This is a real problem ... Most people have no idea how marketing works, and especially PR ...
PR - it's always the case of the experiment, and the result is not guaranteed! Moreover, the experiment is a long-term: during long period, eliminating inefficient places and approaches (predict outcome in advance 100% rarely possible - everything is learned only during the tests), there are no clear metrics to measure progress, etc.
Yes, the lack of control (in opposite to usual case of purchasing goods, etc.) and the need to rely on trust - is unpleasant and unusual for most of us - but it is inevitable with PR - there is simply no other way.
The only effective PR - is a hidden PR. Publishing reports of PR-work - at times reduces their results and even makes them negative. It's small area - very few people who had experience with it. But who is doing this in practice knows no other way simply does not exist. You have to take risks and do spendings, including inefficient spending.
The only way to increase efficiency and reduce risks - do it on a regular basis, accumulating experience and constantly "adjusting" the company. Ordinary people never done it personally, and they think that PR-people are just trying to cheat them. And the only way - is to trust the professionals (even if they will deceive you in final).
I've earned my living with CPA-marketing for many years - it has similar principals to PR:
1. You spend $ 1000 for testing-1 and get $ 800 loss.
2. You adjust campaign and spend another $ 1000 for testing-2 and get $ 300 loss.
3. You adjust campaign and spend another $ 1000 for testing-3 and get $ 10 loss.
4. You adjust campaign and spend another $ 1000 for testing-4 and get $ 100 profit. Congrats - you are professional and found profitable scheme after just testing-4!!!
5. Now you can set budget of your found campaign-scheme to $ 100.000 and earn $ 10.000 protif.
This how it works and this is the only way to do it (except some rear luck cases). There is no way even for super professionals to earn $ 10.100 profit without this "$ 1.110 losses during testing".