Five mistakes ad operators make (and how to stop)

After running audits across dozens of Google Ads and Meta Ads accounts, the same five mistakes keep showing up. None require advanced skills to fix. All five are made by senior operators too — usually because they're under pressure to ship.

1. Setting a target CPA without checking the historical floor

Most accounts have a tCPA set by whoever ran the campaign last. Often that number was a guess made on day one. If your bid strategy is 'Target CPA' and the actual conversion floor over the past 60 days has been 3x your tCPA, Google will buy almost no traffic and you'll wonder why volume dropped.

Fix: pull a 60-day CPA distribution. Set tCPA at the 70th percentile of historical conversions, not the median. Loosen it 10-20% the first week to give the algorithm room to find the right buyers.

2. Scaling winners too aggressively

Found a campaign at 4x ROAS? Great. The instinct is to 5x the budget. Don't. ROAS at $100/day is not the same business as ROAS at $500/day — the audience is wider, the marginal buyer is colder, the conversion rate drops faster than the spend grows. Scale at 20-25% per week, not per day.

Fix: set an auto-scale rule that bumps budget +20% per week ONLY if 7-day ROAS stays within 20% of the prior 7-day window. The moment ROAS slips, freeze the increase. The FixAdSpend bid-optimizer agent does this automatically.

3. Treating broad match like a search term

Google's broad match is the most misunderstood targeting type. A query like 'leather wallet' on broad match might trigger your ad for 'wallet repair', 'wallet review video', or 'wallet emoji meaning'. Most accounts we audit have $200-2000/month going to irrelevant broad-match interpretations.

Fix: run the search-terms report weekly. Sort by spend descending. Add the top 10 irrelevant terms as exact-match negatives. Repeat until the list of new offenders shrinks below 3/week.

4. Ignoring creative fatigue until ROAS tanks

Creative fatigue starts about 7-14 days before ROAS visibly drops. CTR is the early warning sign. If CTR is down 20%+ week-over-week, ROAS will follow within 2 weeks unless you refresh.

Fix: track CTR week-over-week per ad. The moment it drops 20%, queue 2-3 new variants. Don't wait for ROAS to confirm — by then you've already burned 2 weeks of suboptimal spend.

5. Running ads to a slow landing page

We see $5k/month campaigns sending traffic to landing pages with a Largest Contentful Paint > 4 seconds. Google rewards fast pages with cheaper clicks (better Quality Score); Meta does too via 'destination experience'. A 2-second improvement in LCP often drops CPM 10-15%.

Fix: run PageSpeed Insights on your top-spend landing page. Anything above 4s LCP is costing you money in lower Quality Score AND in higher bounce. Compress hero images. Defer non-critical JS. The fix is often a 20-minute change with a months-long payoff.

What FixAdSpend does about all five

Each of these mistakes is detected automatically by a specific agent: profit-margin agent (#1), bid-optimizer (#2), negative-keyword agent (#3), creative-refresh + anomaly-sentinel (#4), and the landing-page-perf scoring is on our roadmap (#5). Each one surfaces a plain-English recommendation with the expected dollar impact.

But you don't need our tool to catch them. The audit takes about an hour the first time you do it manually, faster on repeat. Try the four checks in the 60-second post first; if you still want help, the free tier gives you the heuristic scan with no card required.