I've seen this exact mess before. Country subdirectories with nearly identical content - it's a classic cannibalisation trap. Google doesn't just penalise you, it serves the wrong variant to users. Hreflang might help but isn't a silver bullet if content is practically the same.
You're right that subdomains are a resource sink. They fragment link equity and force you to rebuild authority from scratch for each country. traffic loss estimates? Unless you've got solid baseline data on organic traffic per country and can model the redistribution of link signals, any number you give your boss is guesswork. I'd push back hard: ask for at least three months of crawl logs and rank tracking per directory before modelling.
Someone in the thread mentioned better luck with subdirectories - that matches my experience too. Subdirectories keep link juice intact and make hreflang implementation cleaner. the real fix is reducing content duplication, but if legislation ties your hands, focus on unique meta data, structured markup, and internal linking that reinforces country signals. Don't split into subdomains unless you have budget for a dedicated PR campaign per market - and even then, expect 6-12 months of pain.
for what it's worth, I've run migrations both ways. Subdomain flip lost 40 % of organic traffic in the first quarter, subdirectory restructure cost about 10 % but recovered within two months. Your mileage will vary, but nobody should pretend there's a clean answer without running a controlled test on a single country first.
