1. Scaling budget too fast
A campaign starts performing and it's tempting to double or triple the budget immediately, chasing more volume. The problem: one good day — or one bad day — tells you almost nothing on its own.
Let the campaign run for three to five days minimum before touching budget. Seven days is better if you can afford to wait. Judge the average across that window, not any single day.
If the average holds up, scale by 20-25% max per increase, then hold for another three days before bumping again. Slow, incremental increases protect the algorithm's learning phase. Doubling or tripling overnight almost always resets performance and pushes your CPA back up.
2. Your ad doesn't match your landing page
This kills conversion rate faster than almost anything else. Someone clicks an ad that makes a specific promise, lands on the page, and the messaging doesn't match. That gap creates instant doubt, and they leave.
Whatever your ad promises — offer, price, benefit, hook — your landing page needs to confirm it immediately. Message match isn't optional; it's the difference between a click that converts and a click you paid for and wasted.
3. Killing creatives at the ad level instead of the ad set level
An advertiser sees one creative with a CPA that looks too high and kills it, without checking how the ad set as a whole is performing.
The catch: that "underperforming" creative might be doing the prospecting work — driving cold traffic into the ad set that then converts on your other creatives at a lower CPA. Kill it, and you don't just lose that one ad, you cut off the top of the funnel feeding everything else. Always evaluate performance at the ad set level before touching individual creatives.