Engineers I know who I've worked with on GTM implementations are not big fans. It offers the ability to execute arbitrary JS on your site, does not live in our main repo (unless we export a container and store its JSON there), and there doesn't seem to be any good way to run integration tests on code within GTM itself (obviously you can run tests that observe the output of GTM).

As a marketer, it has indeed enabled me to launch tags and such without an engineer, but in many cases I'd actually prefer to have this live in our code base so it isn't broken by accident by someone who had no idea that some of what they were working on touched a container they had no visibility into.

All of this to say that I get worried when looking at more involved A/B tests that touch pieces of ones product. Sure it might be fine for superficial front-end design tweaks, but testing anything deeper scares me.

Also, random aside, but the author might wish to get some proofing of their article. There's a ton of typos and indicators of a non-native English speaker writing this and made for a more difficult reading experience.