When cost per acquisition is too high, the instinct is to lower bids. That usually reduces volume without improving efficiency. Here are the levers that actually work, in the order I pull them.
0. Verify the tracking first
This is not really a lever, it is a precondition. If page views are marked as conversions, or conversions fire twice, or calls are not counted at all, every subsequent decision is built on fiction. Roughly half the accounts I audit have at least one of these. Fix it before optimising anything.
1. Mine the search terms report weekly
This is where most waste lives. Broad and phrase match will find searches adjacent to your business rather than in it, and left unattended they consume a startling share of budget. Twenty minutes a week adding negatives is the highest-return recurring task in paid search.
2. Fix the match type mix
Broad match with smart bidding and a strong conversion history can work well. Broad match with weak data is an expensive education. If your account has fewer than about thirty conversions a month, tighten toward phrase and exact until the data supports more freedom.
3. Send traffic somewhere relevant
Ad-to-page mismatch is a silent CPA killer. Someone searching for a specific service should land on that service page, not the homepage. Dedicated landing pages routinely improve conversion rate enough to change the maths without touching bids at all.
4. Match bidding to your data volume
Target CPA and Maximise Conversions need conversion volume to work. Below roughly thirty per month they behave erratically. Manual or Maximise Clicks with tight targeting is frequently better for small accounts, whatever the interface recommends.
5. Segment by device, location and hour
Look for the segments quietly spending without converting. Mobile at 2am in a region you do not serve is not a hypothetical; it is in most unmanaged accounts.
6. Improve Quality Score honestly
Relevance between keyword, ad and landing page lowers your cost per click for the same position. It is slower than the other levers and it compounds.
7. Cut what does not work
The hardest one. Campaigns kept alive out of hope consume budget that would perform elsewhere. If something has had a fair run with clean data and is not producing, stop it. That decision is easier when your conversion tracking is trustworthy, which is why it comes first.