"Experiments would take at least several months to run, because it's not till several months that you know whether your business is going to fail solely due to the dev team."
If you use the rather stodgy and formulaic scientific method, sure. But if you approach this as a multi-arm bandit problem, and instead look at the question as "how much information do I need to collect with how much statistical power before I am justified in selecting an outcome?", it becomes quite tractable for a team to just try it out.
It also helps if instead of using merely "productivity" as your one and only goal that you also consider team happiness as a real and going concern. Then "Nobody likes it and it's demoralizing the engineers" becomes a perfectly valid form of feedback to use in the decision about whether to do it.
I also tend to use the metric that the brain is really, really, really good at figuring out what's important, sometimes perhaps even too good, so if we start a stand-up meeting plan, and it just sort of peters out, then I let it go. (Seems like somebody wants to try it every 2 or 3 years or so, I generally just let it happen and then evaluate.) Everybody's brains are coming to the conclusion that it's not important. Rather than fighting it, roll with it. There are some things in life where your rational forebrain must override the deeper instinctive algorithms about what is important, but that's quite cognitively expensive, and I've never seen any reason to believe that stand-up meetings are bringing enough value to justify that.
I emphasize the "I've never seen any reason". I hypothesize that in many ways, the work the teams I've been on have been doing is almost maximally pathologically bad for a stand-up meeting. Almost everyone is working on something independent of everybody else, so blockers are only rarely even possible. If your team is, on the other hand, chock-a-block full of blocking potential it may be useful to you.
Anyways, there's more ways to find out the truth of things that trying to split the universe into two precisely equal parts, make a change in only one of them, and come back ten years later to see what happened, only for someone to complain that a sample size of one isn't enough and we really need to do this a thousand more times to have a valid result.