📎 Webclip
11 Things to Consider for More Accurate Developer Estimates
The article argues that estimates get harder as scope grows, because complexity, ambiguity, and unknowns increase. It separates high-level estimates from detailed ones and focuses on the factors that matter most when giving a quick but informed estimate.
Reading notes#
- High-level estimates are for early decisions, while detailed estimates come after a project has been vetted and can support more defined tasks and timeline
- High-level estimates are used often for projects, features, and bugs, so the post focuses on that kind of estimate
- The first step is to identify the pieces involved, including systems, codebases, APIs, databases, frameworks, plugins, architecture, and patterns
- Familiarity with those pieces affects the estimate, so time to learn documentation, research, and talk to knowledgeable people has to be included
- Changes to existing functionality can increase regression testing, especially when core features are affected
- Existing unit and integration tests can reduce the testing burden, but they may also need to be updated
- Communication and organization take time, including task coordination, progress updates, clarification, and requirement discussions
- Writing unit tests should be counted when new functionality is being built, and TDD includes that effort as part of development
- Integration tests add more work when a team uses them
- Manual testing remains necessary, especially when automated coverage is limited and after each bug fix or requirement change
- Bugs found during testing need triage, prioritization, fixes, retesting, and regression checks
- Documentation updates add effort, especially for larger or more complex solutions
- After go-live, real users and real data can expose more bugs, so stabilization work should be included
- Handing off to operational support can require extra effort and a transition period where the development team still helps
