Microsoft, Google or neither: choosing a host per domain
Once you can put a domain on either host, the interesting question stops being which provider is better and becomes which one this particular domain belongs on. The answer is usually decided by something other than the mail features.
Start with where the person reading it already lives. A domain whose mail goes to someone in Outlook all day should be a Microsoft shared mailbox; one going to someone in Gmail should be a Google Group. Cross-mounting works and nobody enjoys it. This single question settles most domains.
The second question is whether the domain needs an inbox at all. A great many don't. A parked name, a redirect domain, a brand that only ever sends receipts — these need a verified sending identity and nothing else. Onboarding those in send-only mode adds exactly two records, a DKIM key and a pm-bounces CNAME, and never touches MX or SPF. It is the safest option available and the one most portfolios should use for the long tail.
The third is who has to manage it later. Groups have an advantage that rarely gets mentioned: membership is the whole permission model. Adding and removing people is one action, with no mailbox permission to grant, no automapping delay, and no risk of a grant landing on the wrong identity. For a role address that changes hands — treasurer@, a rotating on-call alias — that is worth more than any feature difference.
What should almost never decide it is cost, because on the free paths there isn't one. A Microsoft shared mailbox and a Google Group both cost nothing, on subscriptions you already pay for. The only setup here that consumes a seat is choosing a real Workspace user for a domain, which is worth doing when someone genuinely needs a full Gmail account for that identity — and worth avoiding otherwise.
A mixed estate is fine, and for anyone who has taken over someone else's setup it is inevitable. inboxroot connects both hosts at once and asks which one you want each time you add a domain, because there is no reason to force a portfolio into a single column.