Dependency types
- Finish-to-Start (FS) — most common. B starts after A finishes. (e.g. test after build.)
- Start-to-Start (SS) — B starts when A starts (in parallel). (e.g. design + research kick off together.)
- Finish-to-Finish (FF) — B finishes when A finishes. (e.g. testing finishes when development finishes.)
- Start-to-Finish (SF) — rare. B finishes when A starts. (e.g. old shift ends when new shift starts.)
Baseline tracking
When you approve the schedule, save it as the baseline. As the project progresses, render two bars: the baseline (often gray) and the current (often colored by status). The variance between them is what sponsors actually want to see — not the current bar alone.
Buffer placement
Two schools: per-task buffers (each estimate includes its own pad) or aggregated buffers at the project end (Critical Chain method, Goldratt). The aggregated approach is statistically tighter — task-level buffers tend to get consumed regardless of need (Parkinson's law: work expands to fill the time).
↳ in the wild
Most teams pad estimates and pretend they don't. Be explicit: tell the team "estimates should be 50/50, the buffer lives at the milestone level." Honest individual estimates + a visible aggregate buffer is a more truthful schedule than padded individuals + a hidden Gantt line.