There have been a lot of posts complaining about why a particular post has lost money over time. I am guessing about how this works, but the best I can tell is that Steemit currently shows what the payout would be right now if no other posts were upvoted. So if you post early in a payout cycle, you get a lot of money that tends to trickle away as other good posts get upvoted. A post gets a portion of a pool of funds during that payout period. The more good posts, the less any particular post will make.
This is a real problem for most people because it triggers their loss aversion feelings. As a simple example, if your post makes $20, but you were told early on that it would only pay $15, you are going to feel great about it. However, if you were told that the post was going to make $25 and you only get $20, then you feel much worse than if you were told $20 or $15 at the beginning.
I propose that Steemit change the algorithm for showing author awards to take into account how far from payout the post is and how many more posts are expected in the payout cycle. That way a post will show a little less money in the beginning, but would tend to go up as time passes and confidence level of the payout increases. This would make authors happier because they can watch a post increase in value. There's also been proposals not to show any payout at all prior to payout, but that negates the ability for curators to downvote articles with payouts they consider too much. The proposed approach, consistently applied, will still let curators be able to recognize post payouts they consider abusive and at the same time negate the average posters feeling of loss aversion.
There are other mundane reasons why a post does not make money. The biggest is just dumb luck. You posted a great article and the whales were sleeping or just missed it. This also triggers feelings of loss aversion, in the form of why bother if nothing happens. But the real truth is that if you don't post you don't get anything. So basically you need to just keep trying and writing interesting material and the votes will come.