Really good point on the licensing-vs-custom-dev distinction — that's honestly the more accurate framing than what I had. "Self-hosted = cheaper" only holds if you're not quietly rebuilding half the compliance/localization layer yourself, and that work doesn't show up in any TCO comparison until a team is already mid-implementation. The ops-overhead point on self-hosting is fair too — I probably undersold that in the article. "You control your infra" and "you're now responsible for your infra" are the same sentence read two different ways, and the second one is the one that actually shows up in a 2am incident. Useful to hear DPP is already showing up in real client conversations, not just regulatory reading. "Plan the data model now, don't rebuild later" is a much better way to frame the urgency than what I wrote — makes it a now-decision about schema/extensibility rather than a someday-decision about compliance. Might be worth its own follow-up post honestly. Curious what "plan the data model now" looks like concretely for you — is it mainly about keeping attributes extensible enough for GS1/passport fields later, or does it go deeper into how categories/relations are structured?
