You’re running into a major issue after you’ve reduced bids.
You participate in high-converting auctions less often. This happens because other advertisers already optimized their bids. They bid higher for the same keyword when a conversion rate probability is higher, so they win the good quality auctions. For instance, they bid higher when a user is searching for a new Ferrari from a high-end device, is located in Beverly Hills, etc.
You decrease bids expecting your conversion rate to remain the same but it decreases. But your lower bid is only enough for showing your ad about new Ferrari to teenagers from poor neighborhoods. As a result, your overall performance decreases.
Reducing bid will also decrease the number of clicks you get, which in turn decreases your opportunities to earn more. Low number of clicks reduces your chances to optimize efficiently because your results are statistically insignificant.
Does this mean you should never reduce bids? No. It means you have to take into account a lot of other factors besides the current conversion rate. The actual search query, location, time of the day, age, gender, audience have a major impact on profitability. Plus, you have to make decisions only when your stats per such entity are significant (and they aren’t significant in most cases). It’s a heavy task for manual bid management even if you use automated rules.
What to do instead
Master performance-based bidding strategies. There is a guide to modern bid management in the advanced course. But there is something you should know before you jump into the bid management. There are 2 ways you can address the poor performance of a keyword.
The first way is to remove a keyword or to reduce its bid. When specialists choose this way, soon, they end up with just a few to zero working keywords.
The second way is to keep the keywords yet just fix the causes of their poor performance.