If your strategy is repeatable/algorithmic, i.e. you'll take the same type trade many times, rather than a completely discretionary strategy where each trade is a once off analysis, then understanding the long run expectation of the trade is important.
To use an extreme example to demonstrate the point, say your strategy is successful only 10% of the time, i.e. 1 out of 10 trades wins. So if the size of the loss is -$1, then you need the size of a win to be at least $9 or you will lose money in the long run. In this case, taking profit early (say at $6) on a winning trade will result in a losing strategy overall unless taking profit early significantly improves your win rate from 10% to 15% of trades taken.
Also look at time-based exits which close after a certain time past the signal occurring. In my experience these often product a superior profit factor than setting defined s/l or t/p levels. Time based exits assess how long past your entry signal the trade will have been exhausted and then always exist after that duration. e.g. most of the market reaction to an interest rate change might occur within x minutes/hours/days of the event, so simply closing after x will capture both a massive move and a small move. Similarly, a bounce off a support level will on average take x bars to reach resistance, so just close of x bars. Time-based exits obviously should be backed up with a catastrophic stop loss as well to defend against an extreme outcome.
Finally, in my experience, stop losses in general degrade the performance of a strategy, rather than improve it. I'm not saying don't have a stop loss! What I am saying is that stop losses are a risk management tool, not a profitability management tool and most strategies will perform better (but with unacceptable risk parameters) if a stop loss isn't in place and the trade is just left to expire after a certain time period or when it hits it's t/p level.
Ian