It starts with an acknowledgement that the company has a problem with X. That's actually incredibly hard, because there's nobody in the company willing to say "I'm anti-X". If they did, you'd just fire them and call it a job well done.
Instead, you need to realize that the problem with X is that it's no individual's fault, but the cumulative effect of a lot of little things. Each of those little things is easily dismissed as irrelevant. The job of the X Officer is to care about all of those things at once.
That doesn't make their job easy. Simple changes rarely fix the problem -- if they could, you'd have already done them. They require large changes that often seem antiproductive, especially when you've defined "productive" in ways that you're convinced are objective but just happen to systematically be anti-X.
A common example: coding tests. "We don't exclude women. It's just that men happen to be more on both ends of bell curves, so it's just too bad that far more men pass this test than women. The test is objective, after all." Except that the test doesn't really test what you do for a living. So why insist on it? Is it because you're sexist, or just lazy? I'm tempted to call it the latter, but if it's pointed out by an "X Officer", they'll be accused of affirmative action, misanthropy, etc.
A diversity officer will lose more battles than they win, so it's hard to say what a job well done looks like. In a lot of ways they're doing their job well just by making people actually oppose them out loud. Their best victories look like things other people consider discriminatory against them, because the things that discriminate against them are intolerable while the things that discriminate against other people are just things that happen.
The hope is that collectively they'll put enough people in enough positions of authority to be able to gradually diminish the constant throb of small injustices that collectively have brought about an overwhelming white maleness to all authority positions. Even a large company is only a tiny fraction of society, so it takes the full cumulative effect over decades to actually achieve genuine success.
I hope that answers your question. It's not an easy thing to describe, and it's easy to dismiss the problems they're trying to fix and not worth solving.