating the same script: Exchange openness for scale, exchange scale for control, and use control to monetize . Developers do not "use the platform", essentially refusing to make the "material" in this play.

This reminds me of a phenomenon: the China team that has built a sea would rather spend three times more manpower to build its own payment, push, statistics, and customer service systems than accept the local "Super App" SDK. Indonesia's GoPay, Vietnam's ZaloPay, and Brazil's Pix have full documents, full sandboxes, and fast settlement, but the China team still prefers Stripe + Firebase + self-built backend. Why? Because most of the overseas teams have suffered losses from domestic platforms, they have developed muscle memory of "controllable preferences". Once you have tasted the sweetness of "data can be moved at any time, rules can be set by yourself, and leave at any time if you don't want to do it", you can no longer stand the pain of "reviewing for two weeks, refusing without reason, and migrating to death." This is not a technical choice, it is a post-traumatic stress response.

So back to the paradox at the beginning: the platform is desperately trying to create entrances, tools, and incentives, but developers collectively vote with their feet. It's not that developers don't appreciate it, it's that the platform defines "cooperation" as "inclusion in jurisdiction." The real platform thinking is not to lock developers in walls as traffic materials, but to be like AWS, Stripe, and Vercel-- do water, electricity, coal, not landlords; sell shovels, not digging for gold; set standards, and determine fate. . Such a platform has not yet grown on the Internet in China, because in the existing market, the ROI of "being a landlord" is too high.

One last question: If one day WeChat opens the "User Relationship Chain Export API", Douyin opens the "De-Platform Distribution Protocol", and cloud vendors promise to "unconditional full data migration"-developers will flock in, or still choose to build it? I bet on the latter. Because once trust is overdrawn, no matter how complete the document is and no matter how standard the interface is, it cannot fill in the hole called "sense of security." If the platform wants developers to truly "use the platform", they must first return the deed to others-this is the most expensive and scarce infrastructure.