Budget allocation is something I've struggled with too. I now prefer putting more money behind campaigns that have already demonstrated good conversion quality instead of spreading the budget equally across everything.
One of the biggest lessons I've learned is that PPC doesn't exist separately from the website. You can have excellent ads, but if the website doesn't build trust, explain the offer clearly, or make contacting the business easy, conversions will suffer.
I agree with testing, but I think tests need enough time and data. I've seen people declare an ad a winner after only a few days because it had a slightly better CTR. That's not always enough evidence to make a decision.