The three, and why they diverge
- Classic Outlook for Windows (2016–2021, still widely deployed in enterprises) renders HTML email using Word's layout engine, not a browser engine. No real support for background-image on most elements without a VML fallback, no reliable margin/padding on <div>, and a box model that ignores a lot of ordinary CSS. This is the one that actually breaks templates.
- New Outlook for Windows is a from-scratch rebuild on WebView2: a real Chromium-based renderer. It handles modern CSS the new Outlook client the way a browser would, and most of classic Outlook's workarounds are unnecessary here (and sometimes visually wrong when applied unnecessarily).
- Outlook for Mac is WebView-based too (closer to Safari's engine), but it isn't the same codebase as new Outlook for Windows; small gaps show up in each.
A template built and only checked against new Outlook or Outlook for Mac will look fine to you and then break for every recipient still on the classic desktop client, which, in a lot of enterprise and finance environments, is still the majority.
What to actually build for
- Use table-based layout for anything structural you need to survive classic Outlook: it is the one genuine constraint that hasn't gone away.
- Add a VML fallback for any background-image you need visible in classic Outlook; a plain CSS background-image silently disappears there.
- Don't over-correct: workarounds written for classic Outlook (nested tables, inline everything) aren't required for new Outlook or Outlook for Mac, and can make a template look worse on the clients that didn't need the fix.
- Check all three, not one. "Looks fine in Outlook" is only a real claim once it's been verified against all three engines specifically.
This is the exact reason Faulterly checks six real client engines instead of the usual four: Gmail, classic Outlook for Windows, new Outlook for Windows, Outlook for Mac, Apple Mail, and Yahoo/AOL, each rendered against what that engine actually supports, not a single generic "Outlook" bucket.