1. If there is nobody who knows how the previous conf mgmt system was configured, or how the technology works.
2. If there is nobody willing to adopt a cluster using conf mgmt that they do not prefer.
We had an admin that set up an ELK cluster with cfengine. All of the other admins do not claim to understand it, and hesitate to maintain it, update it, make changes to it, and otherwise perform their job duties. One of the admins is partial to Chef, and the other wants to maintain the ELK cluster with Kubernetes. Only problem is neither has done so, and they have made little progress. I hope to see that change.
It can be, depending on how complex your environment is and the availability of people that understand it. Chef doesn't make this easy by storing so much "state" (nodes, environment, databags, etc).
But at the end of the day, these are just scripts. You can reverse-engineer the environment and replicate in Ansible or what have you.
One red flag is, if this process is found to be too difficult(as opposed to just time-consuming), then you have a very incomplete understanding of your system. That's dangerous.
The Wayback Machine is a project by Archive.org to archive the internet and allow users to browse old versions of pages.
The Archive Team is a community of archivists who do their own thing, but frequently contribute their collections back to Archive.org. There's a lot of cross-pollination between the two.
Same for Salt Stack I'd say. Just a yaml wrapper around execution modules which themselves are either wrappers around cli's or pure python versions of them.