How to Systematically Build Successful Technical Products for Developers

Reflecting on years of experience—from creating Iceland in 2017, contributing to ICE in the open‑source community, to building AppWorks in 2020—the author outlines a systematic approach to building technical products for developers.

Designing the Product

The first step is defining what to build, based on market and user research to identify urgent problems and existing solutions. Feature classification follows the "We too / We better / We only" framework.

Key deliverables include market research reports, user research reports, product interaction drafts, website drafts, and documentation outlines.

Designing the Architecture

Next, decide how to build, using competitive analysis and architecture diagrams. UML structural diagrams (class, component, deployment) and behavioral diagrams (use case, activity, sequence, state machine) help visualize static and dynamic aspects.

Examples include the architecture of the VS Code Time Master plugin and the Ant Could IDE research report.

Managing the Project

Organizing Development

Repository layout (monorepo vs multi‑repo) should anticipate future scaling. Branch management adopts a Git workflow—commonly Gitflow—with clear branch types and merge rules.

Commit message conventions follow the pattern

<type>[optional scope]: <subject>

, where

type

can be

feat

,

fix

,

docs

,

style

,

test

,

refactor

,

chore

, or

revert

. Tools like

commitlint

,

@iceworks/spec

, and

husky

enforce these rules.

Engineering Practices

Standard scripts cover

setup

,

dev

,

lint

,

test

,

build

, and

publish

. Front‑end projects often use React (ice.js) or Rax (rax‑app) templates, component scaffolding, or full‑stack solutions with Midway Hooks.

CI/CD pipelines—commonly GitHub Actions—run tests, build artifacts, and deploy to CDNs or package registries. In Alibaba, DEF provides unified CI/CD workflows.

Collaboration Mechanisms

Establish RFC processes for design discussions, PR guidelines for quality and review efficiency, and CR (code review) guidelines to balance rigor and productivity.

Documentation

Effective documentation improves developer experience. Production channels include Yuque, self‑hosted sites, and Git repositories, each with trade‑offs. Large or open‑source projects often adopt a static‑site approach.

Developing the Website

Static site generators such as Docusaurus, VuePress, and Docsify are evaluated on theme support, customizability, multi‑language/version handling, example rendering, onboarding difficulty, and deployment cost. Deployment commonly uses GitHub Pages or internal platforms like DEF.

Operating the Product

Product operation requires authenticity and moderation—promoting real value without spamming. Content types range from technical deep‑dives to feature announcements, with titles that attract readership.

Managing Demand and Defects

Post‑launch, issues and feature requests are handled via Issue templates, labeling, and community support channels (GitHub Discussions, Discord, DingTalk). Internally, tools like Aone and Research Bee track tickets and provide Q&A services.

Conclusion

Building a technical product is akin to crafting a wooden toy: the basic shape is easy, but polishing it into a beloved tool requires patience, iteration, and continuous refinement.