How do you calculate schedule compression ratio?
From McConnell's book, Rapid Development, he describes compression ratios from researches involving hundreds to thousands of projects. So I think you can't just calculate the compression ratio of one or two projects, as you can't know how much the project would cost without the compression.
Another method that he cites, is to make a specific research with the same problem, and give different priorities to five teams, and compare them, but again, it's just for research purposes. Maybe a more useful method that you can apply is to gather your company's historical data and match projects of about the same size and compare their schedules length.
When and why would you do it? What does it tell you? What benefit does this bring to a project manager?
I would do it to verify the effectiveness of new process or tools that are not well known or don't have enough data to prove its effectiveness, mainly with the "silver bullet branded" ones. The benefit is that you can assert by yourself if the new process/tool really fits well in your project, or if it does't justify its costs and should be avoided in future projects of the same kind.
Are there heuristics with the the ratio that are to be generally followed?
Yes. Compression ratio is not an exact number, you can't say that a practice will compress your schedule by 18.32%. It may have compressed 18% in one project, 5% in another and none at all in another one. Again, based on McConnell's book, mainly in the Best Practices chapter, he describes compression ratios (potential reduction from nominal schedule) as possibly:
- None (0%)
- Fair (0%-10%)
- Good (10%-20%)
- Very Good (20%-30%)
- Excellent (30%+)
So, it's better you classify the ratio by these heuristic statements, rather than by numbers, as it shows the inexactness of the ratio.